Minutes interim-2024-eimpact-01: Thu 16:00
minutes-interim-2024-eimpact-01-202402151600-01

Administrivia
=============

Suresh described the scope of this first session to be existing drafts, where
they exist, how they fit together and to identify any gaps. He also mentioned
that Jari and him will be updating the Gap analysis slides during the meeting
slot and will summarize at the end. Jari also wanted all the participants to
chime in with their contributions during the gap analysis discussions.

Map/survey of existing drafts
=============================

Alexander Clemm presented a Map/survey of existing drafts in the sustainability
space. The slides for this presentation can be found at
https://datatracker.ietf.org/meeting/interim-2024-eimpact-01/materials/slides-interim-2024-eimpact-01-sessa-alex-map-survey-01.pdf

He described the overall landscape of sustainability work in the IETF and the
IRTF and attempted to categorize the drafts into three high level categories:
Foundational, Instrumentation & Visibility, and Programmability & Control.

Jari mentioned that there was a need for some kind of a trade-off between the
cost of collecting some information and the benefit derived from it. He also
felt that there was some missing work related to programmability of the
networks related to these metrics. Alex agreed and also mentioned the
distinction between assurance and fulfilment. Jari agreed about the assurance
part, but wanted to focus a bit more on influencing the outcomes before
assurance. Suresh also agreed with Jari and talked about how metrics can be
used to effect changes such as influencing routing metrics, duty cycles etc.
Suresh promised to add the listed drafts from this presentation into the
eimpact related documents datatracker page, and requested people with drafts in
this space to send a message to the chairs to include their drafts there.
Alexander Pelov wondered why embedded energy in manufacturing the network
devices was not addressed and wanted to understand the timescales for action
being considered. He wanted some significant actions to be done in the near
term. Alex C. mentioned some of the network inventory work being done in ivy as
being relevant for the embedded energy measurements. Suresh mentioned the
lifecycle management drafts from Marisol might also be relevant for the
embedded energy topic. He also mentioned that short term work will head to the
IETF and longer term work will probably need to be done at the IRTF. Marisol
mentioned the POWEFF draft and addressed both embedded energy and dynamic
metrics.

Sustainability Insights & POWEFF
================================

Marisol Palmero and Jan Lindblad presented the topic titled Goals and
Principles of IETF e-impact. The slides for this presentation can be found at
https://datatracker.ietf.org/meeting/interim-2024-eimpact-01/materials/slides-interim-2024-eimpact-01-sessa-marisol-jan-e-impact-principles-01.pdf

He believed that eimpact should maintain a directory of relevant WGs, drafts
and RFCs and detect overlaps. He also thought that the program could help with
education, policymaker engagement, and potentially write down position papers
and architectural principles.

Alex P. wanted to know how we can co-ordinate with other SDOs who are also
doing work in this space. Ali Rezaki wondered how ISOC could play a bigger role
in the sustainability space given that they do not seem to have any focused
activities. Dan York from ISOC mentioned he was partcipating as an individual
in this meeting and there are also others who are looking at these topics. He
also mentioned that the IAB has some input into the future activities of the
ISOC through the board of trustees. Jari mentioned that the key part of the
execution is for eimpact to handoff things to responsible parties and hence
eimpact may not be the place where everything happens. He also mentioned that
the IETF hacakthon would be a good place to start work on some common metrics.
Suresh mentioned the IAB Vision document and would be willing to document our
views on sustainability there once there is community consensus on what to
pursue. Toerless believed that there is a role for ISOC to play and it can be
achieved by individuals participating in the relevant ISOC mailing lists. Vesna
invited Jan to participate in RIPE meetings on the topic and would share
details later.

Power Management YANG
=====================

Ron Bonica presented the draft covering regarding a YANG model for Power
Management. The slides for this presentation can be found at
https://datatracker.ietf.org/meeting/interim-2024-eimpact-01/materials/slides-interim-2024-eimpact-01-sessa-ron-tony-power-management-yang-00.pdf

The goal of this draft is to enable detailed power management for network
elements and be able to report power consumption on a per component basis.

Suresh thought this was a good start but a bit simplistic on the power save
front given that there are multiple different levels of power saving possible
for a lot of components. e.g. turning off transceivers or dropping them to a
lower speed or turning off whole line cards. Tony Li mentioned that their
philosophy is to model this in a recursive manner so that each component can be
individually modeled. Toerless mentioned some of the work in anima and thought
a double commit scheme would be useful here to try out energy savings and
revert if things break. Jari also things that a binary power save is too simple
and may ignore hardware capabilities that could be useful. Alex C. also thinks
the draft is a bit simplistic and would like some hooks in it for
extensibility. He also thought there would be a need to manage changing
dependencies in dynamic scenarios. Tony mentioned that the dependencies can
certainly be updated dynamically but are not expected to be updated at a high
frequency.

Sustainability considerations
=============================

Carlos Pignataro presented the draft covering Sustainability Considerations for
Internetworking. The slides for this presentation can be found at
https://datatracker.ietf.org/meeting/interim-2024-eimpact-01/materials/slides-interim-2024-eimpact-01-sessa-carlos-ali-sustainability-considerations-00.pdf

The document aims to address three major areas in sustainability including
terminology, sustainability considerations, and guidelines for protocol
developers.

