rfc9845xml2.original.xml   rfc9845.xml 
<?xml version="1.0"?> <?xml version='1.0' encoding='UTF-8'?>
<!DOCTYPE rfc [ <!DOCTYPE rfc [
<!ENTITY RFC.2481 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RF <!ENTITY nbsp "&#160;">
C.2481.xml"> <!ENTITY zwsp "&#8203;">
<!ENTITY RFC.3031 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RF <!ENTITY nbhy "&#8209;">
C.3031.xml"> <!ENTITY wj "&#8288;">
<!ENTITY RFC.3095 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RF
C.3095.xml">
<!ENTITY RFC.7950 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RF
C.7950.xml">
<!ENTITY RFC.8402 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RF
C.8402.xml">
<!ENTITY RFC.9330 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RF
C.9330.xml">
]> ]>
<?rfc comments="yes"?> <rfc xmlns:xi="http://www.w3.org/2001/XInclude" category="info" docName="draft-i
<?rfc compact="no"?> rtf-nmrg-green-ps-06" number="9845" updates="" obsoletes="" consensus="true" xml
<?rfc inline="yes"?> :lang="en" ipr="trust200902" submissionType="IRTF" sortRefs="true" symRefs="true
<?rfc sortrefs="yes"?> " tocInclude="true" tocDepth="5" version="3">
<?rfc subcompact="no"?>
<?rfc symrefs="yes"?>
<?rfc toc="yes"?>
<?rfc tocdepth="5"?>
<?rfc tocindent="yes"?>
<?rfc tocompact="yes"?>
<rfc category="info" docName="draft-irtf-nmrg-green-ps-06" ipr="trust200902"
submissionType="IRTF">
<front>
<title abbrev="Management for Green Networking">Challenges and Opportunities in
Management for Green Networking</title>
<author fullname="Alexander Clemm" initials="A." <front>
surname="Clemm" role="editor"> <title abbrev="Management for Green Networking">Challenges and Opportunities
<organization>Independent</organization> in Management for Green Networking</title>
<address> <seriesInfo name="RFC" value="9845"/>
<author fullname="Alexander Clemm" initials="A." surname="Clemm" role="edito
r">
<organization>Independent</organization>
<address>
<postal> <postal>
<city>Los Gatos, CA</city> <city>Los Gatos</city><region>CA</region>
<country>US</country> <country>United States of America</country>
</postal> </postal>
<phone></phone>
<email>ludwig@clemm.org</email> <email>ludwig@clemm.org</email>
</address> </address>
</author> </author>
<author fullname="Carlos Pignataro" initials="C." surname="Pignataro" role=" editor"> <author fullname="Carlos Pignataro" initials="C." surname="Pignataro" role=" editor">
<organization abbrev="NC State University">North Carolina State University <organization abbrev="NCSU / Blue Fern">North Carolina State University &a
</organization> mp; Blue Fern</organization>
<address> <address>
<postal> <postal>
<country>US</country> <country>United States of America</country>
</postal> </postal>
<email>cpignata@gmail.com</email> <email>cmpignat@ncsu.edu</email>
<email>cmpignat@ncsu.edu</email> <email>carlos@bluefern.consulting</email>
</address> </address>
</author> </author>
<author fullname="Cedric Westphal" initials="C" surname="Westphal">
<author fullname="Cedric Westphal" initials="C" <address>
surname="Westphal"> <email>westphal@ieee.org</email>
<address> </address>
<email>westphal@ieee.org</email> </author>
</address> <author fullname="Laurent Ciavaglia" initials="L" surname="Ciavaglia">
</author> <organization>Nokia</organization>
<address>
<author fullname="Laurent Ciavaglia" initials="L" surname="Ciavaglia">
<organization>Nokia</organization>
<address>
<email>laurent.ciavaglia@nokia.com</email> <email>laurent.ciavaglia@nokia.com</email>
</address> </address>
</author> </author>
<author fullname="Jeff Tantsura" initials="J" surname="Tantsura">
<author fullname="Jeff Tantsura" initials="J" surname="Tantsura"> <organization>Nvidia</organization>
<organization>Nvidia</organization> <address>
<address>
<email>jefftant.ietf@gmail.com</email> <email>jefftant.ietf@gmail.com</email>
</address> </address>
</author> </author>
<author fullname="Marie-Paule Odini" initials="M-P" surname="Odini">
<author fullname="Marie-Paule Odini" initials="M-P" surname="Odini"> <address>
<address>
<email>mp.odini@orange.fr</email> <email>mp.odini@orange.fr</email>
</address> </address>
</author> </author>
<date month="August" year="2025"/>
<date day="16" month="3" year="2025" />
<abstract> <workgroup>Network Management</workgroup>
<t>
Reducing humankind's environmental footprint and making technology more envi
ronmentally sustainable are among the biggest challenges of our age.
Networks play an important part in this challenge. On one hand, they enable
applications that help to reduce this footprint. On the other hand, they contri
bute to this footprint themselves in no insignificant way. Therefore, methods t
o make networking technology itself "greener" and to manage and operate networks
in ways that reduce their environmental footprint without impacting their utili
ty need to be explored. This document outlines a corresponding set of opportuni
ties, along with associated research challenges, for networking technology in ge
neral and management technology in particular to become "greener", i.e., more su
stainable, with reduced greenhouse gas emissions and less negative impact on the
environment.
</t>
<t>
This document is a product of the Network Management Research Group (NMRG) of
the Internet Research Task Force (IRTF). This document reflects the consensus
of the research group. It is not a candidate for any level of Internet Standard
and is published for informational purposes.
</t>
</abstract>
</front> <keyword>Sustainability</keyword>
<keyword>Sustainable Networking</keyword>
<keyword>Environmental Sustainability</keyword>
<keyword>Green Networking</keyword>
<middle> <abstract>
<section anchor="intro" title="Introduction"> <t>
Reducing humankind's environmental footprint and making technology more
environmentally sustainable are among the biggest challenges of our age.
Networks play an important part in this challenge. On one hand, they
enable applications that help to reduce this footprint. On the other hand,
they significantly contribute to this footprint themselves.
Therefore, methods to make networking technology itself "greener" and to
manage and operate networks in ways that reduce their environmental
footprint without impacting their utility need to be explored. This
document outlines a corresponding set of opportunities, along with
associated research challenges, for networking technology in general and
management technology in particular to become greener, i.e., more
sustainable, with reduced greenhouse gas emissions and less negative
impact on the environment.
</t>
<t>
This document is a product of the Network Management Research Group (NMRG)
of the Internet Research Task Force (IRTF). This document reflects the
consensus of the research group. It is not a candidate for any level of
Internet Standard and is published for informational purposes.
</t>
</abstract>
</front>
<middle>
<section title="Motivation"> <section anchor="intro">
<t> <name>Introduction</name>
<section>
<name>Motivation</name>
<t>
Climate change and the need to curb greenhouse gas (GHG) emissions have been rec ognized Climate change and the need to curb greenhouse gas (GHG) emissions have been rec ognized
by the United Nations and by most governments as one of the big challenges of ou r time. by the United Nations and by most governments as one of the big challenges of ou r time.
As a result, curbing those emissions is becoming of As a result, curbing those emissions is becoming
increasing importance for society and for many industries. The networking indus increasingly important for society and for many industries. The networking indu
try is no stry is no
exception. exception.
</t> </t>
<t>The science behind greenhouse gas emissions and their relationship with clima
te change <t>The science behind greenhouse gas emissions and their relationship wi
is complex. However, there is overwhelming scientific consensus pointing towards th climate change
a clear is complex. However, there is overwhelming scientific consensus pointing toward
a clear
correlation between climate change and a rising amount of correlation between climate change and a rising amount of
greenhouse gases in the atmosphere. greenhouse gases in the atmosphere.
When we say 'greenhouse gases' or GHG, we are referring to gases in the Earth’s
atmosphere that trap heat and contribute to the greenhouse effect. They include
carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), and Fluorinated gases
(as covered under the Kyoto Protocol and Paris Agreement). In terms of emission
s from human activity, the dominant greenhouse gas is CO2; consequently, it ofte
n becomes shorthand for “all GHGs”. However, other gases are also converted into
“CO2-equivalents”, or CO2e.
One greenhouse gas of particular concern, but by no means the only one, is carbo n One greenhouse gas of particular concern, but by no means the only one, is carbo n
dioxide (CO2). Carbon dioxide is emitted in the process of burning fuels to dioxide (CO2). Carbon dioxide is emitted in the process of burning fuels to
generate energy that is used, for example, to power electrical devices such as generate energy that is used, for example, to power electrical devices such as
networking equipment. Notable here is the use of fossil fuels, such as networking equipment. Notable here is the use of fossil fuels (such as
oil, which releases CO2 that had long been removed from the earth's atmosphere, oil, which releases CO2 that had long been removed from the earth's atmosphere),
as opposed to the use of renewable or sustainable fuels that do not "add" to the as opposed to the use of renewable or sustainable fuels that do not "add" to the
amount of carbon in the atmosphere. amount of CO2 in the atmosphere.
There are additional gases associated with electricity generation, in particular There are additional gases associated with electricity generation, in particular
Methane (CH4) and Nitrous Oxide (N2O). Although in smaller quantities, they hav methane (CH4) and nitrous oxide (N2O). Although they exist in smaller quantitie
e an even higher Global Warming Potential (GWP). s, they have an even higher Global Warming Potential (GWP).
</t> </t>
<t>Greenhouse gas emissions are in turn correlated with the need to power <t>Greenhouse gas emissions are in turn correlated with the need to powe r
technology, including networks. Reducing those emissions can be achieved by red ucing the technology, including networks. Reducing those emissions can be achieved by red ucing the
amount of fossil fuels needed to generate the energy that is needed to power amount of fossil fuels needed to generate the energy that is needed to power
those networks. those networks.
This can be achieved by improving the energy mix to include increasing This can be achieved by improving the energy mix to include increasing
amounts of low-carbon and/or renewable (and hence sustainable) energy sources su ch as wind or solar. amounts of low-carbon and/or renewable (and hence sustainable) energy sources, s uch as wind or solar.
It can also be achieved It can also be achieved
by increasing energy savings and improving energy efficiency so that the same ou tcomes by increasing energy savings and improving energy efficiency so that the same ou tcomes
are achieved while consuming less energy in the first place. are achieved while consuming less energy in the first place.
</t> </t>
<t>The amount of greenhouse gases that an activity adds to the atmosphere, such <t>The amount of greenhouse gases that an activity adds to the atmospher
as CO2 that is emitted in burning fossil fuels to generate the required energy, e, such as CO2 that is emitted in burning fossil fuels to generate the required
is also energy, is also
referred to as greenhouse footprint, or carbon footprint (accounting for greenho referred to as the "greenhouse footprint" or the "carbon footprint" (accounting
uses gases other than CO2 in terms of CO2 equivalents). Reducing this footprint for greenhouse gases other than CO2 in terms of CO2 equivalents). Reducing this
to net-zero is hence a major footprint to net zero is hence a major
sustainability goal. However, sustainability encompasses also other factors beyo sustainability goal. However, sustainability encompasses other factors beyond ca
nd carbon, rbon,
such as sustainable use of other natural resources, the preservation of natural such as the sustainable use of other natural resources, the preservation of natu
habitats ral habitats
and biodiversity, and the avoidance of any form of pollution. and biodiversity, and the avoidance of any form of pollution.
</t> </t>
<t>In the context of this document, we refer to networking technology that helps to improve its own networking <t>In the context of this document, we refer to networking technology th at helps to improve its own networking
sustainability as "green". Green, in that sense, includes technology that help s sustainability as "green". Green, in that sense, includes technology that help s
to lower networking's greenhouse gas emissions including carbon footprint, which turn includes technology that helps to increase efficiency and realize energy s avings as well as facilitating managing networks towards to lower networking's greenhouse gas emissions including the carbon footprint, w hich in turn includes technology that helps increase efficiency and realize ener gy savings as well as facilitates managing networks toward a
stronger use of renewables. stronger use of renewables.
</t> </t>
<t> <t>
Arguably, networks can already be considered a "green" technology in that networ Arguably, networks can already be considered a green technology in that networks
ks enable many applications that allow users and whole industries to save energy enable many applications that allow users and whole industries to save energy a
and thus nd thus
become environmentally more sustainable in a significant way. For example, they become environmentally more sustainable in a significant way. For example, they
allow (at least to an extent) to substitute travel with teleconferencing. They allow (at least to an extent) to substitute travel with teleconferencing. They
enable many employees to work from home and "telecommute", thus reducing the ne enable many employees to work from home and telecommute, thus reducing the need
ed for actual commute. IoT applications that facilitate automated monitoring and for actual commuting. IoT applications that facilitate automated monitoring and
control from remote sites help make agriculture more sustainable by minimizing control from remote sites help make agriculture more sustainable by minimizing
the application of resources such as water and fertilizer as well as land use. the usage of water, fertilizer, and land.
Networked smart buildings allow for greater energy optimization and sparser use Networked smart buildings allow for greater energy optimization and sparser use
of lighting and HVAC (heating, ventilation, air conditioning) than their non-net of lighting and HVAC (heating, ventilation, air conditioning) than their non-net
worked not-so-smart counterparts. That said, calculating precise benefits in te worked, not-so-smart counterparts. That said, calculating precise benefits in t
rms of net sustainability contributions and savings is complex as a holistic pic erms of net sustainability contributions and savings is complex, as a holistic p
ture involves many effects, including substituion effects (perhaps saving on emi icture involves many effects including substitution effects (perhaps saving on e
ssions caused by travel but incurring additional cost associated with additional missions caused by travel but incurring additional costs associated with additio
home office use) as well as behavioral changes (perhaps higher number of meetin nal home office use) as well as behavioral changes (perhaps a higher number of m
gs than if travel were involved). eetings than if travel were involved).
</t>
<t>The IETF has recently initiated a reflection on the energy cost of hosting me
etings three times a year (see for instance <xref target="IETF-Net0" />). It con
ducted a study of the carbon emissions of a typical meeting and found out that 9
9% of the emissions were due to the air travel. In the same vein, <xref target="
Framework" /> compared an in-person with a virtual meeting and found a reduction
in energy of 66% for a virtual meeting. These findings confirm that networking
technology can reduce emissions when acting as virtual substitution for physical
events.
</t>
<t>
That said, networks themselves consume significant amounts of energy. Therefore,
the networking industry has an important role to play in meeting sustainability
goals not just by enabling others to reduce their reliance on energy, but by al
so reducing its own. Future networking advances will increasingly need to focus
on becoming more energy-efficient and reducing carbon footprint, both for econo
mic reasons and for reasons of corporate responsibility. This shift has already
begun, and sustainability is already becoming an important concern for network p
roviders. In some cases, such as in the context of networked data centers, the a
bility to procure enough energy becomes a bottleneck prohibiting further growth
and greater sustainability thus becomes a business necessity.
</t>
<t>
For example, in its annual report, Telefónica reports that in 2021, its network'
s energy consumption per PB of data amounted to 54MWh <xref target="Telefonica20
21"/>. This rate has been dramatically decreasing (a seven-fold factor over six
years) although gains in efficiency are being offset by simultaneous growth in
data volume. In the same report, it is stated as an important corporate goal to
continue on that trajectory and aggressively reduce overall carbon emissions fur
ther.
</t>
</section>
<section title="Approaching the Problem">
<t>
An often-considered gain in networking sustainability can be made with regards t
o improving the efficiency with which networks utilize power during their use ph
ase, reducing the amount of energy that is required to provide communication ser
vices. However, for a holistic approach other aspects need to be considered as w
ell.
</t>
<t>Environmental footprint is determined not by energy consumption alone. The s
ustainability of power sources needs to be considered as well. A deployment tha
t includes devices that are less energy-efficient but that are powered by a sust
ainable energy source can arguably be considered "greener" than a deployment tha
t includes highly efficient device that are powered by Diesel generators. In fa
ct, in the same Telefónica report mentioned earlier, extensive reliance on renew
able energy sources is emphasized.
</t> </t>
<t>Similarly, deployments can take other environmental factors into account that affect carbon footprint. For example, deployments in which factors such as the need for cooling are reduced, or where excessive heat that is generated by equi pment can be put to productive use, will be considered greener than deployments where this is not the case. Examples include deployments in cooler natural surro undings (e.g., in colder climates) where that is an option. Likewise, manufactu ring and recycling of networking equipment are also part of the sustainability e quation, as the production itself consumes energy and results in a carbon cost e mbedded as part of the device itself. Extending the lifetime of equipment may in many cases be preferable over replacing it earlier with equipment that is sligh tly more energy-efficient but that requires the embedded carbon cost to be amort ized over a much shorter period of time. <t>The IETF has recently initiated a reflection on the energy cost of ho sting meetings three times a year (see <xref target="IETF-Net0"/>). It conducted a study of the carbon emissions of a typical meeting and found out that 99% of the emissions were due to air travel. In the same vein, <xref target="Framework" /> compared an in-person with a virtual meeting and found a reduction in energy of 66% for a virtual meeting. These findings confirm that networking technology can reduce emissions when acting as a virtual substitution for physical events.
</t> </t>
<t>Management has an outsized role to play in approaching those problems. To red <t>
uce the amount of energy used, network providers need to maximize ways in which That said, networks themselves consume significant amounts of energy. Therefore,
they make use of scarce resources and eliminate use of resources which are not n the networking industry has an important role to play in meeting sustainability
eeded. They need to optimize the way in which networks are deployed, which resou goals and not just by enabling others to reduce their reliance on energy but by
rces are placed where, how equipment lifecycles and upgrades are being managed - also reducing its own. Future networking advances will increasingly need to fo
all of which constitute classic operational problems. As best practices, method cus on becoming more energy efficient and reducing the carbon footprint, for rea
s, and algorithms are developed, they need to be automated to the greatest exten sons of both corporate responsibility and economics. This shift has already begu
t possible and migrated over time into the network and performed on increasingly n, and sustainability is becoming an important concern for network providers. In
short time scales, transcending management and control planes. some cases, such as in the context of networked data centers, the ability to pr
ocure enough energy becomes a bottleneck, prohibiting further growth, and greate
r sustainability thus becomes a business necessity.
