Minutes interim-2022-eimpactws-02: Thu 14:00
minutes-interim-2022-eimpactws-02-202212081400-00

IAB E-Impact Workshop Session 2: What do we know

5:53 am - 8:00 am  Thursday, December 8, 2022 | (UTC-08:00) Pacific Time (US &
Canada)

Cedric Westphal
heb
Maya Richman
Eve Schooler
Hesham ElBakoury
Jari Arkko  Safiqul
Uri Cummings
John Preuß Mattsson
Michael Welzl
Jens Malmodin
Lars Eggert
Bruce Nordman2
Marisol Palmero Amador
Qin Wu
Call-in User_2
Eric Voit
Louise Krug
Henk Birkholz he/him
Nina Lövehagen
Shayna Robinson
Hosein Badran ISOC
Ali Rezaki
Cindy Morgan
Snezana Mitrovic
Eric Vyncke
Rob Wilton
Selome
Romain
Louis Navarre
Jiankang Yao
Beatrice Siccardi
Fieke
Colin Perkins
Pascal Thubert
olivier.bonaventure
Jan Lindblad
Bruce
Vesna Manojlovic
Peyman Teymoori
Esther Roure Vila
Dom Robinson
Lijun Dong
Carsten Bormann
Daniel Schien
Mohamed Boucadair
Laurent Ciavaglia
Mike Mattera
Paul Shabajee
Brendan Moran
Carlos Pignataro
Chris Adams
Wim Vanderbauwhede
Per Andersson
Gonzalo Salgueiro
Dawn Nafus
Suresh Krishnan
Noa Zilberman
Alex Clemm
Pernilla Bergmark Ericsson

WEBVTT

1
Daniel Schien 00:03:31.654 --> 00:03:34.744
Testing my screen sharing is this visible.

2
Cindy Morgan 00:03:36.124 --> 00:03:36.934
It is.

3
Daniel Schien 00:03:37.144 --> 00:03:41.794
Fantastic. Okay great. Then I'm all set up and I'll stop sharing again.

4
Colin Perkins 00:04:35.135 --> 00:04:40.565
Okay, yeah, good morning. Everyone just the sound check. Good morning. What?
Whether it's a mistake.

5
Cindy Morgan 00:04:42.365 --> 00:04:43.445
Sounds good.

6
Colin Perkins 00:04:43.835 --> 00:04:51.275
Okay, let's see if I can look the screen share.

7
Cindy Morgan 00:05:10.174 --> 00:05:11.494
And we can see your slides.

8
Colin Perkins 00:05:11.704 --> 00:05:17.884
Okay, uh, I put them on the screen where I can't see them, but I guess, uh, if
you can see them those progress.

9
Eve Schooler 00:05:22.414 --> 00:05:28.264
And that's exactly what happened to me calling the other day. I, it was very
perplexing. I was like, where have my slides go.

10
Eve Schooler 00:05:31.924 --> 00:05:38.374
And also, just all sorts of Webex related pop ups. I was like, something has
taken over my machine.

11
Colin Perkins 00:05:39.664 --> 00:05:45.724
And see, where is columns, we can complain about Webex. We frequently enough.

12
Colin Perkins 00:05:52.924 --> 00:05:56.314
Pop up, that tells me that they shouldn't have more than 1 screen. So that's
helpful.

13
Colin Perkins 00:06:22.294 --> 00:06:31.744
It gives people a few minutes to join. Um, we have the slides from, um,
Michael, the other slides.

14
Colin Perkins 00:06:33.574 --> 00:06:38.974
People are presenting, can send slides to, I guess, Gary and myself, we'll try
and put them on.

15
Colin Perkins 00:09:29.015 --> 00:09:45.215
All right, so the, uh, which people are, joining seems to have slowed down. So
I guess we should get started. Um, hopefully you can all hear me fine. Uh, this
is the, uh, the 2nd session and the Internet architecture board's workshop.

16
Colin Perkins 00:09:45.815 --> 00:09:59.165
The environmental impacts of Internet applications and systems, uh, and I'm
Colin Perkins and 1 of the, the, the group who's, uh, organizing this workshop
with Gary and, uh, both of us.

17
Colin Perkins 00:10:01.235 --> 00:10:15.425
Um, so the, the 2nd session isn't a total of what do we know and the goal I
think today is to try and understand, um, what we know about the environmental
impact. Um, and to some extent, also what we don't know from what are the open
questions.