Suresh wanted to separate the advice parts of the document from the background
information parts. He also mentioned that IAB informational documents will not
be able to put requirements on the IETF stream. Carlos mentioned that he was
open to splitting this document into different pieces. Jari agreed with Suresh
and wanted to spend some more time scoping this document. He thought that
agreeing what to put into a recommendation is more important than the process
of getting the IETF to add such recommended text. Ali thought that thinking
about how the draft would be used could be a good guideline on how to divide
the draft into different sections/documents. Suresh agreed with Ali and thought
that any architectural guidance, such as tradeoffs between availability or
performance & sustanability, resulting out of this document would be
potentially in scope for the IAB. Marisol mentioned that Jan Lindblad shared
some sustainability principles at the Prague eimpact meeting and these might be
useful to include in the draft. Carlos recapped and agreed with Vesna that the
document could be a bit overwhelming and hence might be good to partition it.

ICMP Environmental Impact extensions
====================================

Jainam Parikh presented the draft proposing ICMP Extensions for Environmental
Impact. The slides for this presentation can be found at
https://datatracker.ietf.org/meeting/interim-2024-eimpact-01/materials/slides-interim-2024-eimpact-01-sessa-carlos-icmp-extensions-eimpact-00.pdf

The draft defines the Environmental Impact Object, an ICMP extension object
that can be appended to selected ICMPv4 and ICMPv6 messages.

Michael Welzl wanted to avoid redefining power metrics for ICMP and preferred
using previously defined YANG models to ask for specific metrics by reference.
Jari agreed with Michael that sharing the concepts and formats of metrics was a
good idea. He also wanted to spend some time to think what are the right
protocols (ICMP might be one of them) to carry this sort of information. Rudolf
van der Berg thought that this was not useful as it does not provide any
actionable intelligence to other networks since they may not have complete
information about the network originating this message. Carlos mentioned that
this is not scoped to be deployed Internet-wide but rather in constrained use
cases under a single administrive domain. Toerless thought it would be good to
start with the use cases where this work could be used. Suresh addressed
Rudolf's comments and thought it would be good for Carlos and Jainam to
investigate further on how to collate information from multiple sources so that
they could be made useful.

Gap Analysis
============

The chairs had identified some gaps prior to the meeting and had updated the
gaps with information presented during the meeting. Jari presented the high
level gap analysis. The gaps can be divided broadly into three categories:

a) Metrics
* Observability is step 1. For that we need to get metrics *done*
* Need to engage & work on open source to help this, not just standards
* We’re maybe overly focused on energy, what about other aspects of
sustainability

b) Other
* We need an ability to influence systems as well, not just observe
* Lots of details in different power save modes
* Consider how you transport information & integration to other YANG models etc.
* Consider what data can be trusted and how (cross-domain in particular)

c) Social
* Perfect is the enemy of good! Let’s not overdo it!
* More interaction with policy makers, ISOC, rest of the world
* Find a way to publish documents from this program (when not technical YANG
docs. etc)

Open Mic
========

Toerless mentioned variances between multiple router vendors on operating
states and believed that having an enumeration of different states could be a
good way to start simple and still be extensible. Suresh mentioned that people
might be modeling power states differently as well (and pointed to the Ron/Tony
presentation earlier). Alexander Pelov wanted to focus on the short term simple
things so that we can have an immediate effect, and work on more complex
solutions and research in parallel. Suresh mentioned that a lot of research has
already happened in this area and we should be open to using the results for
near term solutions. Marisol mentioned that she was working with customers on
the metrics defined in the POWEFF draft and would like to discuss refining
those metrics in a future side meeting and/or an open source project. Vesna
wanted the program to influence ongoing IETF/IRTF work on technologies and
guide them towards sustainable practices and also push back on unsustainable
work. The example provided was Bitcoin - not a product of IETF/IRTF work but
Vesna noted that IETF/IRTF might be able to influence work at other venues as
well by providing inputs. Jari also pointed out that the IETF is a consensus
based organization and we may not always be able to gather consensus against
unsustainable practices. Alex P. mentioned that it is important to focus on
simple proposals that can provide large improvements in sustainability and
worry about small incremental gains later. Suresh mentioned that any
improvements in sustainability are useful, and it might be OK to go for smaller
gains as long as they are not blocking work on larger gains. Alex P. asked
about open sourcing the collected data. Suresh mentioned that the data might be
sensitive and might need to be anonymized and it needs to be done by the
operators and not the vendors and hence an operator forum might be a good place
to discuss that. Ali mentioned that one of our goals needs to be collecting
granular metrics for the future that would allow companies to better report
their GHG impacts. He also wanted to work on metrics that would help users to
be more aware of their impact and potentially change their behaviors. Michael
Welzl talked about the practice of overprovisioning in networks given that it
is the cheaper and easier way to provide given service levels. He wanted the
networks to move to more intelligent ways of management to make them more
sustainable. Marisol agreed with Ali's message regarding reporting. She
believed that we could provide accurate metrics that would avoid a lot of
assumptions currently used in reporting. On the point of overprovisioning
Rudolf mentioned that there are economic, legal and regulatory constraints that
may make it difficult to avoid overprovisioning. He mentioned that the
economics are best left to individual companies. Suresh mentioned that our goal
would be to provide the tools for individual companies to make better
decisions. Toerless mentioned that unlike CPUs recently we still have room in
networking gear to still be able to improve performance while reducing
environmental impacts. He also believed that in addition to run time metrics we
need to help with tools for planning departments of operators so that they can
study the financial impacts and build out more efficient networks from the
start.