</t> </t>
</section> <t>
<section title="Structuring the Problem Space"> For example, in its annual report, Telefónica reports that in 2024, its n
<t> etwork's energy consumption per PB of data amounted to 38 megawatt-hours (MWh) <
From a technical perspective, multiple vectors along which networks can be made xref target="Telefonica2024"/>. This rate has been dramatically improving from v
"greener" should be considered: alues that were reported for earlier years, for example 115 MWh/PB in 2019 or 40
<list style="symbols"> 0 MWh/PB in 2015 <xref target="Telefonica2020"/>. While portions of the gains i
<t> n efficiency are being offset by simultaneous growth in data volume, Telefónica
Equipment level:<br /><br />Perhaps the most promising vector for improving netw states that an important corporate goal is continuing on that trajectory and agg
orking sustainability concerns the network equipment itself. ressively reducing overall greenhouse gas emissions further.</t>
At the most fundamental level, networks (even softwarized ones) involve applianc
es, i.e., equipment that relies on electrical power to perform its function. The
re are two distinct layers with different opportunities for improvement:
<list style="symbols">
<t>Hardware: Reducing embedded carbon during material extraction and manufac
turing, improving energy efficiency and reducing energy consumption during oper
ations, and reuse, repurpose, and recycle motions.</t>
<t>Software: Improving software energy efficiency, maximizing utilization of
processing devices, allowing for software to interact with hardware to improve
sustainability.</t>
</list>
Beyond making network appliances merely more energy-efficient, there are other i
mportant ways in which equipment can help networks become greener. This include
s aspects such as support for port power saving modes or down-speeding of links
to reduce power consumption for resources that are not fully utilized. To fully
tap into the potential of such features requires accompanying management functi
onality, for example to determine when it is "safe" to down-speed a link or ente
r a power saving mode, and manage the network in such a way that conditions to d
o so are maximized.
<br /><br /> </section>
Most importantly from a management perspective, improving sustainability at the <section>
equipment level involves providing management instrumentation that allows to pre <name>Approaching the Problem</name>
cisely monitor and manage power usage and doing so at different levels of granul
arity, for example accounting separately for the contributions of CPU, memory, a
nd different ports. This enables (for example) controller applications to optimi
ze energy usage across the network and that leverage control loops to assess the
effectiveness (e.g. in terms of reduction in power use) of measures that are ta
ken.
<br /><br /> <t>
As a side note, the terms "device" and "equipment", as used in the context of th One way in which gains in network sustainability can be achieved involves reduci
is draft, are used to refer to networking equipment. We are not taking into con ng the amount of energy needed to provide communication services and improving t
sideration end-user devices and endpoints such as mobile phones or computing equ he efficiency with with networks utilize power during their use phase. However,
ipment. for a holistic approach, other aspects need to be considered as well.
</t> </t>
<t> <t>The environmental footprint is not determined by energy consumption a
Protocol level:<br /><br />Energy-efficiency and greenness are aspects that are lone. The sustainability of power sources needs to be considered as well. A de
rarely considered when designing network protocols. This suggests that there ma ployment that includes devices that are less energy efficient but powered by a s
y be plenty of untapped potential. ustainable energy source can arguably be considered greener than a deployment th
Some aspects involve designing protocols in ways that reduce the need for redund at includes highly efficient devices that are powered by diesel generators. In
ant or wasteful transmission of data to allow not only for better network utiliz fact, in the same Telefónica report mentioned earlier, extensive reliance on ren
ation, but greater goodput per unit of energy being consumed. Techniques might ewable energy sources is emphasized.
include approaches that reduce the "header tax" incurred by payloads as well as
methods resulting in the reduction of wasteful retransmissions. Similarly, ther
e may be cases where chattiness of protocols may be preventing equipment from go
ing into sleep mode. Designing protocols that reduce chattiness in such scenari
os, for example, that reduce dependence on periodic updates or heartbeats, may r
esult in greener outcomes. Likewise, aspects such as restructuring addresses in
ways that allow to minimize the size of lookup tables and associated memory size
s and hence energy use can play a role as well. <br /><br />
Another role of protocols concerns the enabling of management functionality to i
mprove energy efficiency at the network level, such as discovery protocols that
allow for quick adaptation to network components being taken dynamically into an
d out of service depending on network conditions, as well as protocols that can
assist with functions such as the collection of energy telemetry data from the n
etwork.
</t> </t>
<t>
Network level<br /><br />Perhaps the greatest opportunities to realize power sav <t>Similarly, deployments can take other environmental factors into acco
ings exist at the level of the network as whole. Many of these opportunities are unt that affect the carbon footprint. For example, deployments where the need f
directly related to management functionality. For example, optimizing energy ef or cooling is reduced or where excessive heat generated by equipment can be put
ficiency may involve directing traffic in such a way that it allows for isolatio to a productive use will be considered greener than deployments where this is no
n of equipment that might not be needed at certain moments so that it can be pow t the case. Examples include deployments in cooler natural surroundings (e.g., i
ered down or brought into power-saving mode. By the same token, traffic should b n colder climates) where that is an option. Likewise, manufacturing and recycli
e directed in a way that requires bringing additional equipment online or out of ng networking equipment are also part of the sustainability equation, as the pro
power-saving mode in cases where alternative traffic paths are available for wh duction itself consumes energy and results in a carbon cost embedded as part of
ich the incremental energy cost would amount to zero. Likewise, some networking the device itself. Extending the lifetime of equipment may in many cases be pref
devices may be rated less "green" and more power-intensive than others or power erable over replacing it earlier with equipment that is slightly more energy eff
ed by less-sustainable energy sources. Their use might be avoided unless during icient, but that requires the embedded carbon cost to be amortized over a much s
periods of peak capacity demands. Generally, incremental carbon emissions can horter period of time.
be viewed as a cost metric that networks should strive to minimize and consider
as part of routing and of network path optimization.
</t> </t>
<t> <t>Network management has an outsized role to play in approaching those
Architecture level<br /><br />The current network architecture supports a wide r problems.
ange of applications, but does not consider energy efficiency as one of its desi To reduce the amount of energy used, network providers
gn parameters. One can argue that the most energy efficient shift of the last tw need to maximize the use of scarce resources and eliminate the use of
o decades has been the deployment of Content Delivery Network overlays: while th resources that are not strictly needed. They need to optimize the way in which
ese were set up to reduce latency and minimize bandwidth consumption, from a net networks are deployed, which resources are placed where, and how equipment life
work perspective, retrieving the content from a local cache is also much greener cycles and upgrades are being managed -- all of which constitute classic operati
. What other architectural shifts can produce energy consumption reduction? onal problems. As best practices, methods, and algorithms are developed, they ne
ed to be automated to the greatest extent possible, migrated over time into the
network, and performed on increasingly short timescales, transcending management
and control planes.
</t> </t>
</list> </section>
<section>
<name>Structuring the Problem Space</name>
<t>
From a technical perspective, multiple vectors along which networks can be made
greener should be considered:
</t> </t>
<t> <ul spacing="normal">
In this document, we will explore each of those vectors in further detail and at <li>
tempt to articulate specific challenges that could make a difference when addres <t>Equipment level:</t>
sed. As our starting point, we borrow some material from a prior paper, <xref t <t>Perhaps the most promising vector for improving networking
arget="GreenNet22"/>. For this document, this material has been both expanded (f sustainability concerns the network equipment itself. At the most
or example, in terms of some of the opportunities) and pruned (for example, in t fundamental level, networks (even softwarized ones) involve
erms of background on prior scholarly work). appliances, i.e., equipment that relies on electrical power to
perform its function. There are two distinct layers with different
opportunities for improvement:</t>
<ul spacing="normal">
<li>
<t>Hardware: Reducing embedded carbon during material extraction
and manufacturing; improving energy efficiency and reducing
energy consumption during operations; and increasing reuse, repurposing,
and
recycling.</t>
</li>
<li>
<t>Software: Improving software energy efficiency, maximizing
utilization of processing devices, and allowing for software to
interact with hardware to improve sustainability.</t>
</li>
</ul>
<t>Beyond making network appliances merely more energy efficient,
there are other important ways in which equipment can help
networks become greener. This includes aspects such as supporting
port power-saving modes or down-speeding links to reduce
power consumption for resources that are not fully utilized.
To fully tap into the potential of such features requires
accompanying management functionality, for example to determine
when it is "safe" to down-speed a link or enter a power saving
mode, and operate the network in such a way that conditions to do
so are maximized.
</t>
<t>Most importantly, from a management perspective, improving
sustainability at the equipment level involves providing
management instrumentation that allows for precise monitoring and
managing power usage and doing so at different levels of
granularity, for example, accounting separately for the
contributions of CPU, memory, and different ports. This enables
(for example) controller applications to optimize energy usage
across the network and to leverage control loops to assess the
effectiveness (e.g., in terms of reducing power use) of
the measures that are taken.</t>
<t>As a side note, the terms "device" and "equipment", as used in
the context of this document, are used to refer to networking
equipment. We are not taking into consideration end-user devices
and endpoints such as mobile phones or computing equipment.</t>
</li>
<li>
<t>Protocol level:</t>
<t>Energy-efficiency and "greenness" are aspects that are rarely
considered when designing network protocols. This suggests that
there may be plenty of untapped potential. Some aspects involve
designing protocols in ways that reduce the need for redundant or
wasteful transmission of data, allowing not only for better network
utilization but for greater goodput per unit of energy being
consumed. Techniques might include approaches that reduce the
"header tax" incurred by payloads as well as methods resulting in
the reduction of wasteful retransmissions. Similarly, there may
be cases where chattiness of protocols may be preventing equipment
from going into sleep mode. Designing protocols that reduce
chattiness in such scenarios, for example, that reduce dependence
on periodic updates or heartbeats, may result in greener
outcomes. Likewise, aspects such as restructuring addresses in
ways that minimize the size of lookup tables,
associated memory sizes, and hence energy use, can play a role as
well.</t>
<t>Another role of protocols concerns the enabling
of management functionality to improve energy efficiency at the
network level, such as discovery protocols that allow for quick
adaptation to network components being taken dynamically into and
out of service depending on network conditions, as well as
protocols that can assist with functions such as the collection of
energy telemetry data from the network.</t>
</li>
<li>
<t>Network level:</t>
<t>Perhaps the greatest opportunities to realize power savings
exist at the level of the network as whole. Many of these
opportunities are directly related to management
functionality. For example, optimizing energy efficiency may
involve directing traffic in such a way that it allows the
isolation of equipment that might not be needed at certain moments
so that it can be powered down or brought into power-saving
mode. By the same token, traffic should be directed in a way that
requires bringing additional equipment online or out of
power-saving mode in cases where alternative traffic paths are
available for which the incremental energy cost would amount to
zero. Likewise, some networking devices may have a lower
sustainability rating, be less energy-efficient, or be powered
less-sustainable energy sources than others. Their use might be avoi
ded
except during periods of peak capacity demands. Generally,
incremental carbon emissions can be viewed as a cost metric that
networks should strive to minimize and consider as part of routing
and network path optimization.</t>
</li>
<li>
<t>Architecture level:</t>
<t>The current network architecture supports a wide range of
applications but does not consider energy efficiency as one of
its design parameters. One can argue that the most energy
efficient shift of the last two decades has been the deployment of
Content Delivery Network overlays: while these were set up to
reduce latency and minimize bandwidth consumption, from a network
perspective, retrieving the content from a local cache is also
much greener. What other architectural shifts can produce energy
consumption reduction?</t>
</li>
</ul>
<t>In this document, we will explore each of those vectors in further de
tail and attempt to articulate specific challenges that could make a difference
when addressed. As our starting point, we borrow some material from "Challenges
and Opportunities in Green Networking" <xref target="GreenNet22"/>. For this do
cument, this material has been both expanded (for example, in terms of some of t
he opportunities) and pruned (for example, in terms of background on prior schol
arly work).
</t> </t>
<t> <t>
This document is a product of the Network Management Research Group (NMRG) of th e Internet Research Task Force (IRTF). This document reflects the consensus of the research group and was discussed and presented multiple times, each time rec eiving positive feedback and no objections. It is not a candidate for any level of Internet Standard and is published for informational purposes. This document is a product of the Network Management Research Group (NMRG) of th e Internet Research Task Force (IRTF). This document reflects the consensus of the research group and was discussed and presented multiple times, each time rec eiving positive feedback and no objections. It is not a candidate for any level of Internet Standard and is published for informational purposes.
</t> </t>
</section>
</section>
<section>
<name>Definitions and Acronyms</name>
<t>Below you find acronyms used in this document: </t>
<dl newline="false" spacing="normal">
<dt>Carbon Footprint:</dt>
<dd>As used in this document, the amount of carbon emissions
associated with the use or deployment of technology, usually
correlated with the amount of energy consumption</dd>
<dt>CDN:</dt>
<dd>Content Delivery Network</dd>
<dt>CPU:</dt>
<dd>Central Processing Unit (that is, the main processor in a server)</d
d>
<dt>DC:</dt>
<dd>Data Center</dd>
<dt>FCT:</dt>
<dd>Flow Completion Time</dd>
<dt>GHG:</dt>
<dd>Greenhouse Gas</dd>
<dt>GPU:</dt>
<dd>Graphical Processing Unit</dd>
<dt>HVAC:</dt>
<dd>Heating, Ventilation, Air Conditioning</dd>
<dt>ICN:</dt>
<dd>Information-Centric Network</dd>
<dt>IGP:</dt>
<dd>Interior Gateway Protocol</dd>
<dt>IoT:</dt>
<dd>Internet of Things</dd>
</section> <dt>LEED:</dt>
</section> <dd>Leadership in Energy and Environmental Design (a green building rati
ng system)</dd>
<dt>LEO:</dt>
<dd>Low Earth Orbit</dd>
<dt>LPM:</dt>
<dd>Longest Prefix Match (a method to look up prefixes in a forwarding e
lement)</dd>
<dt>MPLS:</dt>
<dd>Multiprotocol Label Switching</dd>
<dt>MTU:</dt>
<dd>Maximum Transmission Unit (the largest packet size that can be trans
mitted over a network)</dd>
<dt>NIC:</dt>
<dd>Network Interface Card</dd>
<dt>QoE:</dt>
<dd>Quality of Experience</dd>
<dt>QoS:</dt>
<dd>Quality of Service</dd>
<!--[rfced] Regarding "Definitions and Acronyms":
<section title="Definitions and Acronyms"> b) Regarding the expansion of "QUIC". We were told by the
QUIC WG of the IETF that it is not an acronym. May this entry be
updated as follows or otherwise? Also, would you like to add
RFC 9000 as an informative reference?
<t> Original:
Below you find acronyms used in this draft: QUIC: Quick UDP Internet Connections
<list style="hanging" hangIndent='7'>
<t hangText="Carbon Footprint:"><br />As used in this document, the amount of ca Perhaps:
rbon emissions associated with the use or deployment of technology, usually corr QUIC: the name of a UDP-based, stream-multiplexing, encrypted
elated with the amount of energy consumption</t> transport protocol.
<t hangText="CDN:">Content Delivery Network</t>
<t hangText="CPU:">Central Processing Unit, that is the main processor in a serv
er</t>
<t hangText="DC:">Data Center</t>
<t hangText="FCT:">Flow Completion Time</t>
<t hangText="GHG:">Greenhouse Gas</t>
<t hangText="GPU:">Graphical Processing Unit</t>
<t hangText="HVAC:">Heating, Ventilation, Air Conditioning</t>
<t hangText="ICN:">Information Centric Network</t>
<t hangText="IGP:">Interior Gateway Protocol</t>
<t hangText="IoT:">Internet of Things</t>
<t hangText="IPU:">Infrastructure Processing Units</t>
<t hangText="LEED:">Leadership in Energy and Environmental Design, a green build
ing rating system</t>
<t hangText="LEO:">Low Earth Orbit</t>
<t hangText="LPM:">Longest Prefix Match, a method to look up prefixes in a forwa
rding element</t>
<t hangText="MPLS:">Multi-Path Label Switchin</t>
<t hangText="MTU:">Maximum Transmission Unit, the largest packet size that can b
e transmitted over a network</t>
<t hangText="NIC:">Network Interface Card</t>
<t hangText="QoE:">Quality of Experience</t>
<t hangText="QoS:">Quality of Service</t>
<t hangText="QUIC:">Quick UDP Internet Connections</t>
<t hangText="SNIC:">Smart NIC</t>
<t hangText="SDN:">Software-Defined Networking</t>
<t hangText="TCP:">Transport Control Protocol</t>
<t hangText="TE:">Traffic Engineering</t>
<t hangText="TPU:">Tensor Processing Unit</t>
<t hangText="WAN:">Wide Area Network</t>
</list> author: RE: b) What you say is correct, agree with that, and we can add RFC 9000
</t> as informative reference (referenced from that definition)
</section>
<!-- what is "from that definition"? the original or new?
<section title="The Dynamics of Combining Sustainability and Network Management"
> left snic per Carlos's prefernece
<t>
As it was mentioned and will be further explored in detail, improving energy
efficiency is a key way to improve sustainability. However, this process and re
duction is only meaningful within its larger context.
</t>
<t>
There is a duality and dialectic in this interaction. First, in order to pro
vide those hardware and software improvements, research and development need to
take place, which bring their own carbon footprint. Further, if the improvement
is in a much more energy efficient Chip, CPU, GPU, or piece of hardware, deployi
ng the next generation of hardware (material acquisition, manufacturing, supply-
chain, transportation) and end-of-lifing the older generation creates and added
carbon footprint. Consequently, there are values that maximize sustainability at
a global basis, which are neither at both ends: it is not about leaving old har
dware and software for overextended periods, and it is also not about replacing
and rearchitecting weekly.
</t>
<t>
Similarly, at the protocol and architectural levels there is an equivalent d
uality: what is the "sweet-spot" on which protocols can be improved (for them to
be less energy-wastful), and when to use protocol interactions (e.g., extension
s) to carry sustainability data and information (i.e., sustainability telemetry
and metrics) and to execute actions.
</t>
<t>
Finally, there needs to be given consideration to Jevons paradox, particular
ly as the energy efficiency optimization (in hardware as well as in software) ca
n bring important operational savings in cost.
</t>
</section>
--> -->
<dt>QUIC:</dt>
<dd>the name of a UDP-based, stream-multiplexing, encrypted
transport protocol <xref target="RFC9000"/>.</dd>
<dt>SDN:</dt>
<dd>Software-Defined Networking</dd>
<dt>TCP:</dt>
<dd>Transport Control Protocol</dd>
<dt>TE:</dt>
<dd>Traffic Engineering</dd>
<dt>TPU:</dt>
<dd>Tensor Processing Unit</dd>
<dt>WAN:</dt>
<dd>Wide Area Network</dd>
</dl>
</section>
<section title="Network Energy Consumption Characteristics and Implications" anc <section anchor="sec_implications">
hor="sec:implications"> <name>Network Energy Consumption Characteristics and Implications</name>
<t> <t>
Carbon footprint and, with it, greenhouse gas emissions are determined by severa The carbon footprint and, with it, greenhouse gas emissions are determined by se
l factors. A main factor is network energy consumption, as the energy consumed c veral factors. A main factor is network energy consumption, as the energy consum
an be considered a proxy for the burning of fuels required for corresponding pow ed can be considered a proxy for the burning of fuels required for corresponding
er generation. Network energy consumption by itself does not tell the whole sto power generation. Network energy consumption by itself does not tell the whole
ry, as it does not take the sustainability of energy sources and energy mix into story, as it does not take the sustainability of energy sources and the energy
account. Likewise, there are other factors such as hidden carbon cost reflecti mix into account. Likewise, there are other factors such as the carbon cost exp
ng the carbon footprint expended in manufacturing of networking hardware. Nonet ended in the manufacturing of networking hardware. Nonetheless, network energy
heless, network energy consumption is an excellent predictor for carbon footprin consumption is an excellent predictor of a carbon footprint and its reduction, w
t and its reduction key to sustainable solutions. Exploring possibilities to im hich is key to sustainable solutions. Hence, exploring possibilities to improve
prove energy efficiency is hence a key factor for greener, more sustainable, les energy efficiency is a key factor for greener, more sustainable, less carbon-in
s carbon-intensive networks. tensive networks.
