Re: [E-impact] Lecture materials on Sustainable Networking

[Dom Robinson <[e-mail]>]

Hey Hesham 

So in terms of the top line question it’s a little ambiguous: I think you could apply CLCA or ALCA approaches to underlying models in each scenario. I don’t think any are inherently ‘just’ CLCA. For example again we are talking about energy savings from reduced data transmission rears its head again in #5. That is absolutely attributional. There is no demonstrable  consequential reduction in energy that I know of that relates to reducing traffic (not the same as ’turning off kit that is generating traffic’). 

Not increasing capacity might have a deferred infrastructure benefit, but even that is hard to prove with efficiency increasing as kit innovates.

Im not going to unpick them all given I may not quite understand the question, but a comment on #1:

#1 - 5G provides 100x capacity (Gbps), but in order to deliver this you need 3-5x the power of 4G. Consequence is that 5G is 3-5x more energy hungry than 4G. Attributional model would say that 5G is 15-25x more efficient per Gbps “WHEN THE NETWORK IS AT 100% utilisation” but the attribution model ignores that the 5G network, when carrying only the traffic that was formerly on 4G will be 99% underutilised. So whichever way you look at it - for a long time at least (until the network is put to full use)  - 5G is going to (pointlessly) burn a hole in our CLCA even if Attributional models look fun on marketing.

Not sure if that helps, but it’s a few thoughts!

Dom




-- 
Dom Robinson
www.id3as.co.uk <http://www.id3as.co.uk/> 
www.greeningofstreaming.org <http://www.greeningofstreaming.org/> 
uk.linkedin.com/in/domrobinson <http://uk.linkedin.com/in/domrobinson>
Meet >> https://calendly.com/id3as 

