[SCION] Carbon aware SCION networking

[Seyedali Tabaeiaghdaei]

Hi Chris,

It is 7 GW, so it is 7 GJ/sec. Yes, our calculation showed that we can save
around 30 % of core network power consumption. By core I do not mean SCION
core, I mean IP-over-WDM networks including core routers. I shared the
document (an overleaf doc) with your email address. Our base numbers and
model come from previous papers in this field and I tried to be as accurate
and as realistic as possible. I would appreciate it if you help us on
finding more realistic/accurate/up-to-date numbers or models.

Best Regards,

Seyedali

On Wed, Mar 24, 2021 at 2:29 PM Chris Adams <chris at thegreenwebfoundation.org>
wrote:

> Thanks for this  Seyadali,
>
> It’s really helpful.
>
> The first one which is my Master's thesis exactly investigates what you
> asked in your email. We're putting CO2 emission per bit of intra-AS paths
> between ingress and egress interfaces of each AS hop into the PCB. The
> total CO2 emission per bit of the path would be the sum of these numbers
> for all hops on the path. A simple beaconing algorithm is also needed to
> select and propagate the greenest set of paths. (We cannot propagate all
> paths due to scalability issue, so we should propagate the greenest ones)
>
> We are also investigating the effect of traffic shift from dirty ASes to
> green ASes on the behvior of  dirty ones and their willingness to become
> greener. A simple model is proposed for that in my thesis. We will use a
> more sophisticated model in the later work.
>
> As this is not yet published, I'll send it to you as soon as it is
> published.
>
> The second project which is older investigates the energy consumption of
> SCION ASes. We showed that SCION can save around 7 GW globally mostly by
> removing routing tables from routers. There is a report for that which is
> not official. We may publish it in the future.
>
>
> Can I check with you on this figure please? Would this be 7GW, in terms of
> draw at a given moment, or 7 gigawatt hours over a given time period, like
> a year?
>
> If it’s 7GW draw, and we’re assuming this would be an average for a year,
> that’s then we end up with (7 * 24 * 365) =~ 61 terawatt hours.
>
> The IEA’s estimates for all the energy use by data transmission networks
> to be around 270 TWh in 2022 - so this would be like wiping 25% from that
> total energy demand 🤯
>
> https://www.iea.org/reports/data-centres-and-data-transmission-networks
>
> Would you mind linking to that report? I hadn’t seen that figure before
> and I’d like to read it in more detail.
>
>
> Chris
>