18
Colin Perkins 00:10:19.955 --> 00:10:38.585
So, to start off today, I just want to recap the, the ground rules and the,
this is the same rules that that Gary puts up on Monday a reminder that we are
recording the session and the recording will be published. Uh, we'll go up on
YouTube after the meeting.

19
Colin Perkins 00:10:39.154 --> 00:10:59.014
Uh, the position papers that you submitted a public and are on the website um,
and, uh, this is a professional meeting, and we expect, uh, people to to abide
by the usual professional standards of behavior. And that's, uh, of course any,
any kind of harassment is is not accepted in this, this meeting.

20
Colin Perkins 00:11:01.055 --> 00:11:20.855
Please try to be polite. Please try to explain your viewpoints in an
understandable manner. Um, please try to learn from other people's viewpoints.
Uh, we've got quite a diverse group of people here. Um, people with a quite a
wide range of backgrounds. Um, some people are from the academic community,
some from industry um.

21
Colin Perkins 00:11:21.514 --> 00:11:42.274
Some who have been heavily involved in the for many years. Some who haven't,
um, some from a, I think, perhaps a wider range of backgrounds than that. Um,
there's a risk that we talk past each other. Um, please do, um, you know, as I
say, take time to explain yourself and, uh, if there are, uh.

22
Colin Perkins 00:11:42.279 --> 00:11:48.874
Patients and misunderstandings, except the drivers are coming from perhaps
different places and we're all here to look.

23
Colin Perkins 00:11:51.034 --> 00:12:06.724
Uh, the URL on the slide has, uh, points us to the, uh, the, the materials, um,
the, um, the slides, uh, will be going up on the, the data tracker as we have
them. Uh, and I'm, I'm sure someone will put a link in the, in the chat.

24
Colin Perkins 00:12:10.564 --> 00:12:29.944
So the agenda for the workshop, we've got 4 sessions uh, this is the 2nd
session. Um, uh, on Monday. We, we looked at the big picture. Um, what we're
focusing on today is what we know. So, trying to to learn what we understand
and what we don't understand about the environment.

25
Colin Perkins 00:12:30.334 --> 00:12:51.364
Mental impact of the network, uh, we're trying to understand what what data we
have or can readily get, whether we understand that data, whether we need to
more measurements more research. And what are the methodologies are sufficient
and where are the gaps? Where are the things we don't know.

26
Colin Perkins 00:12:51.395 --> 00:13:11.645
We're here to try and think about who we might need to talk to, who might need
to connect with that. We're not already talking to, to get access to
information that's needed. Um, and if there are areas where we, we don't
necessarily have the expertise, we need, um, W, where we need to, uh, as I say,
form connections and try and enhance our understand.

27
Colin Perkins 00:13:13.534 --> 00:13:29.464
This will feed into the session tomorrow. Same time tomorrow, which is looking
at, uh, protocol, improvements, implementation, improvements and concrete
actions we can be taking and the session on Monday next week, which is, uh,
starting to look at the future developments and the plans. And what are the
next steps.

28
Colin Perkins 00:13:34.205 --> 00:13:47.975
The, the goals for today, as I say, are to understand what we we do and do not
understand about the situation. Um, and I, I guess the phrase known unknowns
and knowns and so, and we'll all come into this discussion at some point.

29
Colin Perkins 00:13:49.474 --> 00:14:10.324
Uh, 11 thing I did want to highlight, um, from talk on, on Monday, um, there
was a suggestion to, um, put efforts into the actual actions. Um, you know,
it's easy to talk about what, what data we would like to know what measurements
we would like to take. Um, and it's easy to continue.

30
Colin Perkins 00:14:10.354 --> 00:14:25.624
Really focus on measuring and understanding and improving that understanding
and that's certainly interesting, but I would encourage people to focus on, not
necessarily what we would like to know, but what we need to know in order to
make a difference.

31
Colin Perkins 00:14:27.334 --> 00:14:33.364
Right. Let's let's focus on how to achieve concrete action was not just how to
improve our understanding of the situation.

32
Colin Perkins 00:14:37.534 --> 00:14:56.254
Uh, the agenda for today, um, we've got 3, uh, short talks to try and set the
scene. Um, Michael, we'll, we'll begin by talking about, uh, what, what we know
about energy usage. And, uh, if there is any misinformation in the space, um.