</t> </t>
<t>For this, it is important to understand some of the characteristics of power consumption by networks and which aspects contribute the most. This helps to id entify where the greatest potential not just for power savings but also for sust ainability improvements lies. <t>It is important to understand some of the characteristics of power cons umption by networks and which aspects contribute the most. This helps to identi fy where the greatest potential is, not just for power savings but also for sust ainability improvements.
</t> </t>
<t> <t>
Power is ultimately drawn by devices. Devices are not monoliths but are compose Power is ultimately drawn by devices. Devices are not monoliths but are compose
d of multiple components. The power consumption of the device can be divided int d of multiple components. The power consumption of the device can be divided int
o the consumption of the core device - the backplane and CPU, if you will - as w o the consumption of the core device -- the backplane and CPU, if you will -- as
ell as additional consumption incurred per port and line card. In addition, GP well as additional consumption incurred per port and line card. In addition,
U and TPU may be used as well in the network and may have different power consum the GPU and TPU may be used in the network and may have different power consumpt
ption profiles. ion profiles.
Furthermore, it is important to understand the difference between power consumpt ion when a resource is idling versus when it is under load. This helps to unders tand the incremental cost of additional transmission versus the initial cost of transmission. Furthermore, it is important to understand the difference between power consumpt ion when a resource is idling versus when it is under load. This helps to unders tand the incremental cost of additional transmission versus the initial cost of transmission.
</t> </t>
<t> <t>
In typical networking devices, only roughly half of the energy consumption is as sociated with the data plane <xref target="Bolla2011energy"/>. In typical networking devices, only roughly half of the energy consumption is as sociated with the data plane <xref target="Bolla2011energy"/>.
An idle base system typically consumes more than half of the energy over the sam e system running at full load <xref target="Chabarek08"/>, <xref target="Cervero 19"/>. An idle base system typically consumes more than half of the energy that the sam e system would consume when running at full load <xref target="Chabarek08"/> <xr ef target="Cervero15"/>.
Generally, the cost of sending the first bit is very high, as it requires poweri ng up a device, port, etc. Generally, the cost of sending the first bit is very high, as it requires poweri ng up a device, port, etc.
The incremental energy cost of transmission of additional bits (beyond the first ) is many orders of magnitude lower. Likewise, the incremental cost of increment al CPU and memory needed to process additional packets becomes fairly negligible . The incremental energy cost of transmission of additional bits (beyond the first ) is many orders of magnitude lower. Likewise, the incremental cost of the incre mental CPU and memory needed to process additional packets becomes fairly neglig ible.
</t> </t>
<t> This means that a device's energy consumption does not increase linearly wit h the volume of forwarded traffic. Instead, it resembles more of a step functio n in which energy consumption stays roughly the same up to a certain volume of t raffic, followed by a sudden jump when additional resources need to be procured to support a higher volume of traffic. <t> This means that a device's energy consumption does not increase linear ly with the volume of forwarded traffic. Instead, it resembles a step function in which energy consumption stays roughly the same up to a certain volume of tra ffic, followed by a sudden jump when additional resources need to be procured to support a higher volume of traffic.
</t> </t>
<t> <t>
By the same token, it is generally more energy-efficient to transmit a large vol By the same token, it is generally more energy efficient to transmit a large vol
ume of data in one burst (and subsequently turning off or down-speeding the inte ume of data in one burst (and subsequently turn off or down-speed the interface
rface when idling) than to continuously transmit at a lower rate. In that sense when idling) than to continuously transmit at a lower rate. In that sense, it ca
it can be the duration of the transmission that dominates the energy consumption n be the duration of the transmission that dominates the energy consumption -- n
, not the actual data rate. ot the actual data rate.
</t> </t>
<t> <t>
The implications on green networking from an energy-savings standpoint are signi ficant. The implications on green networking from an energy-savings standpoint are signi ficant.
Of utmost importance are schemes that allow for "peak shaving": networks are typ ically dimensioned for periods of peak demand and usage, yet any excess capacity during periods of non-peak usage does not result in corresponding energy saving s. Peak shaving techniques that allow to reduce peak traffic spikes and thus wa ste during non-peak periods may result in outsize sustainability gains. Peak sh aving could be accomplished by techniques such as spreading spikes out over geog raphies (e.g. routing traffic across more costly but less utilized routes) or ov er time (e.g. postponing and buffering non-urgent traffic). Of utmost importance are schemes that allow for "peak shaving": networks are typ ically dimensioned for periods of peak demand and usage, yet any excess capacity during periods of non-peak usage does not result in corresponding energy saving s. Peak shaving techniques that reduce peak traffic spikes and waste during non -peak periods may result in outsize sustainability gains. Peak shaving could be accomplished by techniques such as spreading spikes out over geographies (e.g., routing traffic across more costly but less utilized routes) or over time (e.g. , postponing and buffering non-urgent traffic).
</t> </t>
<t>Likewise, large gains can be made whenever network resources can effectively be taken offline for at least some of the time, managing networks in a way that enables resources to be removed from service so they can be powered down (or put into a more energy-saving state, such as when down-speeding ports) while not ne eded. Of course, any such methods need to take into account the overhead of tak ing resources offline and bringing them back online. This typically takes some amount of time, requiring accurate predictive capabilities to avoid situations i n which network resources are not available at times when they would be needed. In addition, there is additional overhead such as synchronization of state to b e accounted for. <t>Likewise, large gains can be made whenever network resources can effect ively be taken offline for at least some of the time, managing networks in a way that enables resources to be removed from service so they can be powered down ( or put into a more energy-saving state, such as when down-speeding ports) while not needed. Of course, any such methods need to take into account the overhead of taking resources offline and bringing them back online. This typically takes some amount of time, requiring accurate predictive capabilities to avoid situat ions in which network resources are not available at times when they would be ne eded. In addition, there is additional overhead, such as synchronization of sta te, to be accounted for.
</t> </t>
<t> <t>
At the same time, any non-idle resources should be utilized to the greatest exte At the same time, any non-idle resources should be utilized to the greatest exte
nt possible as the incremental energy cost is negligible. Of course, this needs nt possible, as the incremental energy cost is negligible. Of course, this needs
to occur while still taking other operational goals into consideration, such as to occur while still taking other operational goals into consideration, such as
protection against failures (allowing for readily available redundancy and spare protection against failures (allowing for readily available redundancy and spar
capacity in case of failure) and load balancing (for increased operational robu e capacity in case of failure) and load balancing (for increased operational rob
stness). ustness).
As data transmission needs tend to fluctuate wildly and occur in bursts, any opt As data transmission needs tend to fluctuate wildly and occur in bursts, any opt
imization schemes need to be highly adaptable and allow for control loops at ver imization schemes need to be highly adaptable and allow control loops at very fa
y fast time scales. st time scales.
</t> </t>
<t> <t>
Similarly, for applications where this is possible, it may be desirable to repla Similarly, for applications where this is possible, it may be desirable to repla
ce continuous traffic at low data rates with traffic that is sent in burst at hi ce continuous traffic at low data rates with traffic that is sent in bursts at h
gh data rates in order to potentially maximize the time during which resources c igh data rates in order to potentially maximize the time during which resources
an be idled. can be idled.
</t> </t>
<!--[rfced] For readability, may this sentence be rephrased as follows?
Specifically, we suggest rephrasing "emphasis needs to be given to technology"
and making the sentence into shorter ones.
<t> Original:
As a result, emphasis needs to be given to technology that allows for example to As a result, emphasis needs to be given to technology that allows for
(at the device level) exercise very efficient and rapid discovery, monitoring, example to (at the device level) exercise very efficient and rapid
and control of networking resources so that they can be dynamically be taken off discovery, monitoring, and control of networking resources so that
line or back into service, without (at the network level) requiring extensive co they can be dynamically be taken offline or back into service,
nvergence of state across the network or recalculation of routes and other optim without (at the network level) requiring extensive convergence of
ization problems, and (at the network equipment level) support rapid power cycle state across the network or recalculation of routes and other
and initialization schemes. There may be some lessons that can be applied here optimization problems, and (at the network equipment level) support
from IoT, rapid power cycle and initialization schemes.
which has long had to contend with power-constrained end devices that need to sp
end much of their time in power saving states to conserve battery.
</t>
</section>
<section title="Challenges and Opportunities - Equipment Level" anchor="sec:devi Perhaps:
ce"> As a result, we should prioritize technology that efficiently
discovers, monitors, and controls networking resources at the device
level. This allows devices to be dynamically taken offline or brought
back online without extensive network-level state convergence, route
recalculation, or other complex optimizations. At the network equipment
level, it should also support rapid power cycling and initialization.
Or:
Technology that has the following characteristics should be prioritized:
* At the device level, it does efficient and rapid discovery, monitoring,
and control of networking resources so they can be dynamically taken
offline or online.
* At the network level, it does not require extensive convergence of
state across the network or recalculation of routes and other
optimization problems.
* At the network equipment level, it supports rapid power cycle and
initialization schemes.
updated per author request, but perhaps s/emphasis/priority?
-->
<t> <t>
As a result, emphasis needs to be given to technology that enables, for
example, very efficient and rapid
discovery, monitoring, and control of networking resources.
This allows devices to be dynamically taken offline or brought
back online without extensive network-level state convergence, route
recalculation, or other complex optimizations at the network level.
To facilitate such schemes, rapid power cycle and initialization schemes shoul
d be supported at the device level. There may be some lessons that can be appl
ied here from IoT,
which has long had to contend with power-constrained end devices that need to sp
end much of their time in power-saving states to conserve battery.
</t>
</section>
<section anchor="sec_device">
<name>Challenges and Opportunities - Equipment Level</name>
<t>
We are categorizing challenges and opportunities to improve sustainability at th e network equipment level along the following lines: We are categorizing challenges and opportunities to improve sustainability at th e network equipment level along the following lines:
<list style="symbols">
<t>Hardware and manufacturing. Related opportunities are arguably among the most
obvious and perhaps "largest". However, solutions here lie largely outside the
scope of networking researchers. </t>
<t>Visibility and instrumentation. Instrumenting equipment to provide visibility
into how they consume energy is key to management solutions and control loops t
o facilitate optimization schemes. </t>
</list>
</t> </t>
<section title="Hardware and Manufacturing"> <ul spacing="normal">
<t> <li>
<t>Hardware and manufacturing: Related opportunities are arguably amon
Perhaps the most obvious opportunities to make networking technology more energy g the most obvious and perhaps "largest". However, solutions here lie largely ou
efficient exist at the equipment level. After all, networking involves physica tside the scope of networking researchers. </t>
l equipment to receive and transmit data. Making such equipment more power effi </li>
cient, have it dissipate less heat to consume less energy and reduce the need fo <li>
r cooling, making it eco-friendly to deploy, sourcing sustainable materials and <t>Visibility and instrumentation: Instrumenting equipment to provide
facilitating recycling of equipment at the end of its lifecycle all contribute t visibility into how they consume energy is key to management solutions and contr
o making networks greener. More specific and unique to networking are schemes t ol loops to facilitate optimization schemes. </t>
o reduce energy usage of transmission technology from wireless (antennas) to opt </li>
ical (lasers). </ul>
<section anchor="sec_hw_and_man">
<name>Hardware and Manufacturing</name>
<t>
Perhaps the most obvious opportunities to make networking technology more energy
efficient exist at the equipment level. After all, networking involves physica
l equipment to receive and transmit data. Making such equipment more power ef
ficient, having it dissipate less heat to consume less energy and reduce the nee
d for cooling, sourcing sustainable materials, and facilitating the recycling of
equipment at the end of its lifecycle all contribute to making networks greener
. Reducing the energy usage of transmission technology, from wireless (antennas)
to optical (lasers), is a strategy that is unique to networking.
</t> </t>
<t>One critical aspect of the energy cost of networking is the cost to manufactu re and deploy the networking equipment. In addition, even the development proces s itself comes with its own carbon footprint. This is outside of the scope of t his document: we only consider the energy cost of running the network that appli es during the operational part of the equipment's lifecycle. However, a holistic approach would include into this the embedded energy that is included in the ne tworking equipment. As part of this, aspects such as the impact of deploying new protocols on the rate of obsolescence of existing equipment should be considere d. For instance, incremental approaches that do not require to replace equipment right away - or even extend the lifetime of deployed equipment - would have a l ower energy footprint. This is one important benefit also of technologies such as Software-Defined Networking and Network Function Virtualization, as they may allow support of new networking features through software updates without requir ing hardware replacements. <t>One critical aspect of the energy cost of networking is the cost to m anufacture and deploy the networking equipment. In addition, even the developmen t process itself comes with its own carbon footprint. This is outside of the sc ope of this document: we only consider the energy cost of running the network du ring the operational part of the equipment's lifecycle. However, a holistic appr oach would include the embedded energy that is included in the networking equipm ent. As part of this, aspects such as the impact of deploying new protocols on t he rate of obsolescence of existing equipment should be considered. For instance , incremental approaches that do not require replacing equipment right away -- o r even that extend the lifetime of deployed equipment -- would have a lower carb on footprint. This is one important benefit also of technologies such as Softwa re-Defined Networking and network function virtualization, as they may allow sup port for new networking features through software updates without requiring hard ware replacements.
</t> </t>
<t>An attempt to compute not only the energy of running a network, but also the energy embedded into manufacturing the equipment is described in <xref target="E mergy"/> . This is denoted by "emergy", a portmanteau for embedded energy. Likew ise, an approach to recycling equipment and a proof of concept using old cell ph ones recycled into a "junkyard" data center are described in <xref target="Junky ard"/>. <t><xref target="Emergy"/> describes an attempt to compute not only the energy of running a network but also the energy embedded into manufacturing the equipment. This is denoted by "emergy", a portmanteau for embedded energy. Likew ise, <xref target="Junkyard"/> describes an approach to recycling equipment and a proof of concept using old mobile phones recycled into a "junkyard" data cente r.
</t> </t>
<t>One trade-off to consider at this level is the selection of a platform that c an be hardware-optimized for energy efficiency vs a platform that is versatile a nd can run multiple functions. For instance, a switch could run on an efficient hardware platform, or run as a software module (container) over some multi-purpo se platform. While the first one is operationally more energy efficient, it may have a higher embedded energy from a smaller scale, less efficient production pr ocess, as well as a shorter shelf life once new functions need to be added to th e platform. <t>One trade-off to consider at this level is the selection of a platfor m that can be hardware-optimized for energy efficiency versus a platform that is versatile and can run multiple functions. For instance, a switch could run on a n efficient hardware platform or run as a software module (container) over some multipurpose platform. While the first one is operationally more energy efficien t, it may have a higher embedded energy from a smaller scale, a less efficient p roduction process, as well as a shorter shelf life once new functions need to be added to the platform.
</t> </t>
</section> </section>
<section title="Visibility and Instrumentation" anchor="sec:instrum"> <section anchor="sec_instrum">
<t> <name>Visibility and Instrumentation</name>
Beyond "first-order" opportunities as outlined in the previous subsection, netwo <t>
rk equipment just as importantly plays an important role to enable and support g Beyond "first-order" opportunities, as outlined in <xref target="sec_hw_and_man"
reen networking at other levels. Of prime importance is the equipment's ability />, network equipment just as importantly plays a role in enabling and supportin
to provide visibility to management and control plane into its current energy us g green networking at other levels. Of prime importance is the equipment's abili
age. Such visibility enables control loops for energy optimization schemes, allo ty to provide visibility to the management and control planes into its current e
wing applications to obtain feedback regarding the energy implications of their nergy usage. Such visibility enables control loops for energy optimization schem
actions, from setting up paths across the network that require the least increme es, allowing applications to obtain feedback regarding the energy implications o
ntal amount of energy to quantifying metrics related to energy cost used to opti f their actions, from setting up paths across the network that require the least
mize forwarding decisions. Absent an actual measurement of energy usage (and unt incremental amount of energy to quantifying metrics related to energy cost to o
il such measurement is put in place), the network equipment could advertise some ptimize forwarding decisions. Absent an actual measurement of energy usage (and
proxy of its power consumption (say, a labelling scheme as silver, gold, platin until such measurement is put in place), the network equipment could advertise s
um similar to the LEED sustainability metric in building codes or the Energy Sta ome proxy of its power consumption. For example, it could use a labeling scheme
r label in home appliances; or a description of the type of the device as using of silver, gold, or platinum similar to the LEED sustainability metric in buildi
CPU vs GPU vs TPU processors with different power profiles). ng codes, or the Energy Star label in home appliances, or a description of the t
ype of the device as using CPU vs. GPU vs. TPU processors with different power p
rofiles.
</t> </t>
<t> <t>
One prerequisite to such schemes is to have proper instrumentation in place that One prerequisite to such schemes is to have proper instrumentation in place that
allows to monitor current power consumption at the level of networking devices allows for monitoring current power consumption at the level of networking devi
as a whole, line cards, and individual ports. Such instrumentation should also ces as a whole, line cards, and individual ports. Such instrumentation should a
allow to assess the energy efficiency and carbon footprint of the device as a wh lso allow for assessing the energy efficiency and carbon footprint of the device
ole. In addition, it will be desirable to relate this power consumption to data as a whole. In addition, it will be desirable to relate this power consumption
rates as well as to current traffic, for example, to indicate current energy con to data rates and to current traffic, for example, to indicate current energy co
sumption relative to interface speeds, as well as for incremental energy consump nsumption relative to interface speeds, as well as for incremental energy consum
tion that is expected for incremental traffic (to aid control schemes that aim t ption that is expected for incremental traffic (to aid control schemes that aim
o "shave" power off current services or to minimize the incremental use of power to "shave" power off current services or to minimize the incremental use of powe
for additional traffic). This is an area where the current state of the art is r for additional traffic). This is an area where the current state of the art i
sorely lacking and standardization lags behind. For example, as of today, stan s sorely lacking, and standardization lags behind. For example, as of today, st
dardized YANG data models <xref target="RFC7950"/> for network energy consumptio andardized YANG data models <xref target="RFC7950"/> for network energy consumpt
n that can be used in conjunction with management and control protocols have yet ion that can be used in conjunction with management and control protocols have y
to be defined. et to be defined.