> On 5 Jan 2024, at 10:50, Hesham ElBakoury <[e-mail]> wrote:
> 
> What do you think of the following 5 scenarios for consequential LCA (I got these scenarios using Google bard).
> 
> Scenario #1: A telecommunications company is considering upgrading its network infrastructure to 5G technology. They want to understand the potential environmental impacts of this decision, taking into account both direct and indirect effects.
> 
> Scenario #2: A government is considering a policy to promote the expansion of fiber-optic broadband internet to rural areas. They want to assess the potential environmental impacts of this policy, including both positive and negative consequences.
> 
> Scenario #3: A large data center operator is considering investing in renewable energy sources to power its facilities. They want to quantify the potential environmental benefits of this decision, taking into account the full life cycle impacts of various energy options.
> 
> Scenario #4: A satellite internet provider is planning to launch a new constellation of satellites to expand global broadband coverage. They want to assess the environmental footprint of this project, considering both the direct impacts of manufacturing, launching, and operating the satellites, as well as the indirect impacts on energy consumption, resource use, and potential space debris.
> 
> Scenario #5: An online video streaming service is considering adding a feature that automatically adjusts video quality based on viewers' network bandwidth. They want to understand the potential environmental implications of this feature, considering both energy savings from reduced data transmission and potential rebound effects from increased video playback time.
> 
> Hesham
> 
> On Wed, Jan 3, 2024, 4:16 AM Dom Robinson <[e-mail] <mailto:[e-mail]>> wrote:
>> Great point Sebastien : I hold my hand up to the fact that in Greening of Streaming we are myopic (on purpose) about Operational Energy (allow me to coin ‘Op-Energy’ for brevity).  (Our members who work with physical supply chains pick us up on embedded / embodied emissions as we go, but the strategies are largely controlled by individual members and their supply chains rather than having a ‘generic streaming strategy’ that we can collectively work to in our org.. )
>> 
>> My own car analogy breaks down when we move the model back to network services. There are no parallels to  ‘cars’ that become active purely for a specific task and then get parked.  There is an always-on network and an always-on cloud - infrastructure running like busses all the time. 
>> 
>> This makes identifying consequential sustainability impact models for network applications architects very difficult. (Attributional models are by comparison trivial since all you have to do is pick a KPI and imagine a CO equivalence you feel comfy defending) 
>> 
>> My personal list of consequential activities relating to network streaming looks like this at the moment (happy to hear thoughts of others):
>> 
>> Not building in the first place is … too late - we already have an Internet. we can (easily) ‘undo’ that. (We also have a pile of other resources (lithium and more) that have been mined. We shouldn’t be myopic about Carbon.)
>> Upgrading frivolously is clearly negatively consequential. Who needs 1Gbps to watch 16k 240fps Pepper Pig on a phone… Upgrading just to 'out-piss' the competitions marketing is tired. The internet is big enough for most purposes at the moment... (bold, but I’ll defend that quite hard).
>> However never upgrading is also missing out on the consequential benefit of new more energy efficient generations of technology. Imagine EC2 running on Pentium IIs (!) There are moments / generative cycles when tungsten bulbs need to be replaced with LEDs (or whatever the network upgrade equivalent is)
>> And of course upgrading with recycling can be consequential to the embodied / embedded supply chains.
>> So  'turning off’ unused systems remains #1 consequential action IMHO. Even accounting for Embodied emissions in the systems, the Op Energy is continuing to generate CO2 - at least until 100% renewable energy available everywhere. Turning off halts that continuation. Focus on load concentration, spin-down of unused compute resources and possibly a lot of review of SLAs and how they effect ‘over-availability’ (triple redundancy etc) all seem like central considerations for network service operators seeking to improve sustainability positions.
>> 
>> Those, in my mind, are the areas where we can make consequential differences in network systems / service architecture.
>> 