33
Colin Perkins 00:14:58.114 --> 00:15:18.004
Get James and Nina, uh, I think it's Jones will be, uh, apologies for
mispronouncing that, uh, we'll be talking about, um, network, uh, energy
consumption and trends and then Daniel will be talking about the role of peak
demand. And the goal here is to give you some information highlights some open
questions and.

34
Colin Perkins 00:15:18.065 --> 00:15:39.125
Uh, the, uh, we should then have an hour or so for open discussion after that.
Um, and I've, I've got Q slides to structure that discussion. The focus of the
discussion will be, um, on what we know. Uh, what are the known unknowns? What
are the things we know we don't know and need to find more into.

35
Colin Perkins 00:15:39.159 --> 00:15:40.924
In order to have an impact here.

36
Colin Perkins 00:15:42.185 --> 00:15:54.965
What are the risks and concerns and who should we be talking to? What sort of
outreach do we need in order to complete our understanding? And in order to
help us understand what what we know.

37
Colin Perkins 00:15:58.685 --> 00:16:07.055
Uh, with that, uh, I think that's all I have, uh, are there any questions, uh,
about the agenda? Um, before we move on.

38
Colin Perkins 00:16:22.144 --> 00:16:29.494
And I didn't see any questions. So, the 1st book is, uh, can you show you the
slides.

39
Michael Welzl 00:16:31.084 --> 00:16:40.594
Yes, let's see sharing PowerPoint does that work?

40
Colin Perkins 00:16:40.654 --> 00:16:41.044
Yep.

41
Michael Welzl 00:16:42.664 --> 00:16:43.714
Okay, I, uh.

42
Michael Welzl 00:16:45.634 --> 00:16:50.134
Content, blah, blah, blah so, is that you're seeing the full screen or.

43
Colin Perkins 00:16:52.144 --> 00:16:53.044
Seeing the full screen.

44
Michael Welzl 00:16:53.374 --> 00:16:57.724
Okay. All right. Thank you. Um, so this is about the Internet.

45
Michael Welzl 00:17:00.814 --> 00:17:20.673
Um, uh, which is supposedly a bit smaller, but maybe well, depends and a little
bit about misinformation. People have, uh, already done some spoiling actually
on the mailing list. Regarding some things I was gonna talk about. Uh.

46
Michael Welzl 00:17:20.854 --> 00:17:41.824
Here I'm doing my making an effort to give a very, very precise overview of
magnitude relationships. Exactly. Like that. You are that tiny spot your own
footprint this could be a city. This could be a country. This could be the
Internet and that's, uh, the Internet.

47
Michael Welzl 00:17:41.884 --> 00:18:02.014
All of, depending on stupid claims, now we're going to get back to these
relationships. But actually, the point is that these, uh, maybe not completely
off. Um, we'll see. So, I'll begin with nonsense just to get it done to get it
out of the way.

48
Michael Welzl 00:18:03.065 --> 00:18:23.525
Uh, there are some examples of nonsense, and I also did, like this email thing
that, uh, people started already writing about, in, uh, in emails. Um, so,
Mike, honestly, the brother of Tim bersly, he has been talking to the press.
Apparently about this, uh, issue quite a bit.

49
Michael Welzl 00:18:24.304 --> 00:18:38.464
And, um, I just came across these press statements some time ago. Actually,
from a presentation I gave for students were, I was trying to make the point
that there was just lots and lots and lots of nonsense and depressed that you
don't need to believe.

50
Michael Welzl 00:18:39.845 --> 00:18:52.385
Giving very concrete precise numbers about an email, um, that to begin with is
maybe idiotic. Just as an idea I would say. Um, but.

51
Michael Welzl 00:18:53.740 --> 00:19:11.045
It ranges from 0.3, Grand 24 gram and then, you know, some other message it
says from 0.03 to 26 and so forth. And then much later, uh, there is a press
article. Um, microphone is Lee himself actually told the Financial Times that.

52
Michael Welzl 00:19:12.069 --> 00:19:33.124
Well, this is kind of a rolling back thing. It seems that, uh, he must have
gotten responses to these, uh, things that he, that that went out to the press,
because he then said that this is back of the, and the envelope maths and maybe
it's useful to start conversations. But there are really bigger questions. I
guess the main point of this slide is that there are many URLs here.

53
Michael Welzl 00:19:33.364 --> 00:19:54.364
All these statements coming from various sources, and it goes on here, it says
typical office worker sends and receive so many emails per day. You know,
what's a typical number anyway here? But over the course of the year, this is
as much as flying from London to Bruce, where I don't even think there's a
direct flight to begin with. But, uh, it's watching 955.