</t> </t>
<t>To remedy this situation, efforts to define sets of green networking metrics <xref target="I.D.draft-cx-green-green-metrics"/> as well as YANG data models th at include objects that provide visibility into power measurements (e.g. <xref t arget="I.D.draft-li-green-power"/>) are getting underway. Agreed sets of metric s and corresponding data models will provide the basis for further steps, beginn ing with their implementation as part of management and control instrumentation. <t>To remedy this situation, efforts to define sets of green networking metrics <xref target="I-D.cx-green-green-metrics"/> as well as YANG data models that include objects that provide visibility into power measurements (e.g., <xre f target="I-D.li-green-power"/>) were underway in 2024. Agreed sets of metrics and corresponding data models will provide the basis for further steps, beginnin g with their implementation as part of management and control instrumentation.
</t> </t>
<t> <t>
Instrumentation should also take into account the possibility of virtualization, Instrumentation should also take into account the possibility of virtualization,
introducing layers of indirection to assess the actual energy usage. For exampl introducing layers of indirection to assess the actual energy usage. For exampl
e, virtualized networking functions could be hosted on containers or virtual mac e, virtualized networking functions could be hosted on containers or virtual mac
hines which are hosted on a CPU in a data center instead of a regular network ap hines that are hosted on a CPU in a data center instead of a regular network app
pliance such as a router or a switch, leading to very different power consumptio liance such as a router or a switch, leading to very different power consumption
n characteristics. For example, a data center CPU could be more power efficient characteristics. For example, a data center CPU's power consumption could be mo
and consume power more proportionally to actual CPU load. Instrumentation needs re efficient and more proportional to actual CPU load. Instrumentation needs to
to reflect these facts and facilitate attributing power consumption in a correct reflect these facts and facilitate attributing power consumption in a correct ma
manner. nner.
</t> </t>
<t> <t>
Beyond monitoring and providing visibility into power consumption, control knobs Beyond monitoring and providing visibility into power consumption, control knobs
are needed to configure energy saving policies. For instance, power saving mode are needed to configure energy-saving policies. For instance, power-saving mode
s are common in endpoints (such as mobile phones or notebook computers) but sore s are common in endpoints (such as mobile phones or notebook computers) but sore
ly lacking in networking equipment. ly lacking in networking equipment.
</t> </t>
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Basic equipment categorization as "energy-efficient" (or not) as a fir
st step to identify immediate potential improvements, akin to the EnergyStar pro
gram from the US's Environmental Protection Agency. </t>
<t>Equipment instrumentation advances for improved energy-awareness; definit
ion and standardization of granular management information.</t>
<t>Virtualized energy and carbon metrics and assessment of their effectivene
ss in solutions that optimize carbon footprint also in virtualized environments
(including SDN, network slicing, network function virtualization, etc.). </t>
<t>Certification and compliance assessment methods that ensure that green in
strumentation cannot be manipulated to give false and misleading data.</t>
<t>Methods that allow to account for energy mix powering equipment, to facil
itate solutions that optimize carbon footprint and minimize pollution beyond mer
e energy efficiency <xref target="Hossain2019"/>.</t>
</list>
</t> </t>
</section> <ul spacing="normal">
</section> <li>
<section title="Challenges and Opportunities - Protocol Level" anchor="sec:proto <t>Basic equipment categorization as "energy efficient" (or not) as
col"> a first step to identify immediate potential improvements, akin to the Energy St
<t> ar program from the US's Environmental Protection Agency. </t>
There are several opportunities to improve network sustainability at the protoco </li>
l level. We characterize them along several categories. The first and arguably <li>
most impactful category concerns protocols that enable carbon footprint optimiza <t>Equipment instrumentation advances for improved energy awareness;
tion schemes at the network level and management towards those goals. Other cate definition and standardization of granular management information.</t>
gories concern protocols designed to optimize data transmission rates under ener </li>
gy considerations, protocols designed to reduce the volume of data to be transmi <li>
tted, and protocol aspects related to network addressing schemes. While those ca <t>Virtualized energy and carbon metrics and assessment of their eff
tegories may be less impactful, even areas with smaller gains should not be left ectiveness in solutions that optimize carbon footprints in virtualized environme
unexplored. nts (including SDN, network slicing, network function virtualization, etc.). </t
>
</li>
<li>
<t>Certification and compliance assessment methods that ensure that
green instrumentation cannot be manipulated to give false and misleading data.</
t>
</li>
<li>
<t>Methods that allow for the energy mix of the power sources that a
re used to power equipment to be taken into account, in order to facilitate solu
tions that optimize the actual carbon footprint and minimize pollution beyond me
re energy efficiency <xref target="Hossain2019"/>.</t>
</li>
</ul>
</section>
</section>
<section anchor="sec_protocol">
<name>Challenges and Opportunities - Protocol Level</name>
<t>
There are several opportunities to improve network sustainability at the protoco
l level, which can be categorized as follows. The first and arguably most impact
ful category concerns protocols that enable carbon footprint optimization scheme
s at the network level and management towards those goals. Other categories conc
ern protocols designed to optimize data transmission rates under energy consider
ations, protocols designed to reduce the volume of data to be transmitted, and p
rotocol aspects related to network addressing schemes. While those categories ma
y be less impactful, even areas with smaller gains should be explored.
</t> </t>
<t> <t>
There is also substantial work in the area of IoT, which has had to contend with There is also substantial work in the area of IoT, which has had to contend with
energy-constrained devices for a long time. Much of that work was motivated not energy-constrained devices for a long time. Much of that work was motivated not
by sustainability concerns but practical concerns such as battery life. However , by sustainability concerns but practical concerns such as battery life. However ,
many aspects appear to also apply in the context of sustainability, such as redu cing many aspects appear to also apply in the context of sustainability, such as redu cing
chattiness to allow IoT equipment to go into low-power mode. Accordingly, there is chattiness to allow IoT equipment to go into low-power mode. Accordingly, there is
opportunity to extend IoT work to more generalized scenarios. The an opportunity to extend IoT work to more generalized scenarios. The
use of power-constrained protocols into the wider Internet happens use of power-constrained protocols in the wider Internet happens
regularly. For instance, ARM-based chipsets initially designed for regularly. For instance, ARM-based chipsets initially designed for
energy-efficiency in battery-operated mobile devices have been energy efficiency in battery-operated mobile devices have been
embraced in data centers for a similar trajectory. embraced in data centers for a similar trajectory.
</t> </t>
<section title="Protocol Enablers for Carbon Optimization Mechanisms" anchor="Ne <section anchor="NetworkEnergySaving">
tworkEnergySaving"> <name>Protocol Enablers for Carbon Optimization Mechanisms</name>
<t> <t>
As will be discussed in <xref target="sec:network"/>, energy-aware and pollution As discussed in <xref target="sec_network"/>, energy-aware and pollution-aware s
-aware schemes can help improve network sustainability but require awareness of chemes can help improve network sustainability but require awareness of related
related data. To facilitate such schemes, protocols are needed that are able to data. To facilitate such schemes, protocols are needed that are able to discove
discover what links are available along with their energy efficiency. For insta r what links are available along with their energy efficiency. For instance, lin
nce, links may be turned off in order to save energy and turned back on based up ks may be turned off in order to save energy and turned back on based upon the e
on the elasticity of the demand. Protocols should be devised to discover when th lasticity of the demand. Protocols should be devised to discover when this happe
is happens, and to have a view of the topology that is consistent with frequent ns and to have a dynamic view of the topology that keeps up with frequent update
topology updates due to power cycling of the network resources. s due to power cycling of the network resources.
</t> </t>
<t> <t>
Also, protocols are required to quickly converge onto an energy-efficient path o Also, protocols are required to quickly converge onto an energy-efficient path o
nce a new topology is created by turning links on/off. Current routing protocols nce a new topology is created by turning links on/off. Current routing protocols
may provide for fast recovery in the case of failure. However, failures are hop may provide for fast recovery in the case of failure. However, failures are hop
efully relatively rare events, while we expect an energy efficient network to ag efully relatively rare events, while we expect an energy-efficient network to ag
gressively try to turn off links. There may be synergies with time-variant rout gressively try to turn off links. There may be synergies with Time-Variant Rout
ing <xref target="I.D.draft-ietf-tvr-requirements"/> that can be explored, in wh ing <xref target="I-D.ietf-tvr-requirements"/> that can be explored, in which th
ich topology varies over time with nodes and links turned on or off according to e topology varies over time with nodes and links turned on or off according to a
a schedule. There may even be overlaps in use cases, for example in cases where schedule. There may be overlaps in use cases, for example, when regular changes
regular changes in the energy mix (and hence carbon footprint) of some nodes oc in the energy mix (and hence carbon footprint) of some nodes occur that coincid
cur that coincide with the time of day (such as switching from solar to fossil f e with the time of day (such as switching from solar to fossil fuels at night).
uels at night).
</t> </t>
<t> <t>
Some mechanism is needed to present to the management layer a view of the networ Some mechanism is needed to present to the management layer a view of the networ
k that identifies opportunities to turn resources off (routers/links) while stil k that identifies opportunities to turn off resources (e.g., routers or links) w
l providing an acceptable level of Quality of Experience (QoE) to the users. Thi hile still providing an acceptable level of Quality of Experience (QoE) to the u
s gets more complex as the level of QoE shifts from the current Best Effort deli sers. This gets more complex as the level of QoE shifts from the current best-ef
very model to more sophisticated mechanisms with, for instance, latency, bandwid fort delivery model to more sophisticated mechanisms with, for instance, latency
th or reliability guarantees. , bandwidth, or reliability guarantees.
</t> </t>
<t>Similarly, schemes might be devised in which links across paths with a favora ble energy mix are preferred over other paths. This implies that the discovery o f topology should be able support corresponding parameters. More generally spea king, any mechanism that provides applications with network visibility is a cand idate for scrutinization as to whether it should be extended to provide support for sustainability-related parameters. <t>Similarly, schemes might be devised in which links across paths with a favorable energy mix are preferred over other paths. This implies that the dis covery of topology should be able support corresponding parameters. More genera lly speaking, any mechanism that provides applications with network visibility i s a candidate for scrutinization as to whether it should be extended to provide support for sustainability-related parameters.
</t> </t>
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Protocol advances to enable rapidly taking down, bring back online, and d
iscover availability and power saving status of networking resources while minim
izing the need for reconvergence and propagation of state.</t>
<t>An assessment of which protocols could be extended with energy- and susta
inability-related parameters in ways that would enable "greener" networking solu
tions, and exploration of those solutions. </t>
</list>
</t> </t>
<!-- CW: QoS, QoE and then EoS Energy of Service? --> <ul spacing="normal">
<li>
<t>Protocol advances to enable rapidly taking down, bringing back on
line, and discovering availability and power-saving status of networking resourc
es while minimizing the need for reconvergence and propagation of state.</t>
</li>
<li>
<t>An assessment of which protocols could be extended with energy- a
nd sustainability-related parameters in ways that would enable greener networkin
g solutions, and an exploration of those solutions. </t>
</li>
</ul>
</section> </section>
<!-- End of Enabling Energy Saving Mechanisms subsection-->
<section title="Protocol Optimization" anchor="TrafficAdaptation"> <section anchor="TrafficAdaptation">
<t> <name>Protocol Optimization</name>
The second category involves designing protocols in such a way that the rate of <t>
transmission is chosen to maximize energy efficiency. For example, Traffic Engi The second category involves designing protocols in such a way that the rate of
neering (TE) can be manipulated to impact the rate adaptation mechanism <xref ta transmission is chosen to maximize energy efficiency. For example, Traffic Engi
rget="Ren2018jordan"/>. By choosing where to send the traffic, TE can artificial neering (TE) can be manipulated to impact the rate adaptation mechanism <xref ta
ly congest links so as to trigger rate adaptation and therefore reduce the total rget="Ren2018jordan"/>. By choosing where to send the traffic, TE can artificial
amount of traffic. Most TE systems attempt to minimize Maximal Link Utilization ly congest links so as to trigger rate adaptation and therefore reduce the total
(MLU) but energy saving mechanisms could decide to do the opposite (maximize mi amount of traffic. Most TE systems attempt to minimize Maximum Link Utilization
nimal link utilization) and attempt to turn off some resources to save power. but energy-saving mechanisms could decide to do the opposite (i.e., maximize Mi
nimum Link Utilization) and attempt to turn off some resources to save power.
</t> </t>
<t>Another example is to set up the proper rate of transmission to minimize the flow completion time (FCT) so as to enable opportunities to turn off links. In a wireless context, <xref target="TradeOff" /> studies how setting the proper ini tial value for the congestion window can reduce the FCT and therefore allow the equipment to go faster into a low-energy mode. By sending the data faster, the e nergy cost can be significantly reduced. This is a simple proof of concept, but protocols that allow for turning links into a low-power mode by transmitting the data over shorter periods could be designed for other types of networks beyond Wi-Fi access. This should be done carefully: in the limit, a high rate of transm ission over a short period of time may create bursts that the network would need to accommodate, with all attendant complications of bursty traffic. We conjectu re there is a sweet spot between trying to complete flows faster while controlli ng for burstiness in the network. It is probably advisable to attempt to send tr affic paced yet in bulk rather than spread out over multiple round trips. This i s an area of worthwhile exploration. <t>Another example is to set up the proper rate of transmission to minim ize the flow completion time (FCT) so as to enable opportunities to turn off lin ks. In a wireless context, <xref target="TradeOff"/> studies how setting the pro per initial value for the congestion window can reduce the FCT and therefore all ow the equipment to go faster into a low-energy mode. By sending the data faster , the energy cost can be significantly reduced. This is a simple proof of concep t, but protocols that allow for turning links into a low-power mode by transmitt ing the data over shorter periods could be designed for other types of networks beyond Wi-Fi access. This should be done carefully: a sudden very high rate of t ransmission over a short period of time may create bursts that the network would need to accommodate, with all attendant complications of bursty traffic. We con jecture there is a sweet spot between trying to complete flows faster while cont rolling for burstiness in the network. It is probably advisable to attempt to se nd traffic paced yet in bulk rather than spread out over multiple round trips. T his is an area of worthwhile exploration.
</t> </t>
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Protocol advances that allow greater control over traffic pacing to accou
nt for fluctuations in carbon cost, i.e., control knobs to "bulk up" transmissio
n over short periods or to smoothen it out over longer periods. </t>
<t>Protocol advances that allow to optimize link utilization according to di
fferent goals and strategies (including maximizing minimal link utilization vs m
inimizing maximal link utilization, etc.)</t>
<t>Assessments of the carbon impact of such strategies.</t>
</list>
</t> </t>
<!-- CW: most transport protocols attempt to find the operating point where traf <ul spacing="normal">
fic is sent without dropping packet. TCPs seesaw mechanism is rather crude for t <li>
his purpose. New protocols such as BBR should be more efficient in that respect. <t>Protocol advances that allow greater control over traffic pacing
I feel this section can be beefed up a bit. Also we can circle back to your poi to account for fluctuations in carbon cost, i.e., control knobs to "bulk up" tra
nt in the intro that bursty transmission of traffic is more energy efficient. Th nsmission over short periods or to smooth it out over longer periods. </t>
is is true at the edge only, I would assume, since bursts are multiplexed in the </li>
core. That said: having some burst when there are few flows, and then some more <li>
fluid behavior when there are many could be suggested. For instance, having som <t>Protocol advances for optimizing link utilization according to di
e "energy gateway" that buffers the burst from the domain with few flows and the fferent goals and strategies (including maximizing Minimal Link Utilization vs.
n transmit them to the domain with more flows, and back on the other end of the minimizing Maximal Link Utilization, etc.)</t>
connection --> </li>
<li>
<t>Assessments of the carbon impact of such strategies.</t>
</li>
</ul>
</section> </section>
<section title="Data Volume Reduction" anchor="DataVolume"> <section anchor="DataVolume">
<t> <name>Data Volume Reduction</name>
The first category involves designing protocols in such a way that they reduce t
he volume of data that needs to be transmitted for any given purpose. Loosely sp <t>
eaking, by reducing this volume, more traffic can be served by the same amount o The third category involves designing protocols in such a way that they reduce t
f networking infrastructure, hence reducing overall energy consumption. Possibil he volume of data that needs to be transmitted for any given purpose. Loosely sp
ities here include protocols that avoid unnecessary retransmissions. At the app eaking, by reducing this volume, more traffic can be served by the same amount o
lication layer, protocols may also use coding mechanisms that encode information f networking infrastructure, hence reducing overall energy consumption. Possibil
close to the Shannon limit. Currently, most of the traffic over the Internet c ities here include protocols that avoid unnecessary retransmissions. At the app
onsists of video streaming and encoders for video are already quite efficient an lication layer, protocols may also use coding mechanisms that encode information
d keep improving all the time, resulting in energy savings as one of many advant close to the Shannon limit <xref target="Shannon"/>. Currently, most of the tra
ages (of course being offset by increasingly higher resolution). However, it is ffic over the Internet consists of video streaming. Video encoders are already q
not clear that the extra work to achieve higher compression ratios for the paylo uite efficient and keep improving all the time. This results in energy savings
ads results in a net energy gain: what is saved over the network may be offset b as one of many benefits, even if some of those savings may be partially offset b
y the compression/decompression effort. Further research on this aspect is neces y increasingly higher resolution. It is not clear that the extra work to achieve
sary. higher compression ratios for the payloads results in a net energy gain: what i
s saved over the network may be offset by the compression/decompression effort.
Further research on this aspect is necessary.
</t> </t>
<t>At the transport protocol layer, TCP and to some extent QUIC react to congest ion by dropping packets. This is a highly energy inefficient method to signal co ngestion, since the network has to wait one RTT to be aware that the congestion has occurred, and since the effort to transmit the packet from the source up unt il it is dropped ends up being wasted. This calls for new transport protocols t hat react to congestion without dropping packets. ECN <xref target="RFC2481"/> i s a possible solution, however, it is not widely deployed. DC-TCP <xref target=" Alizadeh2010DCTCP"/> is tuned for Data Centers, L4S is an attempt to port simila r functionality to the Internet <xref target="RFC9330"/>. Qualitative Communicat ion <xref target="QUAL"/> <xref target="Westphal2021qualitative"/> allows the no des to react to congestion by dropping only some of the data in the packet, ther eby only partially wasting the resource consumed by transmitted the packet up to this point. Novel transport protocols for the WAN can ensure that no energy is wasted transmitting packets that will be eventually dropped. <t>At the transport protocol layer, TCP and to some extent QUIC react to congestion by dropping packets. This is an extremely energy inefficient method to signal congestion because (a) the network has to wait one RTT to be aware tha t the congestion has occurred, and (b) the effort to transmit the packet from th e source up until it is dropped ends up being wasted. This calls for new transp ort protocols that react to congestion without dropping packets. ECN <xref targe t="RFC3168"/> is a possible solution, however, it is not widely deployed. DCTCP <xref target="Alizadeh2010DCTCP"/> is tuned for data centers; Low Latency, Low L oss, and Scalable Throughput (L4S) is an attempt to port similar functionality t o the Internet <xref target="RFC9330"/>. Qualitative Communication <xref target= "QUAL"/> <xref target="Westphal2021qualitative"/> allows the nodes to react to c ongestion by dropping only some of the data in the packet, thereby only partiall y wasting the resource consumed by transmitted the packet up to that point. Nov el transport protocols for the WAN can ensure that no energy is wasted transmitt ing packets that will be eventually dropped.