>> Other measures i see relating digital media to sustainability might have ‘attributional’ effects, but if they are not directly correlating to the consequential outcomes on the list above then I struggle to see what the attribution model value is, other than relieving guilt because we can account things "off our scope 2 respoinsibilitiy" or making for some good greenwashing resources..
>> 
>> … 
>> 
>> But yes: Embedded / embodied CO in the infrastructure was not captured in my car analogy!
>> 
>> Dom
>> -- 
>> Dom Robinson
>> www.id3as.co.uk <http://www.id3as.co.uk/> 
>> www.greeningofstreaming.org <http://www.greeningofstreaming.org/> 
>> uk.linkedin.com/in/domrobinson <http://uk.linkedin.com/in/domrobinson>
>> Meet >> https://calendly.com/id3as 
>> 
>>> On 2 Jan 2024, at 21:16, Sébastien Rumley <sebastien.rumley=[e-mail] <mailto:[e-mail]>> wrote:
>>> 
>>> Hi all,
>>> 
>>> Happy new year.
>>> 
>>> Thanks to Dom for this explanation about consequential vs. attributional. The train/car example is a good one.
>>> 
>>> Yet... there is something that bothers me. Quoting Dom :
>>> 
>>> "I take the car 5 days to work. Now i work from home 3 days. I have ‘actually’ reduced my CO2 production by 3/5ths."
>>> 
>>> This is actually not true, because that is taking into account "operational CO2" only, and not the embodied CO2. The car is still there, it has been produced. And... will its life expectancy be extended by 3/5th ?
>>> 
>>> And how to account for the CO2 embodied in building the supply chain for assembling the car ? And the roads ? Should one expect eventually the supply chain/roads to be scaled-down by 3/5th of a unit ? Any hints on how-to apply the attributional/consequential on embodied CO2?
>>> 
>>> 
>>> 
>>> This reminds me the subtle question "should embodied CO2 expenses be amortized over the years of the life expectancy, or accounted full dollar at acquisition time" ? I guess consequential would mean : I get the full embodied CO2 expense at the moment I pay the car (if I end up no buying a new car --> no CO2...). Attributional : I can amortize over the years.... ?
>>> 
>>> Quoting Dom again : "I actually think this issue is endemic and without MUCH clearer separation in discussion we will waste a lot of our time and energy focussing on attributional fuzzy strategies rather than on real world consequential ones."
>>> 
>>> I fully agree.
>>> 
>>> Sébastien
>>> 
>>> 
>>> 
>>> 
>>> 
>>> On 24.12.23 08:07, Dom Robinson wrote:
>>>> Carlos at al.
>>>> 
>>>> Sustainability Considerations for Internetworking
>>>> datatracker.ietf.org
>>>> <ietf-logo-nor-180.png>
>>>>  <https://datatracker.ietf.org/doc/html/draft-cparsk-eimpact-sustainability-considerations>Sustainability Considerations for Internetworking <https://datatracker.ietf.org/doc/html/draft-cparsk-eimpact-sustainability-considerations>
>>>> datatracker.ietf.org <https://datatracker.ietf.org/doc/html/draft-cparsk-eimpact-sustainability-considerations>	<ietf-logo-nor-180.png> <https://datatracker.ietf.org/doc/html/draft-cparsk-eimpact-sustainability-considerations>
>>>> 
>>>> Great doc.
>>>> 
>>>> I think the terms section should break out Consequential LCA and Attributional LCA.
>>>> 
>>>> IMHO ‘most’ of all discussion around sustainability and Internet service provisioning are conflating the two, and causing serious problems. 
>>>> 
>>>> Attributional models lead to offsetting and guilt reduction, dis-ownership of the responsibility through accounting and crafty ‘scope 3’ delegation in order to self badge ‘net zero’ for marketing cachet. It is pretty much synonymous with Green Washing in my view.
>>>> 
>>>> Consequential models are much less involved with green washing and seek to ‘actually’ reduce impacts. These are the projects worth doing because they make a real difference. They take ownership of the problem and fix it.
>>>> 
>>>> But while both models use same terms and frames of reference, measures  and language they are prone to conflation and misleading intention for that reason.
>>>> 
>>>> Attributional: I take the train 5 days a week to work, now i work from home 3 days a week: i have reduced my own carbon foot print by 3/5, but the train is still producing the carbon. Less guilt and great PR but no real world reduction in CO2...
>>>> 
>>>> Consequential: I take the car 5 days to work. Now i work from home 3 days. I have ‘actually’ reduced my CO2 production by 3/5ths.
>>>> 
>>>> It really reminds me of the energy / GB debates that have at times raged in this group.
>>>> 
>>>> Carbon calculators are almost universally attributional. So any organisation using them to measure success ends up working toward alleviating guilt, and disowning responsibility. ‘How much can we move to scope 3 by accounting / offsetting’ etc. It leads to ‘Net Zero’ rather than ‘Zero’