54
Michael Welzl 00:19:54.395 --> 00:20:15.485
Movies all that kind of stuff um, really simply put um, you can try and
calculate these things and compare them but trust me I did actually try and
there's just nothing here that makes any sense at all all these numbers don't
match. So, I'll just nonsense. Um, there's also this.

55
Michael Welzl 00:20:15.634 --> 00:20:36.634
Saying that climate change is going to make us more stupid. I just leave that
here without say more about it. Um, the reason for me to talk about this is
that I do agree with also channels and call me who has said that this is a
harmful thing. People go out saying these things to depressed.

56
Michael Welzl 00:20:36.695 --> 00:20:57.785
Get spread and it may actually have a negative effect for the cost that we are
having here, which is to, uh, consider the, the environmental impact of the
Internet. It may cost some people not to take this whole, this, whole kind of
effort serious. And, uh, especially when these people work at funding.

57
Michael Welzl 00:20:58.024 --> 00:21:14.314
Is done, this is not a good thing. We can also consider bundling this law here,
which is also called the bullshit Estimator principle. Basically the fact that
it takes an order of magnitude more effort to refute bullshit and to put it out.

58
Michael Welzl 00:21:16.174 --> 00:21:35.914
Um, Jonathan qormi has done a good job at, uh, pointing at these things. So I
put some links here to these keynote slides that have many examples. And this
does not compute the article and also the separating fact from fiction article
that was sent to the workshop mailing list in an attachment, just maybe a
quarter before this.

59
Michael Welzl 00:21:37.324 --> 00:21:39.514
Uh, I think it's interesting reading. Indeed.

60
Michael Welzl 00:21:41.529 --> 00:22:02.644
Done with this already. Let's look at reality. Uh, my take is that, um, if we
can come with some conservative estimates that are possibly reasonably
realistic, and possibly go out to depressed with, uh, with them. Then, uh, we
may be able to, you know, counter these things a little bit. And maybe this is
a better thing than having these.

61
Michael Welzl 00:22:02.679 --> 00:22:22.924
Crazy exaggerations. Now, what are these estimates? Um, I have made a start
with this table that our submission has. And, uh, what I did is, I didn't
really try to well, try to gather data myself but, uh, just looked at reports
that others have made.

62
Michael Welzl 00:22:23.974 --> 00:22:44.614
Now, when you look at reports, actually, um, you know, we have already
discussed this, I think, to some degree in the 1st meeting that, uh, there is
no, even no notion of the Internet. What is the Internet you know, it
considers, uh, it consists of, of devices. The Wired part, the wireless part
devices or customer premises equipment, user equipment.

63
Michael Welzl 00:22:45.215 --> 00:23:05.585
There is the embodied energy of devices, which is from constructing the thing
that we're using. Uh, there are data centers, which, you know, may maybe, uh,
well, yeah, things made it for quite a bit. And, uh, the study is also
different age. So, what I did is, I.

64
Louise Krug 00:23:05.585 --> 00:23:05.885
Just.

65
Michael Welzl 00:23:05.885 --> 00:23:06.125
Try to.

66
Louise Krug 00:23:06.154 --> 00:23:06.334
It's.

67
Michael Welzl 00:23:06.754 --> 00:23:24.004
Find a couple of sources, and from these sources I arrived at, the range of
0.5% to 1.17% of the goal was your 2 footprint. Now my point here is not that
this is the number. My point is that it's probably not.

68
Louise Krug 00:23:24.004 --> 00:23:24.994
Percent is probably not.

69
Michael Welzl 00:23:24.994 --> 00:23:26.674
0.001%.

70
Michael Welzl 00:23:27.304 --> 00:23:48.424
But it's probably something in that rough magnitude 1, possible revision just
to make this a bit concrete, is that, uh, you can take the smarter 20, 30
report. That's been cited a lot which states that, uh, in general has a
footprint of 2.7% of global emissions in 2020 and, uh, there are numbers from
2012.

71
Michael Welzl 00:23:48.484 --> 00:23:52.204
Say, telecom electricity is I see there's a 3rd of.

72
Michael Welzl 00:23:52.804 --> 00:23:58.984
No, this may have changed chances on it. Now it's bigger, but, uh, if now it's
a 3rd and.