</t> </t>
<t>Another solution to reduce the bandwidth of network protocols by reducing the ir header tax, for example applying header compression. An example in IETF is <x ref target="RFC3095"/>. Again, reducing protocol header size saves energy to for ward packets, but at the cost of maintaining a state for compression/decompressi on, plus computing these operations. The gain from such protocol optimization fu rther depends on the application and whether it sends packets with large payload s close to the MTU (the header tax and any savings here are very limited), or wh ether it sends packets with very small payload size (making the header tax more pronounced and savings more significant). <t>Another solution to reduce the bandwidth of network protocols is by r educing their header tax, for example, by applying header compression. An exampl e in IETF is RObust Header Compression (ROHC) <xref target="RFC3095"/>. Again, r educing protocol header size saves energy to forward packets, but at the cost of maintaining a state for compression/decompression, plus computing these operati ons. The gain from such protocol optimization further depends on the application and whether it sends packets with (a) large payloads close to the MTU, thus lim iting the header tax and any savings, or (b) very small payload size, thus incre asing the header tax and the savings.
</t> </t>
<t> <t>
An alternative to reducing the amount of protocol data is to design routing prot An alternative to reducing the amount of protocol data is to design routing prot
ocols that are more efficient to process at each node. For instance, path-based ocols that are more efficient to process at each node. For instance, path-based
forwarding/labels such as MPLS <xref target="RFC3031"/> facilitate the next hop forwarding/labels such as MPLS <xref target="RFC3031"/> facilitate the next hop
look-up, thereby reducing the energy consumption. It is unclear if some state at lookup, thereby reducing the energy consumption. It is unclear if some state at
router to speed up look up is more energy efficient that "no state + lookup" th router to speed up lookup is more energy efficient than "no state + lookup", whi
at is more computationally intensive. Other methods to speed up a next-hop looku ch is more computationally intensive. Other methods to speed up a next-hop looku
p include geographic routing (e.g., <xref target="Herzen2011PIE"/>). Some networ p include geographic routing (e.g., <xref target="Herzen2011PIE"/>). Some networ
k protocols could be designed to reduce the next hop look-up computation at a ro k protocols could be designed to reduce the next hop lookup computation at a rou
uter. It is unclear if Longest Prefix Match (LPM) is efficient from an energy p ter. It is unclear whether Longest Prefix Match (LPM) is energy efficient or a
oint of view or if constitutes a significant energy burden for the operation of significant energy burden for router operation.
a router.
</t> </t>
<t>Beyond the volume of data itself, another consideration is the number of mess ages and chattiness of the protocol. Some protocols rely on frequent periodic u pdates or heartbeats, which may prevent equipment to go into sleep mode. In such cases, it makes sense to explore the use of feasible alternatives that rely on different communication patterns and fewer messages. <t>Beyond the volume of data itself, another consideration is the number of messages and chattiness of the protocol. Some protocols rely on frequent pe riodic updates or heartbeats, which may prevent equipment from going into sleep mode. In such cases, it makes sense to explore the use of feasible alternatives that rely on different communication patterns and fewer messages.
</t> </t>
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Assessments of energy-related tradeoffs regarding protocol design space a
nd tradeoffs, such as maintaining state versus more compact encodings or extra c
omputation for transcoding operations versus larger data volume. </t>
<t>Protocol advances for improving the ratio of goodput to throughput and to
reduce waste: reduction in header tax, in protocol verbosity, in need for retra
nsmissions, improvements in coding, etc. </t>
<t>Protocols that allow to manage transmission patterns in ways that facilit
ate periods of link inactivity, such as burstiness and chattiness. </t>
</list>
</t> </t>
</section> <ul spacing="normal">
<li>
<section title="Network Addressing" anchor="Addressing"> <t>Assessments of energy-related trade-offs regarding protocol desig
<t> n space and trade-offs, such as maintaining state versus more compact encodings,
There may be other ways to shave off energy usage from networks. One example co or extra computation for transcoding operations versus larger data volume. </t>
ncerns network addressing. Address tables can get very large, resulting in larg </li>
e forwarding tables that require considerable amount of memory, in addition to l <li>
arge amounts of state needing to be maintained and synchronized. From an energy <t>
footprint perspective, both can be considered wasteful and offer opportunities Protocol advances for improving the ratio of goodput to throughput
for improvement. At the protocol level, rethinking how addresses are structure and to reduce waste; this includes advances such as coding improvements,
d can allow for flexible addressing schemes that can be exploited in network dep reductions in header tax, lower protocol verbosity, and reduced need for
loyments that are less energy-intensive by design. This can be complemented by retransmissions.
supporting clever address allocation schemes that minimize the number of require </t>
d forwarding entries as part of deployments. </li>
<li>
<t>Protocols that allow for managing transmission patterns in ways t
hat facilitate periods of link inactivity, such as burstiness and chattiness. </
t>
</li>
</ul>
</section>
<section anchor="Addressing">
<name>Network Addressing</name>
<t>
Network addressing is another way to shave off energy usage from networks. Addre
ss tables can get very large, resulting in large forwarding tables that require
considerable amount of memory, in addition to large amounts of state needing to
be maintained and synchronized. Memory as well as the processing needed to main
tain and synchronize state both consume energy. Exploring ways to reduce the am
ount of memory and synchronization of state that is required offers opportunitie
s to reduce energy use.
At the protocol level, rethinking how addresses are structured can allow for fle
xible addressing schemes that can be exploited in network deployments that are l
ess energy-intensive by design. This can be complemented by supporting clever a
ddress allocation schemes that minimize the number of required forwarding entrie
s as part of deployments.
</t> </t>
<t>Alternatively, the address could be designed to allow for more efficient proc essing than LPM. For instance, a geographic type of addressing (where the next h op is computed as a simple distance calculation based on the respective position of the current node, of its neighbors and of the destination) <xref target="Her zen2011PIE"/> could be potentially more energy efficient. <t>Alternatively, the addressing could be designed to allow for more eff icient processing than LPM. For instance, a geographic type of addressing (where the next hop is computed as a simple distance calculation based on the respecti ve position of the current node, of its neighbors and of the destination) <xref target="Herzen2011PIE"/> could be potentially more energy efficient.
</t> </t>
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Devise methods to assess the magnitude of the carbon footprint that is as
sociated with addressing schemes.</t>
<t>Devise methods to improve addressing schemes, as well as address assignme
nt schemes, to minimize their footprint.</t>
</list>
</t> </t>
</section> <ul spacing="normal">
</section><!-- End of Protocol Level Section--> <li>
<t>Devise methods to assess the magnitude of the carbon footprint th
at is associated with addressing schemes.</t>
</li>
<li>
<t>Devise methods to improve addressing schemes, as well as address
assignment schemes, to minimize their footprint.</t>
</li>
</ul>
</section>
</section>
<section title="Challenges and Opportunities - Network Level" anchor="sec:networ <section anchor="sec_network">
k"> <name>Challenges and Opportunities - Network Level</name>
<section title="Network Optimization and Energy/Carbon/Pollution-Aware Networkin <section anchor="sec_ean">
g" anchor="sec:ean"> <name>Network Optimization and Energy/Carbon/Pollution-Aware Networking<
<t> /name>
Networks have been optimized for many years under many criteria, for example to <t>
optimize (maximize) network utilization and to optimize (minimize) cost. Hence, Networks have been optimized for many years under many criteria, for example, to
it is straightforward to add optimization for "greenness" (including energy eff optimize (maximize) network utilization and to optimize (minimize) cost. Hence
iciency, power consumption, carbon footprint) as important criteria. , it is straightforward to add optimization for greenness (including energy effi
ciency, power consumption, carbon footprint) as important criteria.
</t> </t>
<t> <t>
This includes assessing the carbon footprints of paths and optimizing those path s so that overall footprint is minimized, then applying techniques such as path- aware networking or segment routing <xref target="RFC8402"/> to steer traffic al ong those paths. (As mentioned earlier, other proxy measures This includes assessing the carbon footprints of paths and optimizing those path s so that overall footprint is minimized, then applying techniques such as path- aware networking or segment routing <xref target="RFC8402"/> to steer traffic al ong those paths. (As mentioned earlier, other proxy measures
could be used for carbon footprint, such as an energy-efficiency ratings of trav could be used for carbon footprint, such as energy-efficiency ratings of travers
ersed equipment.) ed equipment.)
It also includes aspects such as considering the incremental carbon footprint in It also includes aspects such as considering the incremental carbon footprint in
routing decisions. Optimizing cost has a long tradition in networking; many of routing decisions. Optimizing cost has long been an area of focus in networkin
the existing mechanisms can be leveraged for greener networking simply by intro g; many of the existing mechanisms can be leveraged for greener networking simpl
ducing carbon footprint as a cost factor. Low-hanging fruit include the inclusio y by introducing the carbon footprint as a cost factor. Low-hanging fruit includ
n of carbon-related parameters as a cost parameter in control planes, whether di es adding carbon-related parameters as a cost parameter in control planes, wheth
stributed (e.g., IGP) or conceptually centralized via SDN controllers. Likewise, er distributed (e.g., IGP) or conceptually centralized via SDN controllers. Like
there are opportunities in right-placing functionality in the network. An exam wise, there are opportunities to correctly place functionality in the network fo
ple concerns placement of virtualized network functions in carbon-optimized ways r optimal effectiveness.
- for example, cohosted on fewer servers in close proximity to each other in or <!--[rfced] Will the term "right-placing" be clear to the reader?
der to avoid unnecessary overhead in long-distance control traffic. It has not been used before in the RFC series and does not appear
in the dictionary. It appears twice in this document, as follows.
Please consider whether the term should be replaced or explained,
as follows or otherwise.
Section 6.1:
there are opportunities in right-placing functionality in the network.
current: opportunities to correctly place functionality in the network.
Section 7:
..., right-placing of computational intelligence, ...
Perhaps:
S6.1: there are opportunities for "right-placing" functionality, i.e.,
ensuring that it is in the most suitable location for optimal effectiveness.
[or]
there are opportunities for deliberate placement of functionality
within the network.
S7: ensuring the most suitable placement of computational intelligence
within the network design
S7: updated as noted.
Carlos suggested:
... replace "right-placing" (there are, I believe, two instances) with something
using "correct".
6.1: From Alex:
Perhaps:
Likewise, there are opportunities in right-placing functionality in the networ
k, i.e. place functionality in a way that minimizes energy usage without comprom
ising other aspects of that functionality (such as performance or utility). An
example concerns placement of virtualized network functions in carbon-optimized
ways. For example, virtualized network functions can be cohosted on fewer server
s to achieve higher server utilization, which is more effective from an energy a
nd carbon perspective than larger numbers of servers with lower utilization. Li
kewise, they can be placed in close proximity to each other in order to avoid un
necessary overhead in long-distance control traffic.
combination of the above with "correctly place"
-->
An example is placement of virtualized network functions in carbon-optimized wa
ys. for exmaple, virtualized network functions can be cohosted on fewer servers
to achieve higher server utilization, which is more effective from an energy an
d carbon perspective than larger numbers of servers with lower utilization. Lik
ewise, they can be placed in close proximity to each other in order to avoid unn
ecessary overhead in long-distance control traffic.
</t> </t>
<t> <t>
Other opportunities concern adding carbon-awareness to dynamic path selection sc Other opportunities concern adding carbon awareness to dynamic path selection sc
hemes. This is sometimes also referred to as "energy-aware networking" (respect hemes. This is sometimes referred to as "energy-aware networking" (or "pollutio
ively "pollution-aware networking" <xref target="Hossain2019"/> or "carbon-aware n-aware networking" <xref target="Hossain2019"/> or "carbon-aware networking", w
networking", when carbon footprint related parameters beyond pure energy consum hen parameters beyond simply energy consumption are taken into account). Again,
ption are taken into account). Again, considerable energy savings can potential considerable energy savings can potentially be realized by taking resources off
ly be realized by taking resources offline (e.g., putting them into power-saving line (e.g., putting them into power-saving or hibernation mode) when they are no
or hibernation mode) when they are not currently needed under current network d t needed under current network demand and load conditions.
emand and load conditions. Therefore, weaning such resources from traffic become Therefore, weaning resources from traffic is an important consideration
s an important consideration for energy-efficient traffic steering. This contra for energy-efficient traffic steering. This approach contrasts and indeed conf
sts and indeed conflicts with existing schemes that typically aim to create redu licts with
ndancy and load-balance traffic across a network to achieve even resource utiliz existing schemes that typically aim to to create redundancy and load-balance t
ation. This usually occurs for important reasons, such as making networks more r raffic across a network to achieve even resource utilization across larger numbe
esilient, optimizing service levels, and increasing fairness. One of the big ch rs of network resources as a means to increase network resilience,
allenges hence concerns how resource weaning schemes to realize energy savings c optimize service levels, and ensure fairness.
an be accommodated while preventing the cannibalization of other important goals Thus, a big challenge is how resource-weaning schemes to realize energy savings
, counteracting other established mechanisms, and avoiding destabilization of th can be accommodated without cannibalizing other important goals, counteracting o
e network. ther established mechanisms, or destabilizing the network.
</t> </t>
<t>An opportunity may lie in making a distinction between "energy modes" of diff
erent domains. For instance, in a highly trafficked core, the energy challenge i <t>An opportunity may lie in making a distinction between "energy modes"
s to transmit the traffic efficiently. The amount of traffic is relatively fluid of different domains. For instance, in a highly trafficked core, the energy cha
(due to multiplexing of multiple sessions) and the traffic is predictable. In t llenge is to transmit the traffic efficiently. The amount of traffic is relative
his case, there is no need to optimize on a per session basis nor even at a shor ly fluid (due to multiplexing of multiple sessions) and the traffic is predictab
t time scale. In the access networks connecting to that core, though, there are le. In this case, there is no need to optimize on a per-session basis or at a sh
opportunities for this fast convergence: traffic is much more bursty, less predi ort timescale. In the access networks connecting to that core, though, there are
ctable and the network should be able to be more reactive. Other domains such as opportunities for this fast convergence: traffic is much more bursty and less p
DCs may have also more variable workloads and different traffic patterns. redictable, and the network should be able to be more reactive. Other domains su
ch as DCs may have more variable workloads and different traffic patterns.
</t> </t>
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Devise methods for carbon-aware traffic steering and routing; treat carbo
n footprint as a traffic cost metric to optimize.</t>
<t>Apply ML and AI methods to optimize networks for carbon footprint; assess
applicability of game theoretic approaches.</t>
<t>Articulate and, as applicable, moderate tradeoffs between carbon awarenes
s and other operational goals such as robustness and redundancy. </t>
<t>Extend control-plane protocols with carbon-related parameters. </t>
<t>Consider security issues imposed by greater energy awareness, to minimize
the new attack surfaces that would allow an adversary to turn off resources or
to waste energy.</t>
</list>
</t> </t>
</section> <ul spacing="normal">
<section title="Assessing Carbon Footprint and Network-Level Instrumentation"> <li>
<t> <t>Devise methods for carbon-aware traffic steering and routing; tre
As an important prerequisite to capture many of the opportunities outlined in <x at carbon footprint as a traffic cost metric to optimize.</t>
ref target="sec:ean"/>, good abstractions (and corresponding instrumentation) th </li>
at allow to easily assess energy cost and carbon footprint will be required. The <li>
se abstractions need to account for not only for the energy cost associated with <t>Apply Machine Learning (ML) and AI methods to optimize networks f
packet forwarding across a given path, but related cost for processing, for mem or carbon footprint; assess applicability of game theoretic approaches.</t>
ory, for maintaining of state, to result in a holistic picture. </li>
<li>
<t>Articulate and, as applicable, moderate trade-offs between carbon
awareness and other operational goals such as robustness and redundancy. </t>
</li>
<li>
<t>Extend control plane protocols with carbon-related parameters. </
t>
</li>
<li>
<t>Consider security issues imposed by greater energy awareness, to
minimize the new attack surfaces that would allow an adversary to turn off resou
rces or to waste energy.</t>
</li>
</ul>
</section>
<section>
<name>Assessing Carbon Footprint and Network-Level Instrumentation</name
>
<t>
As an important prerequisite to capture many of the opportunities outlined in <x
ref target="sec_ean"/>, good abstractions (and corresponding instrumentation) fo
r easily assessing energy cost and carbon footprint will be required. These abst
ractions need to account for not only the energy cost associated with packet for
warding across a given path, but also the related cost for processing, for memor
y, and for maintaining of state, to result in a holistic picture.
</t> </t>
<t> Optimization of carbon footprint involves in many cases trade-offs that invo
lve not only packet forwarding but also aspects such as keeping state, caching d <t>In many cases, optimization of carbon footprint has trade-offs that i
ata, or running computations at the edge instead of elsewhere. (Note: there may nvolve not only packet forwarding but also aspects such as keeping state, cachin
be a differential in running a computation at an edge server vs. at an hypersca g data, or running computations at the edge instead of elsewhere. (Note: There
le DC. The latter is often better optimized than the latter.) Likewise, other a may be a differential in running a computation at an edge server vs. at a hypers
spects of carbon footprint beyond mere energy-intensity should be considered. Fo cale DC. The latter is often better optimized than the former.) Likewise, other
r instance, some network segments may be powered by more sustainable energy sour aspects of carbon footprint beyond mere energy-intensity should be considered.
ces than others, and some network equipment may be more environmentally friendly For instance, some network segments may be powered by more sustainable energy so
to build, deploy and recycle, all of which can be reflected in abstractions to urces than others, and some network equipment may be more environmentally friend
consider. ly to build, deploy, and recycle, all of which can be reflected in abstractions
to consider.
</t> </t>
<t>Assessing carbon footprint at the network level requires instrumentation that
associates that footprint not just with individual devices (as outline in <xref <t>Assessing carbon footprint at the network level requires instrumentat
target="sec:instrum"/> but relates it also to concepts that are meaningful at t ion that associates that footprint not just with individual devices (as outlined
he network level, i.e., to flows and to paths. For example, it will be useful t in <xref target="sec_instrum"/>) but also with concepts that are meaningful at
o provide visibility into the carbon intensity of a path: Can the carbon cost of the network level, i.e., to flows and to paths. For example, it will be useful
traffic transmitted over the path be aggregated? Does the path include outliers to provide visibility into the carbon intensity of a path: Can the carbon cost o
, i.e., segments with equipment with a particularly poor carbon footprint? f traffic transmitted over the path be aggregated? Does the path include outlier
s, i.e., segments with equipment with a particularly large carbon footprint?