>>>> 
>>>> Even in this group it was only after a year of debates - once Chris Adams mentioned CLCA vs ALCA - that this conflation became apparent to me and I personally realised that there are often two separate discussions going on at cross purposes yet using the same data and same language in interpretations because of this. Since then I have put a concerted effort in to isolating the CLCA from the ALCA discussions and while 75% of the noise is from ALCA 100% of the actual progress toward improvement is framed by CLCA. So much so that we are carefully checking our work at Greening of Streaming so far to make sure we know which model frames each of our efforts (and refocussing on CLCA where possible).
>>>> 
>>>> I actually think this issue is endemic and without MUCH clearer separation in discussion we will waste a lot of our time and energy focussing on attributional fuzzy strategies rather than on real world consequential ones. 
>>>> 
>>>> So i think something of this issue should be included in the ‘LCA’ definition of the document draft you ate working on..
>>>> 
>>>> Seasons greetings all! A great year of discussions in this group: thank you all!
>>>> 
>>>> -- 
>>>> Dom Robinson
>>>> www.id3as.co.uk <http://www.id3as.co.uk/> 
>>>> www.greeningofstreaming.org <http://www.greeningofstreaming.org/> 
>>>> uk.linkedin.com/in/domrobinson <http://uk.linkedin.com/in/domrobinson>
>>>> Meet >> https://calendly.com/id3as 
>>>> 
>>>>> On 23 Dec 2023, at 18:50, Carlos Pignataro <[e-mail]> <mailto:[e-mail]> wrote:
>>>>> 
>>>>> Thanks so much for sharing this, Romain!!!
>>>>> 
>>>>> Maarten,
>>>>> 
>>>>> You raise an important point, and one that we’ve started to explore in Sections 6.1-6.3 of https://datatracker.ietf.org/doc/html/draft-cparsk-eimpact-sustainability-considerations
>>>>> 
>>>>> Thanks!
>>>>> 
>>>>> Carlos.
>>>>> 
>>>>>> On Dec 21, 2023, at 11:33 AM, [e-mail] <mailto:[e-mail]> wrote:
>>>>>> 
>>>>>> Hi Romain,
>>>>>>  
>>>>>> Thanks for sharing. I learned a lot. 
>>>>>>  
>>>>>> One specific issue with the energy consumption of networking/routing/etc is indeed the purposefully build-in redundancy. You refer a bit to it in slide 13, I believe. It just made me think that it may be needed to have a better understanding of “justifiable redundancy” balancing the “adequate redundancy to avoid network outings” against energy use.
>>>>>>  
>>>>>> I believe we have been thinking in terms of “zero tolerance of network outings” more than the costs of energy use that it brings. Another hard nut to crack.
>>>>>>  
>>>>>> Best
>>>>>>  
>>>>>> Maarten
>>>>>>  
>>>>>> Maarten Botterman
>>>>>> GNKS Consult BV
>>>>>>  
>>>>>>  
>>>>>>  
>>>>>>  
>>>>>> From: E-impact <[e-mail] <mailto:[e-mail]>> On Behalf Of Romain Jacob
>>>>>> Sent: Thursday, December 21, 2023 3:28 PM
>>>>>> To: [e-mail] <mailto:[e-mail]>; [e-mail] <mailto:[e-mail]>
>>>>>> Subject: [E-impact] Lecture materials on Sustainable Networking
>>>>>>  
>>>>>> Hello everyone, 
>>>>>> As mentioned before, I prepared and delivered a 2h lecture on Sustainable Networking for Master's students. The material draws quite a bit from references shared and discussions we had on its mailing list, so I'd like to start by thanking you all for that :-) 
>>>>>> The presentation PDF [1] and sources are openly available [2]. You're welcome to take/adapt/reuse. I'd be happy to know if you do so! 
>>>>>> Cheers,
>>>>>> -- 
>>>>>> Romain
>>>>>> [1] https://polybox.ethz.ch/index.php/s/OreTzMWYLmVMrjb
>>>>>> [2] https://osf.io/pnmq5
>>>>>> -- 
>>>>>> Romain JACOB
>>>>>> Postdoctoral Researcher
>>>>>> ETH Zurich
>>>>>> Networked Systems Group (NSG) 
>>>>>> Lead: Prof. Laurent Vanbever
>>>>>> www.romainjacob.net <https://www.romainjacob.net/>
>>>>>> @RJacobPartner <https://twitter.com/RJacobPartner>
>>>>>> @[e-mail] <https://discuss.systems/@jacobr>
>>>>>> Gloriastrasse 35, ETZ G81
>>>>>> 8092 Zurich 
>>>>>> [phone] 
>>>>>> -- 
>>>>>> E-impact mailing list
>>>>>> [e-mail] <mailto:[e-mail]>
>>>>>> https://www.ietf.org/mailman/listinfo/e-impact
>>>>> 
>>>>> -- 
>>>>> E-impact mailing list
>>>>> [e-mail] <mailto:[e-mail]>
>>>>> https://www.ietf.org/mailman/listinfo/e-impact
>>>> 
>>> -- 
>>> Dr. Sébastien RUMLEY, Professor
>>> iCoSys Institute, part of
>>> HEIA-FR - School of Engineering and Architecture - Fribourg, part of
>>> HES-SO - University of Applied Sciences and Arts Western Switzerland
>>> -- 
>>> E-impact mailing list
>>> [e-mail] <mailto:[e-mail]>
>>> https://www.ietf.org/mailman/listinfo/e-impact
>> 
>> -- 
>> E-impact mailing list
>> [e-mail] <mailto:[e-mail]>
>> https://www.ietf.org/mailman/listinfo/e-impact