73
Louise Krug 00:23:59.014 --> 00:23:59.254
We.

74
Michael Welzl 00:23:59.254 --> 00:24:13.054
Have something like 0.9% so, you know, it's probably something in that rough
magnitude. Um, is this a small number? 0.9% doesn't exactly sound very big,
but, uh.

75
Michael Welzl 00:24:14.314 --> 00:24:34.954
To begin with the Internet energy user has been compared against the aviation
industry a lot. I think that is a bad idea, because the aviation industry is
somewhat hard to talk about cause, uh, that number of 0.9% is actually
comparable to the emissions of deviation industry. But then.

76
Michael Welzl 00:24:34.984 --> 00:24:56.104
Aviation, we also have, uh, and other greenhouse gases playing rules. If you
take those into account, then the number of deviation industry is significantly
higher. And then I've recently read, uh, something that, that apparently quite
recent study that finds that, uh, the influence of water vapor is actually
quite significant and there's water vapor.

77
Michael Welzl 00:24:56.135 --> 00:25:17.255
There in the atmosphere stays long and contributes to heating up the planet.
And if you consider this, then this is way above I think, 3%. So it's hard to
talk about. And, uh, I think can lead to confusions instead of that. What? I
did the other day. Was to take a look at, uh, this, our world and data on a
webpage that gives.

78
Michael Welzl 00:25:17.260 --> 00:25:38.405
An overview of emissions per country, which again, you know, is very broad
because what does it mean that country I mean, each country runs Internet
devices right? So even that conversation gets hard. But, uh, anyway, you know,
to get an idea of the of the magnitude, um, according to this page, the 2221
emissions of.

79
Michael Welzl 00:25:38.434 --> 00:25:59.554
The UK was 0.93%, so that's a comparable number Norway where I live, uh, has
0.11 here in Norway as a strong push to, for everybody to, uh, reduce the
personal footprint. There. Are people going to great lengths to do this as a
public press offering calculated.

80
Michael Welzl 00:25:59.584 --> 00:26:20.524
So you can see what is my personal footprint. Of course, it's a good idea to
change the perception of people and make them aware of it and, uh, make more
people, join this. But I think it's important to put things into perspective a
little bit. Population here is 5.4Million. So, roughly the per person
contribution is 0.

81
Michael Welzl 00:26:20.944 --> 00:26:41.704
Manage 02%, it's tiny now, this these are the people here. Um, it's also a
small fraction of the world population, but it is interesting to consider that
if we could with the standard, somehow reduce the internet's power policy but
10%. I mean, even if you think of 1%, right? Maybe that's more realistic.

82
Michael Welzl 00:26:42.155 --> 00:27:01.655
Even then this is this is a big number compared to, uh, you know, a couple of
couple of people now, 10%. That will be, you know, again, going from 0.9% or
0.81, which is 4.5Million regions. And, uh, you can imagine them all, you know,
wearing Viking helmets. And things so many.

83
Michael Welzl 00:27:03.009 --> 00:27:24.154
Okay, I'm stopping here. It does get a bit silly, but the point is that, uh,
the potential of a standard that has an impact, the global impact on the
Internet is truly enormous and it's gonna be enormous. You know, whether this
is 1% 2%, 3% and a half percent. It's very big getting back to the picture from
the beginning.

84
Michael Welzl 00:27:24.274 --> 00:27:34.024
Uh, I don't think we're still wrong with this country and Internet comparison
also, because we have these rings, which means that it's hard to understand
what the size really is.

85
Michael Welzl 00:27:35.464 --> 00:27:51.364
And, uh, you know, snapped on, I don't know, these could be a potential savings
from standards, who knows, you know, time bit smaller than that. But, you know,
it's about this general general, very broad relationships. That's already
everything. I have to say.

86
Colin Perkins 00:27:56.554 --> 00:28:00.424
All right, thank you. My call. Um, I'm curious how many no weekends are there.

87
Colin Perkins 00:28:04.985 --> 00:28:06.965
Is it is fair and a half a 1Million all of them.

88
Michael Welzl 00:28:07.475 --> 00:28:15.065
Uh, all together I thought in in the meeting, uh, no, no, it's a bit more. I
think it's 5 and a half 5 something.