</t> </t>
<t>Similarly, how can the carbon cost of a flow be assessed? That might serve m <t>Similarly, how can the carbon cost of a flow be assessed?
any purposes beyond network optimization, from the option to introduce green bil That might serve many purposes beyond network optimization, from the option to
ling and charging schemes to the ability to raise carbon awareness by end users. introduce green billing and charging schemes that account for the amount of carb
on-equivalent emissions that are attributed to the use of communication services
by particular users to the ability to raise carbon awareness by end users.
</t> </t>
<!-- CW: One key class of network that is potentially more energy intensive, but uses mostly renewable energy is that of satellite networks, and in particular L ow Earth Orbit constellations. Wireless networks are less energy efficient than fiber, as there is more signal loss and the transmitted energy is less focused towards the destination (this has to be qualified for free-space optical transmi ssion of course). Therefore transmitting a signal into space and back comes with an arguably high energy cost. However, satellites cannot be connected to a powe r grid, and therefore have the requirement to be energy independent. Solar array s provide the energy to power up the satellites, with a higher efficienty outsid e of the atmosphere. Any discussion of the "greenness" of such network should ta ke into account the energy cost of building up an infrastructure in space and ho w this energy "capital expense" is amortized over time with the lower "operating expense" of using renewable energy. -->
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Devise methods to assess, to estimate, to predict carbon-intensity of paths
.</t>
<t>Devise methods to account for carbon footprint of flows and networking serv
ices.</t>
</list>
</t> </t>
</section> <ul spacing="normal">
<section title="Dimensioning and Peak Shaving"> <li>
<t> <t>Devise methods to assess, estimate, and predict the carbon intens
As mentioned in <xref target="sec:implications"/>, the overall energy usage of a ity of paths.</t>
network is in large part determined by how the network is dimensioned, specific </li>
ally: which and how many pieces of network equipment are deployed and turned on. <li>
A significant portion of energy is drawn even when simply in idle state. Minim <t>Devise methods to account for the carbon footprint of flows and n
izing the amount of equipment that needs to be turned on in the first place pres etworking services.</t>
ents hence one of the biggest energy saving opportunities. </li>
</ul>
</section>
<section>
<name>Dimensioning and Peak Shaving</name>
<t>
As mentioned in <xref target="sec_implications"/>, the overall energy usage of a
network is in large part determined by how the network is dimensioned, specific
ally: which and how many pieces of network equipment are deployed and turned on.
A significant portion of energy is drawn even when simply in idle state. Hence
, minimizing the amount of equipment that needs to be turned on in the first pla
ce presents one of the biggest energy-saving opportunities.
</t> </t>
<t> <t>
Network deployments are generally dimensioned for periods of peak traffic, resul Network deployments are generally dimensioned for periods of peak traffic, resul
ting in excess capacity during periods of non-peak usage that nonetheless consum ting in excess capacity during periods of non-peak usage that nonetheless consum
es power. Shaving peak usage may thus result in outsized sustainability gains, es power. Shaving peak usage may thus result in outsized sustainability gains,
as it reduces not only energy usage during peak traffic, but more importantly wa as it reduces energy usage during peak traffic but, more importantly, waste duri
ste during non-peak periods. ng non-peak periods.
</t> </t>
<t>While traffic volume is largely a function of demand traffic that network pro viders have little influence over, some peak shaving cand nevertheless be accomp lished by techniques such as spreading spikes out over geographies (e.g. redirec ting some traffic across more costly but less utilized routes, particular in cas es when traffic spikes are of a more local or reginal nature) or over time (e.g. postponing non-urgent traffic, storing or buffering using edge clouds or extra storage where feasible). <t>While traffic volume is largely a function of demand traffic that net work providers have little influence over, some peak shaving can nevertheless be accomplished by techniques such as spreading spikes out over geographies (e.g., redirecting some traffic across more costly but less utilized routes, particula rly in cases when traffic spikes are of a more local or regional nature) or over time (e.g., postponing non-urgent traffic, storing or buffering using edge clou ds or extra storage where feasible).
</t> </t>
<t>To make techniques effective, accurate learning and prediction of traffic pat terns is required. This includes the ability to perform forecasting to ensure th at additional resources can be spun up in time should it be needed. Clearly, th is presents interesting challenges, yet also opportunities for technical advance s to make a difference. <t>To make techniques effective, accurate learning and prediction of tra ffic patterns are required. This includes the ability to perform forecasting to ensure that additional resources can be spun up in time should they be needed. Clearly, this presents interesting challenges, yet also opportunities for techni cal advances to make a difference.
</t> </t>
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Support for methods that allow to monitor and forecast traffic demand, invo
lving new mechanisms and/or performance improvements of existing mechanisms to s
upport the collection of telemetry and generation of traffic matrices at very hi
gh velocity and scale </t>
<t>Additional methods that allow for even traffic load distribution across the
network, i.e. load balancing on a network scale, and enablement of those method
s through control protocol extensions as needed.</t>
</list>
</t> </t>
<ul spacing="normal">
</section> <li>
<section title="Convergence Schemes"> <t>Support methods for monitoring and forecasting traffic demand, in
<t> volving new mechanisms and/or performance improvements of existing mechanisms to
One set of challenges of carbon-aware networking concerns the fact that many sch support the collection of telemetry and generation of traffic matrices at very
emes result in much greater dynamicity and continuous change in the network as r high velocity and scale.</t>
esources may be getting steered away from (when possible) and then leveraged aga </li>
in (when necessary) in rapid succession. This imposes significant stress on con <li>
vergence schemes that results in challenges to the scalability of solutions and <t>Additional methods for distributing traffic load evenly across th
their ability to perform in a fast-enough manner. Network-wide convergence impos e network, i.e., load balancing on a network scale, and enablement of those meth
es high cost and incurs significant delay and is hence not susceptible to such s ods through control protocol extensions as needed.</t>
chemes. In order to mitigate this problem, mechanisms should be investigated tha </li>
t do not require convergence beyond the vicinity of the affected network device. </ul>
Especially in cases where central network controllers are involved that are re </section>
sponsible for aspects such as configuration of paths and the positioning of netw <section>
ork functions and that aim for global optimization, the impact of churn needs to <name>Convergence Schemes</name>
be minimized. This means that, for example, (re-) discovery and update schemes <t>
need to be simplified and extensive recalculation e.g., of routes and paths bas One set of challenges of carbon-aware networking concerns the fact that many sch
ed on the current energy state of the network needs to be avoided. emes result in much greater dynamicity and continuous change in the network, as
resources may be steered away from (when possible) and then leveraged again (whe
n necessary) in rapid succession. This imposes significant stress on convergenc
e schemes that results in challenges to the scalability of solutions and their a
bility to perform in a fast-enough manner. Network-wide convergence imposes high
cost and incurs significant delay and thus is not susceptible to such schemes.
In order to mitigate this problem, mechanisms should be investigated that do not
require convergence beyond the vicinity of the affected network device. The im
pact of churn needs to be minimized, especially in cases where central network c
ontrollers (responsible for the configuration of paths and the positioning of ne
twork functions and that aim for global optimization) are involved. This means
that, for example, discovery, rediscovery, and update schemes need to be simplif
ied, and extensive recalculation (e.g., of routes and paths based on the current
energy state of the network) needs to be avoided.
</t> </t>
<t>The following summarizes some challenges and opportunities in this space that <t>The following summarizes some challenges and opportunities in this sp
can provide the basis for advances in greener networking: ace that can provide the basis for advances in greener networking:
<list style="symbols">
<t>Protocols that facilitate rapid convergence (per <xref target="NetworkEne
rgySaving"/>).</t>
<t>Investigate methods that mitigate effects of churn, including methods tha
t maintain memory or state as well as methods relying on prediction, inference,
and interpolation.</t>
</list>
</t> </t>
</section> <ul spacing="normal">
<li>
<t>Protocols that facilitate rapid convergence (per <xref target="Ne
tworkEnergySaving"/>).</t>
</li>
<li>
<t>Investigate methods that mitigate effects of churn, including met
hods that maintain memory or state as well as methods relying on prediction, inf
erence, and interpolation.</t>
</li>
</ul>
</section>
<!-- Alex, feel free to comment out this following section, we can talk more abo <section>
ut it. --> <name>The Role of Topology</name>
<t>
One of the most important network management constructs is that of the n
etwork topology. A network topology can usually be represented as a database or
as a mathematical graph, with vertices or nodes, edges or links, representing ne
tworking nodes, links connecting their interfaces, and all their characteristics
. Examples of these network topology representations include routing protocols'
link-state databases (LSDBs) and service function chaining graphs.
</t>
<section title="The Role of Topology"> <t>
<t> As part of adding carbon and energy awareness into networks, it is useful also f
One of the most important network management constructs is that of the n or topology information to provide visibility into sustainability data. Such ca
etwork topology. A network topology can usually be represented as a database or pabilities can help to assess sustainability of the network overall and can enab
as a mathematical graph, with vertices or nodes, edges or links, representing ne le automated applications to improve it.
tworking nodes, links connecting their interfaces, and all their characteristics </t>
. Examples of these network topology representations include routing protocols l <t>
ink-state databases, and service function chaining graphs.
</t>
<t>
As we desire to add carbon and energy awareness into networks, the energ
y proportionality of topologies directly supports sustainability visibility and
improvements via automation.
</t>
<t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Embedding carbon and energy awareness into the representation of topo
logies, whether considering IGP LSDBs (link-state databases) and their advertise
ments, BGP-LS (BGP Link-State), or metadata for the rendering of service functio
n paths in a service chain.</t>
<t>Use of those carbon-aware attributes to optimize topology as a whole
under end-to-end energy and carbon considerations.
</t> </t>
</list> <ul spacing="normal">
</t> <li>
</section> <t>Embedding carbon and energy awareness into the representation of
topologies, whether considering IGP LSDBs and their advertisements, BGP-LS (BGP
</section> Link-State), or metadata for the rendering of service function paths in a servic
e chain.</t>
<section title="Challenges and Opportunities - Architecture Level" anchor="sec:a </li>
rchi"> <li>
<t> <t>Use of those carbon-aware attributes to optimize topology as a wh
Another possibility to improve network energy efficiency is to organize networks ole under end-to-end energy and carbon considerations.
in a way that they allow important applications to reduce energy consumption. E </t>
xamples include facilitating retrieval of content or performing computation in w </li>
ays that minimize the amount of communication that needs to take place in the fi </ul>
rst place, even if energy savings within the network may at least in part be off </section>
set by additional energy consumption elsewhere. The following are some examples </section>
that suggest that it may be worthwhile reconsidering the ways in which networks <section anchor="sec_archi">
are architected to minimize their carbon footprint. <name>Challenges and Opportunities - Architecture Level</name>
<t>
Another possibility to improve network energy efficiency is to organize networks
in a way that they allow important applications to reduce energy consumption. E
xamples include facilitating retrieval of content or performing computation in w
ays that minimize the amount of communication needed in the first place, even if
energy savings within the network may be offset (at least in part) by additiona
l energy consumption elsewhere. The following examples suggest that it may be wo
rthwhile to reconsider the ways in which networks are architected to minimize th
eir carbon footprint.
</t> </t>
<t> <t>
For example, Content Delivery Networks (CDNs) have reduced the energy expenditur For example, Content Delivery Networks (CDNs) have reduced the energy expenditur
e of the Internet by downloading content near the users. The content is sent onl e of the Internet by downloading content near the users. The content is sent onl
y a few times over the WAN, and then is served locally. This shifts the energy c y a few times over the WAN and then is served locally. This shifts the energy co
onsumption from networking to storage. Further methods can reduce the energy usa nsumption from networking to storage. Further methods can reduce the energy usag
ge even more <xref target="Bianco2016energy"/> <xref target="Mathew2011energy"/> e even more <xref target="Bianco2016energy"/> <xref target="Mathew2011energy"/>
<xref target="Islam2012evaluating"/>. Whether overall energy savings are net po <xref target="Islam2012evaluating"/>. Whether overall energy savings are net pos
sitive depends on the actual deployment, but from the network operator's perspec itive depends on the actual deployment, but from the network operator's perspect
tive, at least it shifts the energy bill away from the network to the CDN operat ive, at least it shifts the energy bill away from the network to the CDN operato
or. r.
</t> </t>
<t> <t>
While CDNs operate as an overlay, another architecture has been proposed to prov While CDNs operate as an overlay, another architecture has been proposed to prov
ide the CDN features directly in the network, namely Information Centric Network ide the CDN features directly in the network -- namely, Information-Centric Netw
s <xref target="Ahlgren2012survey"/>, studied as well in the IRTF ICNRG. This ho orks <xref target="Ahlgren2012survey"/>, also studied in the ICNRG of the IRTF.
wever shifts the energy consumption back to the network operator and requires so However, this shifts the energy consumption back to the network operator and req
me power-hungry hardware, such as chips for larger name look-ups and memory for uires some power-hungry hardware, such as chips for larger name lookups and memo
the in-network cache. As a result, it is unclear if there is an actual energy g ry for the in-network cache. As a result, it is unclear if there is an actual e
ain from the dissemination and retrieval of content within in-network caches. nergy gain from the dissemination and retrieval of content within in-network cac
hes.
</t> </t>
<t> <t>
Fog computing and placing intelligence at the edge are other architectural direc Fog computing and placing intelligence at the edge are other architectural direc
tions for reducing the amount of energy that is spent on packet forwarding and i tions for reducing the amount of energy that is spent on packet forwarding and i
n the network. There again, the trade-off is between performing computation in a n the network. There again, the trade-off is between performing computational ta
n energy-optimized data center at very large scale (but requiring transmission o sks (a) in an energy-optimized data center at very large scale (but requiring tr
f significant volumes of data across many nodes and long distances) versus perfo ansmission of significant volumes of data across many nodes and long distances)
rming computational tasks at the edge where the energy may not be used as effici versus (b) at the edge where the energy may not be used as efficiently (less mul
ently (less multiplexing of resources and inherently lower efficiency of smaller tiplexing of resources and inherently lower efficiency of smaller sites due to t
sites due to their smaller scale) but the amount of long-distance network traff heir smaller scale) but the amount of long-distance network traffic and energy r
ic and energy required for the network is significantly reduced. equired for the network is significantly reduced.
Softwarization, containers, microservices are direct enablers for such architect Softwarization, containers, and
ures, and the deployment of programmable network infrastructure (as for instance microservices are direct enablers of such architectures. Their
Infrastructure Processing Units - IPUs or SmartNICs that offload some computati realization will be further aided by the deployment of programmable
ons from the CPU onto the NIC) will help its realization. network infrastructure, such as Infrastructure Processing Units
However, the power consumption characteristics of CPUs are different from those (IPUs) or Smart NICs that offload some computations from the CPU onto
of NPUs, another aspect to be considered in conjunction with virtualization. the NIC.
However, the power consumption characteristics of CPUs are different from those
of NPUs; this is another aspect to be considered in conjunction with virtualizat
ion.
</t> </t>
<t> <t>
Other possibilities concern taking economic aspects into consideration impact, s Other possibilities are taking economic aspects into consideration, such as prov
uch as providing incentives to users of networking services in order to minimize iding incentives to users of networking services in order to minimize energy con
energy consumption and emission impact. An example for this is given in <xref sumption and emission impact. In <xref target="Wolf2014choicenet"/>, an example
target="Wolf2014choicenet"/>, which could be expanded to include energy incentiv is provided that could be expanded to include energy incentives.
es.
</t> </t>
<t> <t>
Other approaches consider performing a late binding of data and functions to be Other approaches consider performing a late binding of the data and the function
performed on the data <xref target="Krol2017NFaaS"/>. The COIN Research Group in s to be performed on it <xref target="Krol2017NFaaS"/>. The COINRG of the IRTF f
IRTF focuses on similar issues. Jointly optimizing for the total energy cost th ocuses on similar issues. Jointly optimizing for the total energy cost that take
at takes into account networking as well as computing (along with the different s into account networking as well as computing (along with the different energy
energy cost of computing in an hyperscale DC vs at an edge node) is still an are cost of computing in a hyperscale DC vs. at an edge node) is still an area of op
a of open research. en research.
</t> </t>
<t> <t>
In summary, rethinking of the overall network (and networked application) archit In summary, rethinking the overall network (and networked application) architect
ecture can be an opportunity to significantly reduce the energy cost at the netw ure can be an opportunity to significantly reduce the energy cost at the network
ork layer, for example by performing tasks that involve massive communications c layer, for example, by performing tasks that involve massive communications clo
loser to the user. To what extend these shifts result in a net reduction of car ser to the user. To what extent these shifts result in a net reduction of carbo
bon footprint is an important question that requires further analysis on a case- n footprint is an important question that requires further analysis on a case-by
by-case basis. -case basis.
</t> </t>
<t> <t>
The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking: The following summarizes some challenges and opportunities in this space that ca n provide the basis for advances in greener networking:
<list style="symbols">
<t>Investigate organization of networking architecture for important classes
of applications (examples: content delivery, right-placing of computational int
elligence, industrial operations and control, massively distributed machine lear
ning and AI) to optimize green foot print and holistic approaches to trade off c
arbon footprint between forwarding, storage, and computation.</t>
<t>Models to assess and compare alternatives in providing networked services
, e.g., evaluate carbon impact relative to alternatives where to perform compute
, what information to cache, and what communication exchanges to conduct.</t>
</list>
</t> </t>
</section> <ul spacing="normal">
<li>
<section title="Conclusions"> <t>Investigate organization of networking architecture for important c
<t> How to make networks "greener" and reduce their carbon footprint is an impor lasses of applications (e.g., content delivery, most suitable placement of compu
tant problem for the networking industry to address, both for societal and for e tational intelligence and network functionality within the network design, indus
conomic reasons. This document has highlighted a number of the technical challen trial operations and control, massively distributed ML and AI) to optimize overa
ges and opportunities in that regard. ll sustainability and holistic approaches to trade off carbon footprint between
</t> forwarding, storage, and computation.</t>
<t> </li>
Of those, perhaps the key challenge to address right away concerns the ability t <li>
o expose at a fine granularity the energy impact of any networking actions. Prov <t>Models to assess and compare alternatives in providing networked se
iding visibility into this will enable many approaches to come towards a solutio rvices, e.g., evaluate carbon impact relative to where to perform computation, w
n. It will be key to implementing optimization via control loops that allow to a hat information to cache, and what communication exchanges to conduct.</t>
ssess the energy impact of decision taken. It will also help to answer question </li>
s such as: is caching - with the associated storage energy - better than retrans </ul>
mitting from a different server - with the associated networking cost? Is compre </section>
ssion more energy-efficient once factoring the computation cost of compression v <section>
s transmitting uncompressed data? Which compression scheme is more energy effici <name>Conclusions</name>
ent? Is energy saving of computing at an efficient hyperscale DC compensated by <t> How to make networks greener and reduce their carbon footprint is an i
the networking cost to reach that DC? Is the overhead of gathering and transmitt mportant problem for the networking industry to address, both for societal and f
ing fine-grained energy telemetry data offset by the total energy gain by ways o or economic reasons. This document has highlighted a number of the technical cha
f better decisions that this data enables? Is transmitting data to a Low Earth O llenges and opportunities in that regard.
rbit (LEO) satellite constellation compensated by the fact that once in the cons
tellation, the networking is fueled by solar energy? Is the energy cost of sendi
ng rockets to place routers in Low Earth Orbit amortized over time?