89
Colin Perkins 00:28:15.905 --> 00:28:16.295
Okay.

90
Colin Perkins 00:28:18.935 --> 00:28:19.085
The.

91
Michael Welzl 00:28:19.085 --> 00:28:20.135
Almost the whole country.

92
Colin Perkins 00:28:20.615 --> 00:28:24.335
It's the whole country. All right. Does anyone have any sensible questions for
Michael?

93
Colin Perkins 00:28:30.635 --> 00:28:34.535
Uh, so those, uh, okay, so.

94
Colin Perkins 00:28:34.774 --> 00:28:45.604
So, custom and Eve have questions in the chat so if if says other particular
standard standards, we should scrutinize these for these savings.

95
Colin Perkins 00:28:51.904 --> 00:28:53.164
I I guess that's possible.

96
Colin Perkins 00:28:54.484 --> 00:28:55.024
But, uh.

97
Eve Schooler 00:28:55.624 --> 00:28:56.344
It's a loaded.

98
Michael Welzl 00:28:56.344 --> 00:28:58.894
Exactly, that's a long it's a large question.

99
Eve Schooler 00:28:59.074 --> 00:29:15.364
But I, but I, but I put it out there mostly to figure out if in your mind you
had certain ones that you, um, feel, uh, you know, could use improvement or
certain techniques that, you know, are well, that are used widely that.

100
Eve Schooler 00:29:15.994 --> 00:29:22.054
You know, we should be thinking about extending or changing, um, places to
start.

101
Michael Welzl 00:29:22.534 --> 00:29:36.574
Yeah, I'm a transport layer person so I've been thinking about, uh, transport
layer protocols that send lots of X very simply, you know, if you could just
imagine reducing. I mean, now we usually send approximately.

102
Michael Welzl 00:29:37.475 --> 00:29:57.815
For every other packet, so I've been, I've been just wondering if we would take
this down to an, for every quarter. I mean, every 4th packet, you know, would
that have a significant input influence on the whole Internet? That's a silly
way to do it, but we could be a bit more intelligent about how many acts we
sent. We could. I mean, I'm interested in multi costs.

103
Michael Welzl 00:29:57.904 --> 00:30:15.064
It's very interesting this paper that wasn't there. I think you're on paper had
interesting suggestions about routing along. Uh, well, not just energy saving,
but, uh, depending on carbon efficiency. So, you know, I think there are plenty
of things that'd be good to.

104
Colin Perkins 00:30:16.744 --> 00:30:29.974
Yep, okay I mean, I think we have time allocated tomorrow to talk about, um,
things we could do and the focus here is perhaps more on understanding what we
know and how to how to measure the impact. Uh, last is that you have that.

105
Lars Eggert 00:30:30.904 --> 00:30:32.734
Yeah, and my head the screen, which is.

106
Colin Perkins 00:30:32.944 --> 00:30:33.334
It is.

107
Lars Eggert 00:30:34.509 --> 00:30:55.264
Webex client, or I, I just don't think different. It's not yellow anymore. It's
confusing. Me. Hell. Anyway. So, 1 sort of piece of feedback. Maybe. I mean,
you, you showed these nonsense, uh, metrics or measurements where that were
reported and, like, many others that I've seen. They all they, they throw
together the bundle up, like the networking stuff and the application.

108
Lars Eggert 00:30:55.684 --> 00:31:16.804
And they say, you know, it's per email, right? And I actually find that a bit
confusing right? And it sort of models the waters. Because if you look at the
data center, and you look at your the stuff that burns energy, right? It's the
stuff that gets hot and that's usually the, and the GPU is it's not so much the
network. Right? And so obviously the so well, not obviously, but I, I assume
that.

109
Lars Eggert 00:31:16.834 --> 00:31:37.834
Means that the application processing part is actually probably contributing
more to the power usage, compared to the communication, which is as far as I
know pretty efficient, but it's getting more efficient since the switch ships
are getting better. And so I wonder if there's anybody who sort of tries to
tease that apart and tries to talk about.

110
Lars Eggert 00:31:38.734 --> 00:31:52.684
Um, power users here, 2 emissions, the different layers, if you will right at
the application layer. Um, and then for the communication part and further
aspects or is it all all reported as, you know, this big lump thing?

111
Michael Welzl 00:31:55.355 --> 00:32:15.995
Okay, uh, well, since I, I don't know if I'm the 1 supposed to answer that, and
this is the broad command for everyone. That's false. But, I mean, uh, this, of
course, obviously, this nonsense studies have lumped everything together. Um,
there are ways to get CPU and memory with our. I was wondering if there are
APIs.

112
Michael Welzl 00:32:16.054 --> 00:32:24.364
That may give something like that in software because that would be actually,
especially interesting for, for I don't know of any.

113
Colin Perkins 00:32:37.085 --> 00:32:43.055
The mute button, and this may be 1 for what we don't know discussion later is
what's the breakdown between different aspects of the system?

114
Suresh Krishnan 00:32:45.274 --> 00:32:48.334
Would you mind if I just give a quick answer to Michael?

115
Colin Perkins 00:32:48.574 --> 00:32:49.654
Of course, jumped in.

116
Suresh Krishnan 00:32:50.014 --> 00:33:05.374
Uh, so, uh, 1 of the things I did in earlier life Michael, right? Like, what we
did is, like, there's no real stuff coming from the chip itself, but on on the
switch box or a router box, uh, we had, uh, some kind of, right? Like, from the
data center, like, so what we can do is we can go to the rabbit and and.

117
Suresh Krishnan 00:33:05.555 --> 00:33:15.635
How much power it consumes, right? So then you have to subtract the steady
state power to get that thing. So it is possible. And and I have at least done
something like that. I can probably share with you, but it's not public.

118
Michael Welzl 00:33:17.525 --> 00:33:17.945
Thanks.

119
Colin Perkins 00:33:19.295 --> 00:33:23.585
Yeah, that sounds useful. Uh, who's that? Bruce?

120
Bruce Nordman2 00:33:24.305 --> 00:33:24.965
Yeah.

121
Bruce Nordman2 00:33:25.534 --> 00:33:46.264
Noting is that all network equipment is, on the order of 110th of all
electronics energies, all electronics are not Internet connected, but
increasing fraction is suggest that so worth a point where it's taking and
electronics are about 14%. Well, this was from 8 years ago of all buildings,
energy.

122
Bruce Nordman2 00:33:47.015 --> 00:34:03.515
So, there's rough order of magnitude from network of women to electronics, and
roughly an order magnitude from electronics to all electricity and buildings
just for context. And I've, I've been doing this for since the late 90 s,
energies of electronics.

123
Colin Perkins 00:34:05.525 --> 00:34:07.415
Um, uh, other breakdown.

124
Colin Perkins 00:34:07.475 --> 00:34:17.195
Uh, uh, people aware of breakdowns of how the different components in data
sensors use the energy for example. So I'm guessing there, there's some of the
big, biggest consumers.

125
Bruce Nordman2 00:34:18.215 --> 00:34:28.415
Data centers are about 10% of all electronics as well. So the, uh, electronics
not in data centers is about 10 times as much as.

126
Bruce Nordman2 00:34:28.569 --> 00:34:29.914
Electronics and data centers.

127
Colin Perkins 00:34:30.844 --> 00:34:31.233
Okay.

128
Bruce Nordman2 00:34:31.564 --> 00:34:32.493
For context.

129
Colin Perkins 00:34:33.454 --> 00:34:34.804
Okay, interesting.

130
Colin Perkins 00:34:37.263 --> 00:34:45.304
All right, um, we, we actually have more time for discussion later. Does anyone
have anything else for Michael before we move on to the next, uh, presentation?

131
Pernilla Bergmark Ericsson 00:34:48.964 --> 00:35:08.884
Yeah, it was you said something and thank you for your presentation around the,
um, opportunity of, of less frequent signaling let's say, and of course the
experience from from the 5 g development where we're signaling has been
significantly less frequent.

132
Pernilla Bergmark Ericsson 00:35:09.604 --> 00:35:30.604
What it enables in terms of sleep mode is something that is worth considering I
also think from for other systems and so on. Uh, and then also, in terms of
standardization you mentioned the, uh, so I'm academia. Um, WH, WH, what kind
of.

133
Pernilla Bergmark Ericsson 00:35:30.635 --> 00:35:37.475
We would need so, I think in terms of specification, how to calculate the.

134
Pernilla Bergmark Ericsson 00:35:37.804 --> 00:35:58.804
Print that is a standard in place, which is quite detailed but then, of course,
some studies look at some look at, uh, digital technologies, which is not so
well, defined some look at Internet. And what how do we define Internet? Is it
from an internal perspective? Is it from.

135
Pernilla Bergmark Ericsson 00:35:58.834 --> 00:36:09.574
This sort of core and what what would be the system boundaries and that I think
we have, uh, much less in terms of standards in relation to assessments. So, I
don't know if you agree with that.