</t>
<t>
Determining where the sweet spots are and optimizing networks along those lines
will be a key towards making networks "greener". We expect to see significant ad
vances across these areas and believe that researchers, developers, and operator
s of networking technology have an important role to play in this.
</t> </t>
</section>
<section title="IANA Considerations"> <t>
<t>This document does not have any IANA requests. </t> Of those, perhaps the key challenge to address right away is the ability to expo
</section> se at a fine granularity the energy impact of any networking actions. Providing
visibility into this will enable many approaches to come towards a solution. It
will be key to implementing optimization via control loops that can assess the e
nergy impact of a decision taken. It will also help to answer questions such as
(but not limited to) the following:</t>
<ul spacing="compact">
<li>Is caching (with the associated storage) better than retransmitting from a
different server (with the associated networking cost)?</li>
<li>Is compression more energy efficient once factoring in the computation cos
t of compression vs. transmitting uncompressed data? Which compression scheme is
more energy efficient?</li>
<li>Is energy saving of computing at an efficient hyperscale DC compensated by
the networking cost to reach that DC?</li>
<li>Is the overhead of gathering and transmitting fine-grained energy telemetr
y data offset by the total energy gain resulting from the better decisions that
this data enables?</li>
<section title="Security Considerations"> <li>Is the energy cost needed to transmit data to a Low Earth Orbit
(LEO) satellite constellation offset by the fact that the constallation and
any networking within it are powered by solar energy?</li>
<li>Is the energy cost of sending rockets to place routers in LEO amortized ov
er time?</li>
</ul>
<t>Security considerations may appear to be orthogonal to green networking consi <t>
derations. However, there are a number of important caveats. Determining where the sweet spots are and optimizing networks along those lines
will be a key towards making networks greener. We expect to see significant adva
nces across these areas and believe that researchers, developers, and operators
of networking technology have an important role to play in this.
</t> </t>
<t>Security vulnerabilities of networks may manifest themselves in compromised e </section>
nergy efficiency. For example, attackers could aim at increasing energy consump <section>
tion to drive up attack victims' energy bill. Specific vulnerabilities will dep <name>IANA Considerations</name>
end on the particular mechanisms. For example, in the case of monitoring energy <t>This document has no IANA actions.</t>
consumption data, tampering with such data might result in compromised energy o </section>
ptimization control loops. Hence any mechanisms to instrument and monitor the ne <section>
twork for such data need to be properly secured to ensure authenticity. <name>Security Considerations</name>
<t>Security considerations may appear to be orthogonal to green networking
considerations. However, there are a number of important caveats.
</t> </t>
<t>In some cases, there are inherent tradeoffs between security and maximal ener gy efficiency that might otherwise be achieved. An example is encryption, which requires additional computation for encryption and decryption activities and se curity handshakes, in addition to the need to send more traffic than necessitate d by the entropy of the actual data stream. Likewise, mechanisms that allow to turn resources on or off could become a target for attackers. <t>Security vulnerabilities of networks may manifest themselves in comprom ised energy efficiency. For example, attackers could aim at increasing energy c onsumption to drive up attack victims' energy bills. Specific vulnerabilities w ill depend on the particular mechanisms. For example, in the case of monitoring energy consumption data, tampering with such data might result in compromised e nergy optimization control loops. Hence, any mechanisms to instrument and monito r the network for such data need to be properly secured to ensure authenticity.
</t> </t>
<t>Energy consumption can be used to create covert channels, which is a security risk for information leakage. For instance, the temperature of an element can b e used to create a Thermal Covert Channel <xref target="TCC"/>, or the reading/s haring of the measured energy consumption can be abused to create a covert chann el (see for instance <xref target="DRAM"/> or <xref target="NewClass"/>). Power information may be used to create side-channel attacks. For instance, <xref targ et="SideChannel"/> provides a review of 20 years of study on this topic. Any new parameters considered in protocol designs or in measurements are susceptible to create such covert or side channel and this should be taken into account while designing energy efficient protocols. <t>In some cases, there are inherent trade-offs between security and maxim al energy efficiency that might otherwise be achieved. An example is encryption , which requires additional computation for encryption and decryption activities and security handshakes, in addition to the need to send more traffic than nece ssitated by the entropy of the actual data stream. Likewise, mechanisms that al low to turn resources on or off could become a target for attackers.
</t> </t>
</section> <t>Energy consumption can be used to create covert channels, which is a se
curity risk for information leakage. For instance, the temperature of an element
<section title="Contributors"> can be used to create a Thermal Covert Channel <xref target="TCC"/>, or the rea
<t> ding/sharing of the measured energy consumption can be abused to create a covert
<list> channel (see for instance <xref target="DRAM"/> or <xref target="NewClass"/>).
<t>Michael Welzl, University of Oslo, michawe@ifi.uio.no</t> Power information may be used to create side-channel attacks. For instance, <xre
</list> f target="SideChannel"/> provides a review of 20 years of study on this topic. A
</t> ny new parameters considered in protocol designs or in measurements are suscepti
</section> ble to create such covert or side channels, and this should be taken into accoun
t while designing energy-efficient protocols.
<section title="Acknowledgments">
<t>
The authors thank Dave Oran for providing the information regarding covert chann
els using energy measurements, and Mike King for an exceptionally thorough revie
w and useful comments.
</t> </t>
</section> </section>
</middle>
<back> </middle>
<back>
<displayreference target="I-D.ietf-tvr-requirements" to="TVR_REQS" />
<displayreference target="I-D.cx-green-green-metrics" to="GREEN_METRICS"/>
<displayreference target="I-D.li-green-power" to="POWER_YANG"/>
<references title="Informative References"> <references>
&RFC.2481;
&RFC.3031;
&RFC.3095;
&RFC.7950;
&RFC.8402;
&RFC.9330;
<reference anchor="I.D.draft-cx-green-green-metrics"> <name>Informative References</name>
<front> <xi:include href="https://bib.ietf.org/public/rfc/bibxml/reference.RFC.316
<title>Green Networking Metrics</title> 8.xml"/>
<author fullname="Alexander Clemm"></author> <xi:include href="https://bib.ietf.org/public/rfc/bibxml/reference.RFC.303
<author fullname="Carlos Pignataro"></author> 1.xml"/>
<author fullname="Eve Schooler"></author> <xi:include href="https://bib.ietf.org/public/rfc/bibxml/reference.RFC.309
<author fullname="Laurent Ciavaglia"></author> 5.xml"/>
<author fullname="Ali Rezaki"></author> <xi:include href="https://bib.ietf.org/public/rfc/bibxml/reference.RFC.795
<author fullname="Greg Mirsky"></author> 0.xml"/>
<author fullname="Jeff Tantsura"></author> <xi:include href="https://bib.ietf.org/public/rfc/bibxml/reference.RFC.840
<date month="October" year="2024"/> 2.xml"/>
</front> <xi:include
</reference> href="https://bib.ietf.org/public/rfc/bibxml/reference.RFC.9330.xml"/>
<xi:include href="https://bib.ietf.org/public/rfc/bibxml/reference.RFC.900
0.xml"/>
<reference anchor="I.D.draft-li-green-power"> <!-- [I-D.cx-green-green-metrics] IESG State: Expired as of 06/27/25-->
<front>
<title>A YANG model for Power Management</title>
<author fullname="Tony Li"></author>
<author fullname="Ron Bonica"></author>
<date month="October" year="2024"/>
</front>
</reference>
<reference anchor="I.D.draft-ietf-tvr-requirements"> <reference anchor="I-D.cx-green-green-metrics" target="https://datatracker.ietf.
<front> org/doc/html/draft-cx-green-green-metrics-00">
<title>TVR (Time-Variant Routing) Requirements</title> <front>
<author fullname="Daniel King"></author> <title>Green Networking Metrics for Environmentally Sustainable Networking
<author fullname="Luis Contreras"></author> </title>
<author fullname="Brian Sipos"></author> <author initials="A." surname="Clemm" fullname="Alexander Clemm" role="edi
<author fullname="L. Zhang"></author> tor">
<date month="September" year="2024"/> <organization>Santa Clara University</organization>
</front> </author>
<author initials="C." surname="Pignataro" fullname="Carlos Pignataro" role
="editor">
<organization>North Carolina State University</organization>
</author>
<author initials="E." surname="Schooler" fullname="Eve Schooler">
<organization>University of Oxford</organization>
</author>
<author initials="L." surname="Ciavaglia" fullname="Laurent Ciavaglia">
<organization>Nokia</organization>
</author>
<author initials="A." surname="Rezaki" fullname="Ali Rezaki">
<organization>Nokia</organization>
</author>
<author initials="G." surname="Mirsky" fullname="Greg Mirsky">
<organization>Ericsson</organization>
</author>
<author initials="J." surname="Tantsura" fullname="Jeff Tantsura">
<organization>Nvidia</organization>
</author>
<date month="October" day="21" year="2024" />
</front>
<seriesInfo name="Internet-Draft" value="draft-cx-green-green-metrics-00" />
</reference> </reference>
<reference anchor="Telefonica2021">
<front>
<title>Telefonica Consolidated Annual Report 2021.</title>
<author fullname="Telefonica"> </author>
<date year="2021"/>
</front>
</reference>
<reference anchor="GreenNet22"> <!-- [I-D.li-green-power] IESG State: Expired as of 06/27/25-->
<front> <xi:include href="https://bib.ietf.org/public/rfc/bibxml3/reference.I-D.li
<title>Challenges and Opportunities in Green Networking</title> -green-power.xml"/>
<author fullname="Alexander Clemm"> </author>
<author fullname="Cedric Westphal"></author>
<date month="June" year="2022" />
</front>
<seriesInfo
name="1st International Workshop on Network Energy Efficiency in
the Softwarization Era" value="IEEE NetSoft 2022"/>
</reference> <!-- [I-D.ietf-tvr-requirements] IESG State: I-D Exists as of 06/27/25-->
<xi:include href="https://bib.ietf.org/public/rfc/bibxml3/reference.I-D.ie
tf-tvr-requirements.xml"/>
<reference anchor="Bolla2011energy"> <!-- [rfced] still being discussed
<front>
<title>Energy Efficiency in the Future Internet: A Survey of Existing Approa
ches and Trends in Energy-Aware Fixed Network Infrastructures</title>
<author fullname="Raffaele Bolla"> </author>
<author fullname="Roberto Bruschi"> </author>
<author fullname="Franco Davoli"> </author>
<author fullname="Flavio Cucchietti"> </author>
<date year="2011" />
</front>
<seriesInfo name="IEEE Communications Surveys and Tutorials" value="Vol.13 N
o.2, pp.223-244"/>
</reference> We found the following URL for [Telefonica2021];
would you like to add it or a different URL to this reference?
<reference anchor="Chabarek08"> https://www.telefonica.com/en/shareholders-investors/financial-reports/historica
<front> l-archive-annual-reports/2021/
<title>Power awareness in network design and routing</title>
<author fullname="J. Chabarek"></author>
<author fullname="J. Sommers"></author>
<author fullname="P. Barford"></author>
<author fullname="D. Tsiang"></author>
<author fullname="S. Wright"></author>
<date year="2008"/>
</front>
<seriesInfo name="IEEE Infocom" value="pp.457-465"/>
</reference> Current:
[Telefonica2021]
Telefonica, "Telefonica Consolidated Annual Report 2021",
2021.
-->
<reference anchor="Cervero19"> <reference anchor="Telefonica2024" target="https://www.telefonica.com/en/w
<front> p-content/uploads/sites/5/2025/02/Consolidated-Annual-Accounts-2024.pdf">
<title>Green Wired Networks</title> <front>
<author fullname="Alfonso Gazo Cervero"></author> <title>Telefonica Consolidated Annual Report 2024</title>
<author fullname="Michele Chincoli"></author> <author>
<author fullname="Lars Dittmann"></author> <organization>Telefonica</organization>
<author fullname="Andreas Fischer"></author> </author>
<author fullname="Alberto Garcia"></author> <date year="2025"/>
<date year="2019"/> </front>
</front> </reference>
<seriesInfo name="Wiley Journal on Large-Scale Distributed Systems and En
ergy Efficiency" value="pp.41-80"/>
</reference> <reference anchor="Telefonica2020" target="https://www.telefonica.com/en/s
hareholders-investors/financial-reports/historical-archive-annual-reports/2020/"
>
<front>
<title>Telefonica Consolidated Management Report 2020</title>
<author>
<organization>Telefonica</organization>
</author>
<date year="2021"/>
</front>
</reference>
<reference anchor="Alizadeh2010DCTCP"> <reference anchor="GreenNet22">
<front> <front>
<title>Data Center TCP (DCTCP) </title> <title>Challenges and Opportunities in Green Networking</title>
<author fullname="Mohammad Alizadeh"></author> <author fullname="Alexander Clemm"> </author>
<author fullname="Albert Greenberg"></author> <author fullname="Cedric Westphal"/>
<author fullname="David Maltz"></author> <date month="June" year="2022"/>
<author fullname="Jitendra Padhye"></author> </front>
<author fullname="Parveen Patel"></author> <seriesInfo name="DOI" value="10.1109/NetSoft54395.2022.9844020"/>
<author fullname="Balaji Prabhakar"></author> <refcontent>1st International Workshop on Network Energy Efficiency in t
<author fullname="Sudipta Sengupta"></author> he Softwarization Era, IEEE NetSoft 2022</refcontent>
<author fullname="Murari Sridharan"></author> </reference>
<date year="2010"/>
</front>
<seriesInfo name="ACM SIGCOMM" value="pp.63-74"/>
</reference> <reference anchor="Bolla2011energy">
<front>
<title>Energy Efficiency in the Future Internet: A Survey of Existing
Approaches and Trends in Energy-Aware Fixed Network Infrastructures</title>
<author fullname="Raffaele Bolla"> </author>
<author fullname="Roberto Bruschi"> </author>
<author fullname="Franco Davoli"> </author>
<author fullname="Flavio Cucchietti"> </author>
<date year="2011"/>
</front>
<refcontent>IEEE Communications Surveys and Tutorials, vol. 13, no. 2, p
p. 223-244</refcontent>
<seriesInfo name="DOI" value="10.1109/SURV.2011.071410.00073"/>
</reference>
<reference anchor="QUAL"> <reference anchor="Chabarek08">
<front> <front>
<title> A Framework for Qualitative Communications using Big Packet Protocol <title>Power Awareness in Network Design and Routing</title>
</title> <author fullname="J. Chabarek"/>
<author fullname="Richard Li"></author> <author fullname="J. Sommers"/>
<author fullname="Kiran Makhijani"></author> <author fullname="P. Barford"/>
<author fullname="Hamed Yousefi"></author> <author fullname="C. Estan"/>
<author fullname="Cedric Westphal"></author> <author fullname="D. Tsiang"/>
<author fullname="Lijun Xong"></author> <author fullname="S. Wright"/>
<author fullname="Tim Wauters"></author> <date year="2008"/>
<author fullname="Filip De Turck"></author> </front>
<date year="2019"/> <refcontent>IEEE INFOCOM 2008 - The 27th Conference on Computer Communic
</front> ations, pp. 457-465</refcontent>
<seriesInfo name="Proceedings ACM Sigcomm Workshop On Networking For Emer <seriesInfo name="DOI" value="10.1109/INFOCOM.2008.93"/>
ging Applications And Technologies" value="pp.22-28"/> </reference>
</reference> <reference anchor="Cervero15">
<front>
<title>Green Wired Networks</title>
<author fullname="Alfonso Gazo Cervero"/>
<author fullname="Michele Chincoli"/>
<author fullname="Lars Dittmann"/>
<author fullname="Andreas Fischer"/>
<author fullname="Alberto Garcia"/>
<author fullname="Jaime Galán-Jiménez"/>
<author fullname="Laurent Lefevre"/>
<author fullname="Hermann de Meer"/>
<author fullname="Thierry Monteil"/>
<author fullname="Paolo Monti"/>
<author fullname="Anne-Cecile Orgerie"/>
<author fullname="Louis-Francois Pau"/>
<author fullname="Chris Phillips"/>
<author fullname="Sergio Ricciardi"/>
<author fullname="Remi Sharrock"/>
<author fullname="Patricia Stolf"/>
<author fullname="Tuan Trinh"/>
<author fullname="Luca Valcarenghi"/>
<date year="2015"/>
</front>
<refcontent>Large-Scale Distributed Systems and Energy Efficiency, pp. 4
1-80</refcontent>
<seriesInfo name="DOI" value="10.1002/9781118981122.ch3"/>
</reference>
<reference anchor="Westphal2021qualitative"> <reference anchor="Alizadeh2010DCTCP">
<front> <front>
<title>Qualitative Communications for Augmented Reality and Virtual Reality <title>Data Center TCP (DCTCP) </title>
</title> <author fullname="Mohammad Alizadeh"/>
<author fullname="Cedric Westphal"></author> <author fullname="Albert Greenberg"/>
<author fullname="Dongbiao He"></author> <author fullname="David Maltz"/>
<author fullname="Kiran Makhijani"></author> <author fullname="Jitendra Padhye"/>
<author fullname="Richard Li"></author> <author fullname="Parveen Patel"/>
<date year="2021"/> <author fullname="Balaji Prabhakar"/>
</front> <author fullname="Sudipta Sengupta"/>
<seriesInfo name="22nd IEEE International Conference on High Performance <author fullname="Murari Sridharan"/>
Switching and Routing (HPSR)" value="pp.1-6"/> <date year="2010"/>
</reference> </front>
<refcontent>ACM SIGCOMM Computer Communication Review, vol. 40, no. 4, p
p. 63-74</refcontent>
<seriesInfo name="DOI" value="10.1145/1851275.1851192"/>
</reference>
<reference anchor="Herzen2011PIE"> <reference anchor="QUAL">
<front> <front>
<title>Scalable routing easy as PIE: A practical isometric embedding protoco <title>A Framework for Qualitative Communications using Big Packet Pro
l</title> tocol</title>
<author fullname="Julien Herzen"></author> <author fullname="Richard Li"/>
<author fullname="Cedric Westphal"></author> <author fullname="Kiran Makhijani"/>
<author fullname="Patrick Thiran"></author> <author fullname="Hamed Yousefi"/>
<date year="2011"/> <author fullname="Cedric Westphal"/>
</front> <author fullname="Lijun Xong"/>
<seriesInfo name="19th IEEE International Conference on Network Protocols <author fullname="Tim Wauters"/>
(ICNP)" value="pp.49-58"/> <author fullname="Filip De Turck"/>
</reference> <date year="2019"/>
</front>
<seriesInfo name="DOI" value="10.1145/3341558.3342201"/>
<refcontent>NEAT'19: Proceedings of the ACM Sigcomm Workshop On Networki
ng For Emerging Applications And Technologies, pp. 22-28</refcontent>
</reference>
<reference anchor="Ren2018jordan"> <reference anchor="Westphal2021qualitative">
<front> <front>
<title>JORDAN: A Novel Traffic Engineering Algorithm for Dynamic Adaptive St <title>Qualitative Communications for Augmented Reality and Virtual Re
reaming over HTTP</title> ality </title>
<author fullname="Jing Ren"></author> <author fullname="Cedric Westphal"/>
<author fullname="Kejie Ren"></author> <author fullname="Dongbiao He"/>
<author fullname="Cedric Westphal"></author> <author fullname="Kiran Makhijani"/>
<author fullname="Jin Wang"></author> <author fullname="Richard Li"/>
<author fullname="Jinfan Wang"></author> <date year="2021"/>
<author fullname="Tongyu Song"></author> </front>
<author fullname="Sucheng Liu"></author> <seriesInfo name="DOI" value="10.1109/HPSR52026.2021.9481793"/>
<author fullname="Jianping Wang"></author> <refcontent>22nd IEEE International Conference on High Performance Switc
<date year="2018"/> hing and Routing (HPSR), pp. 1-6</refcontent>
</front> </reference>
<seriesInfo name="IEEE International Conference on Computing, Networking
and Communications (ICNC)" value="pp.581-587"/>
</reference>
<reference anchor="Bianco2016energy"> <reference anchor="Herzen2011PIE">
<front> <front>
<title>Energy consumption for data distribution in content delivery networks <title>Scalable routing easy as PIE: A practical isometric embedding p
</title> rotocol</title>
<author fullname="Andreas Bianco"></author> <author fullname="Julien Herzen"/>
<author fullname="Reza Mashayekhi"></author> <author fullname="Cedric Westphal"/>
<author fullname="Michele Meo"></author> <author fullname="Patrick Thiran"/>
<date year="2016"/> <date year="2011"/>
</front> </front>
<seriesInfo name="IEEE International Conference on Communications (ICC)" val <seriesInfo name="DOI" value="10.1109/ICNP.2011.6089081"/>
ue="pp.1-6"/> <refcontent>19th IEEE International Conference on Network Protocols (ICN
P), pp. 49-58</refcontent>
</reference>
</reference> <reference anchor="Ren2018jordan">
<front>
<title>JORDAN: A Novel Traffic Engineering Algorithm for Dynamic Adapt
ive Streaming over HTTP</title>
<author fullname="Jing Ren"/>
<author fullname="Kejie Lu"/>
<author fullname="Cedric Westphal"/>
<author fullname="Jin Wang"/>
<author fullname="Jinfan Wang"/>
<author fullname="Tongyu Song"/>
<author fullname="Shucheng Liu"/>
<author fullname="Jianping Wang"/>
<date year="2018"/>
</front>
<seriesInfo name="DOI" value="10.1109/ICCNC.2018.8390337"/>
<refcontent>2018 International Conference on Computing, Networking and C
ommunications (ICNC), pp. 581-587</refcontent>
</reference>
<reference anchor="Mathew2011energy"> <reference anchor="Bianco2016energy">
<front> <front>
<title>Energy-Aware Load Balancing in Content Delivery Networks</title> <title>Energy consumption for data distribution in content delivery ne
<author fullname="Vimal Mathew"></author> tworks</title>
<author fullname="Ramesh Sitaraman"></author> <author fullname="Andreas Bianco"/>
<author fullname="Prashant Shenoy"></author> <author fullname="Reza Mashayekhi"/>
<date year="2011"/> <author fullname="Michele Meo"/>
</front> <date year="2016"/>
<seriesInfo name="CoRR http://arxiv.org/abs/1109.5641" value=""/> </front>
</reference> <refcontent>IEEE International Conference on Communications (ICC), pp. 1
-6</refcontent>
<seriesInfo name="DOI" value="10.1109/ICC.2016.7511356"/>
</reference>
<reference anchor="Islam2012evaluating"> <reference anchor="Mathew2011energy" target="https://arxiv.org/abs/1109.56
<front> 41">
<title>Evaluating Energy Consumption in CDN Servers</title> <front>
<author fullname="Saif ul Islam"></author> <title>Energy-Aware Load Balancing in Content Delivery Networks</title
<author fullname="Jean-Marc Pierson"></author> >
<date year="2012"/> <author fullname="Vimal Mathew"/>
</front> <author fullname="Ramesh Sitaraman"/>
<seriesInfo name="Proceedings of the Second International Conference on I <author fullname="Prashant Shenoy"/>
CT as Key Technology against Global Warming" value="pp.64-78"/> <date year="2011"/>
</reference> </front>
<seriesInfo name="DOI" value="10.48550/arXiv.1109.5641"/>
<seriesInfo name="arXiv:" value="1109.5641v1"/>
</reference>
<reference anchor="Ahlgren2012survey"> <reference anchor="Islam2012evaluating">
<front> <front>
<title>A survey of information-centric networking</title> <title>Evaluating Energy Consumption in CDN Servers</title>
<author fullname="Bengt Ahlgren"></author> <author fullname="Saif ul Islam"/>
<author fullname="Christian Dannewitz"></author> <author fullname="Jean-Marc Pierson"/>
<author fullname="Claudio Imbrenda"></author> <date year="2012"/>
<author fullname="Dirk Kutscher"></author> </front>
<author fullname="Borje Ohlman"></author> <seriesInfo name="DOI" value="10.1007/978-3-642-32606-6_6"/>
<date year="2012"/> <refcontent>Proceedings of the Second International Conference on ICT as Key Tec
</front> hnology against Global Warming, Lecture Notes in Computer Science, vol 7453, pp.
<seriesInfo name="IEEE Communications Magazine" value="Vol.50 No.7"/> 64-78</refcontent>
</reference> </reference>
<reference anchor="Wolf2014choicenet"> <reference anchor="Ahlgren2012survey">
<front> <front>
<title>ChoiceNet: Toward an Economy Plane for the Internet</title> <title>A survey of information-centric networking</title>
<author fullname="Wolf Tilman"></author> <author fullname="Bengt Ahlgren"/>
<author fullname="James Griffioen"></author> <author fullname="Christian Dannewitz"/>
<author fullname="Lenneth Calvert"></author> <author fullname="Claudio Imbrenda"/>
<author fullname="Rudra Dutta"></author> <author fullname="Dirk Kutscher"/>
<author fullname="George Rouskas"></author> <author fullname="Borje Ohlman"/>
<author fullname="Ilya Baldin"></author> <date year="2012"/>
<author fullname="Anna Nagurney"></author> </front>
<date year="2014" month="July"/> <refcontent>IEEE Communications Magazine, vol. 50, no. 7, pp. 26-36</ref
</front> content>
<seriesInfo name="SIGCOMM Computer Communciations Review" value="Vol.44 N <seriesInfo name="DOI" value="10.1109/MCOM.2012.6231276"/>
o.3"/> </reference>
</reference>
<reference anchor="Krol2017NFaaS"> <reference anchor="Wolf2014choicenet">
<front> <front>
<title>NFaaS: Named Function as a Service</title> <title>ChoiceNet: Toward an Economy Plane for the Internet</title>
<author fullname="Michal Krol"></author> <author fullname="Wolf Tilman"/>
<author fullname="Ioannis Psaras"></author> <author fullname="James Griffioen"/>
<date year="2017"/> <author fullname="Lenneth Calvert"/>
</front> <author fullname="Rudra Dutta"/>
<seriesInfo name="ACM SIGCOMM ICN Conference" value=""/> <author fullname="George Rouskas"/>
</reference> <author fullname="Ilya Baldin"/>
<author fullname="Anna Nagurney"/>
<date year="2014" month="July"/>
</front>
<seriesInfo name="DOI" value="10.1145/2656877.2656886"/>
<refcontent>ACM SIGCOMM Computer Communciations Review, vol. 44, no. 3,
pp. 58-65</refcontent>
</reference>
<reference anchor="TCC"> <reference anchor="Krol2017NFaaS">
<front> <front>
<title>Selective Noise Based Power Efficient and Effective Countermeasure Ag <title>NFaaS: Named Function as a Service</title>
ainst Thermal Covert Channel Attacks in Multi-Core Systems</title> <author fullname="Michal Krol"/>
<author fullname="Parisa Rahimi"></author> <author fullname="Ioannis Psaras"/>
<author fullname="Amit Kumar Singh"></author> <date year="2017"/>
<author fullname="Xioahang Wang"></author> </front>
<date year="2022"/> <refcontent>ICN '17: Proceedings of the 4th ACM Conference on Informatio
</front> n-Centric Networking, pp. 134-144</refcontent>
<seriesInfo name="Journal on Low Power Electronics and Applications" valu <seriesInfo name="DOI" value="10.1145/3125719.3125727"/>
e=""/> </reference>
</reference>
<reference anchor="DRAM"> <reference anchor="TCC">
<front> <front>
<title>Robust Covert Channels Based on DRAM Power Consumption</title> <title>Selective Noise Based Power-Efficient and Effective Countermeas
<author fullname="Thales Bandiera Paiva"></author> ure Against Thermal Covert Channel Attacks in Multi-Core Systems</title>
<author fullname="Javier Navaridas"></author> <author fullname="Parisa Rahimi"/>
<author fullname="Routo Terada"></author> <author fullname="Amit Kumar Singh"/>
<date year="2019"/> <author fullname="Xioahang Wang"/>
</front> <date year="2022"/>
<seriesInfo name="In book: Information Security (pp.319-338)" value=""/> </front>
</reference> <seriesInfo name="DOI" value="10.3390/jlpea12020025"/>
<refcontent>Journal on Low Power Electronics and Applications, vol. 12,
no. 2</refcontent>
</reference>
<reference anchor="NewClass"> <reference anchor="DRAM">
<front> <front>
<title>A New Class of Covert Channels Exploiting Power Management Vulnerabil <title>Robust Covert Channels Based on DRAM Power Consumption</title>
ities</title> <author fullname="Thales Bandiera Paiva"/>
<author fullname="S. Karen Khatamifard"></author> <author fullname="Javier Navaridas"/>
<author fullname="Longfei Wang"></author> <author fullname="Routo Terada"/>
<author fullname="Selcuk Kose"></author> <date year="2019"/>
<author fullname="Ulya R. Karpuzcu"></author> </front>
<date year="2018"/> <seriesInfo name="DOI" value="10.1007/978-3-030-30215-3_16"/>
</front> <refcontent>Information Security, ISC 2019, Lecture Notes in Computer Sc
<seriesInfo name="IEEE Computer Architecture Letters" value=""/> ience, vol 11723</refcontent>
</reference> </reference>
<reference anchor="SideChannel"> <reference anchor="NewClass">
<front> <front>
<title>Power Side-Channel Attack Analysis: A Review of 20 Years of Study for <title>A New Class of Covert Channels Exploiting Power Management Vuln
the Layman</title> erabilities</title>
<author fullname="Mark Randolph"></author> <author fullname="S. Karen Khatamifard"/>
<author fullname="William Diehl"></author> <author fullname="Longfei Wang"/>
<date year="2020"/> <author fullname="Selcuk Kose"/>
</front> <author fullname="Ulya R. Karpuzcu"/>
<seriesInfo name="Cryptography 2020, 4, 15" value=""/> <date year="2018"/>
</reference> </front>
<seriesInfo name="DOI" value="10.1109/LCA.2018.2860006"/>
<refcontent>IEEE Computer Architecture Letters, vol. 15, no. 2, pp. 201-
204</refcontent>
</reference>
<reference anchor="Framework"> <reference anchor="SideChannel">
<front> <front>
<title> A framework to estimate emissions from virtual conferences</title> <title>Power Side-Channel Attack Analysis: A Review of 20 Years of Stu
<author fullname="Grant Faber"></author> dy for the Layman</title>
<date year="2021"/> <author fullname="Mark Randolph"/>
</front> <author fullname="William Diehl"/>
<seriesInfo name="International Journal of Environmental Studies, 78:4, 6 <date year="2020"/>
08-623" value=""/> </front>
</reference> <refcontent>Cryptography, vol. 4, no. 2</refcontent>
<seriesInfo name="DOI" value="10.3390/cryptography4020015"/>
</reference>
<reference anchor="Emergy"> <reference anchor="Framework">
<front> <front>
<title>The Energy and Emergy of the Internet</title> <title>A framework to estimate emissions from virtual conferences</tit
<author fullname="Barath Raghavan"></author> le>
<author fullname="Justin Ma"></author> <author fullname="Grant Faber"/>
<date year="2011"/> <date year="2021"/>
</front> </front>
<seriesInfo name="ACM HotNets" value=""/> <seriesInfo name="DOI" value="10.1080/00207233.2020.1864190"/>
</reference> <refcontent>International Journal of Environmental Studies, vol. 78, no.
4, pp. 608-623</refcontent>
</reference>
<reference anchor="Junkyard"> <reference anchor="Emergy">
<front> <front>
<title>Architecture of a Junkyard Datacenter</title> <title>The Energy and Emergy of the Internet</title>
<author fullname="Jennifer Switzer"></author> <author fullname="Barath Raghavan"/>
<author fullname="Ryan Kastner"></author> <author fullname="Justin Ma"/>
<author fullname="Pat Pannuto"></author> <date year="2011"/>
<date year="2021"/> </front>
</front> <seriesInfo name="DOI" value="10.1145/2070562.2070571"/>
<seriesInfo name="arXiv:2110.06870v1, October 2021" value=""/> <refcontent>Proceedings of the 10th ACM Workshop on Hot Topics in Networ
</reference> ks, no. 9, pp. 1-6</refcontent>
</reference>
<reference anchor="TradeOff"> <reference anchor="Junkyard" target="https://arxiv.org/abs/2110.06870v1">
<front> <front>
<title>Not a Trade-Off: On the Wi-Fi Energy Efficiency of Effective Internet <title>Architecture of a Junkyard Datacenter</title>
Congestion Control</title> <author fullname="Jennifer Switzer"/>
<author fullname="Michael Welzl"></author> <author fullname="Ryan Kastner"/>
<date year="2022"/> <author fullname="Pat Pannuto"/>
</front> <date year="2021"/>
<seriesInfo name="IEEE/IFIP WONS" value=""/> </front>
</reference> <seriesInfo name="DOI" value="10.48550/arXiv.2110.06870"/>
<seriesInfo name="arXiv:" value="2110.06870v1"/>
</reference>
<reference anchor="Hossain2019" target="https://ieeexplore.ieee.org/docu <reference anchor="TradeOff">
ment/6779082"> <front>
<title>Not a Trade-Off: On the Wi-Fi Energy Efficiency of Effective In
ternet Congestion Control</title>
<author fullname="Michael Welzl"/>
<date year="2022"/>
</front>
<refcontent>2022 17th Wireless On-Demand Network Systems and Services Co
nference (WONS), pp. 1-4</refcontent>
<seriesInfo name="DOI" value="10.23919/WONS54113.2022.9764413"/>
</reference>
<reference anchor="Hossain2019" target="https://hal.science/hal-02169758/"
>
<front> <front>
<title>Energy, Carbon and Renewable Energy: Candidate Metrics for Gree n-Aware Routing?</title> <title>Energy, Carbon and Renewable Energy: Candidate Metrics for Gree n-Aware Routing?</title>
<author fullname="M.M. Hossain"> </author> <author fullname="M.M. Hossain"> </author>
<author fullname="J.-P. Georges"> </author> <author fullname="J.-P. Georges"> </author>
<author fullname="E. Rondeau"> </author> <author fullname="E. Rondeau"> </author>
<author fullname="T. Divoux"> </author> <author fullname="T. Divoux"> </author>
<date month="June" year="2019"/> <date month="June" year="2019"/>
</front> </front>
<seriesInfo name="DOI" value="10.3390/s19132901"/> <seriesInfo name="DOI" value="10.3390/s19132901"/>
<seriesInfo name="Sensors" value="Vol. 19 No. 3"/> <refcontent>Sensors, vol. 19, no. 3</refcontent>
</reference> </reference>
<reference anchor="IETF-Net0" target="https://www.ietf.org/blog/towards-a-net- <reference anchor="IETF-Net0" target="https://www.ietf.org/blog/towards-a-
zero-ietf/"> net-zero-ietf/">
<front> <front>
<title>Towards a net zero IETF</title> <title>Towards a net zero IETF</title>
<author fullname="Jay Daley"></author> <author fullname="Jay Daley"/>
<date day="6" month="5" year="2022"/> <date day="6" month="5" year="2022"/>
</front> </front>
<seriesInfo name="IETF News" value=""/> <refcontent>IETF News</refcontent>
</reference> </reference>
</references> <!-- [rfced] note that we updated this to list July 1948 only and included DOI i nfo. Please let us know any concerns.
</back> DOI: landing page
https://ieeexplore.ieee.org/document/6773024
-->
<reference anchor="Shannon">
<front>
<title>A Mathematical Theory of Communication</title>
<author fullname="C. E. Shannon">
<organization />
</author>
<date month="July" year="1948" />
</front>
<seriesInfo name="DOI" value="10.1002/j.1538-7305.1948.tb01338.x" />
<refcontent>The Bell System Technical Journal, vol. 27, no. 3, pp. 379-423<
/refcontent>
</reference>
</references>
<section numbered="false">
<name>Acknowledgments</name>
<t>The authors thank <contact fullname="Dave Oran"/> for providing the
information regarding covert channels using energy measurements and
<contact fullname="Mike King"/> for an exceptionally thorough review and
useful comments.</t>
</section>
<section numbered="false">
<name>Contributors</name>
<contact fullname="Michael Welzl">
<organization>University of Oslo</organization>
<address>
<email>michawe@ifi.uio.no</email>
</address>
</contact>
</section>
</back>
</rfc> </rfc>
 End of changes. 193 change blocks. 
1337 lines changed or deleted 1657 lines changed or added

This html diff was produced by rfcdiff 1.48.