136
Michael Welzl 00:36:12.484 --> 00:36:18.244
Uh, agree with the fact that we don't have so many standards defining the
Internet.

137
Pernilla Bergmark Ericsson 00:36:18.664 --> 00:36:19.504
Yeah, I mean.

138
Michael Welzl 00:36:19.504 --> 00:36:20.014
In.

139
Pernilla Bergmark Ericsson 00:36:21.754 --> 00:36:26.134
In relation to the assessment specifically, or if you are aware of the thing
that I'm.

140
Michael Welzl 00:36:26.224 --> 00:36:27.304
I'm I'm not I'm not.

141
Pernilla Bergmark Ericsson 00:36:29.375 --> 00:36:30.395
No, okay. Thank you.

142
Colin Perkins 00:36:33.964 --> 00:36:44.854
All right, uh, I think you have all the questions I think we should probably
move on to the next, uh, presentation. Uh, which I think is, is it, uh, am I
pronouncing your name correctly?

143
Jens Malmodin 00:36:45.544 --> 00:36:47.344
Yes, do you hear me.

144
Colin Perkins 00:36:47.614 --> 00:36:48.604
Yes, we can hear you.

145
Jens Malmodin 00:36:48.904 --> 00:36:53.794
Great so, let me try to share my presentation Share application.

146
Jens Malmodin 00:37:02.404 --> 00:37:03.394
That seems to be.

147
Jens Malmodin 00:37:08.884 --> 00:37:11.794
Window.

148
Jens Malmodin 00:37:15.935 --> 00:37:16.655
Do you see anything.

149
Colin Perkins 00:37:17.794 --> 00:37:18.484
Yet.

150
Jens Malmodin 00:37:27.604 --> 00:37:29.164
I'm not share the screen.

151
Jens Malmodin 00:37:41.494 --> 00:37:42.424
You don't see anything.

152
Colin Perkins 00:37:43.234 --> 00:37:45.364
Uh, I, I, I don't see anything there.

153
Jens Malmodin 00:37:47.435 --> 00:37:50.195
Call.

154
Jens Malmodin 00:37:52.205 --> 00:37:53.105
You see the screen.

155
Colin Perkins 00:37:55.864 --> 00:37:59.554
I am not seeing you sharing anything, but maybe that's just me.

156
Pernilla Bergmark Ericsson 00:38:00.934 --> 00:38:01.294
No.

157
Jari Arkko 00:38:01.354 --> 00:38:01.744
In here.

158
Jens Malmodin 00:38:16.624 --> 00:38:16.894
Yes.

159
Jens Malmodin 00:38:27.965 --> 00:38:31.415
Seems not working Urs sent the presentation to, you.

160
Daniel Schien 00:38:33.904 --> 00:38:37.024
Yes, I can also we can also swap the.

161
Jens Malmodin 00:38:37.954 --> 00:38:39.754
Yeah, Daniel.

162
Daniel Schien 00:38:40.084 --> 00:38:41.224
Okay, no worries.

163
Jens Malmodin 00:38:41.284 --> 00:38:42.454
I don't know if.

164
Daniel Schien 00:38:42.724 --> 00:38:48.334
And then you can you have that you have a moment to sort out your slides and
then.

165
Daniel Schien 00:38:50.495 --> 00:38:52.775
And then you can present from your own screen. Okay.

166
Colin Perkins 00:38:53.705 --> 00:38:57.275
If not send the slides to Gary and me and we will, we can.

167
Pernilla Bergmark Ericsson 00:38:58.205 --> 00:39:01.265
Yes, in case you are using double screens that my.

168
Jens Malmodin 00:39:03.574 --> 00:39:05.854
Yeah, yeah, that's an idea. Yeah.

169
Daniel Schien 00:39:06.574 --> 00:39:09.454
Can you, can you see the slides.

170
Colin Perkins 00:39:09.604 --> 00:39:11.254
Yes, I can see.

171
Daniel Schien 00:39:11.314 --> 00:39:13.984
When you see the presenter view or the, the menu.

172
Colin Perkins 00:39:14.224 --> 00:39:15.274
No, we got the main view.

173
Daniel Schien 00:39:15.604 --> 00:39:27.484
Fantastic. Okay. Thank you. Um, so my name is Daniel sheen. I work at the
University of Bristol in the UK and present work that our group here.

174
Daniel Schien 00:39:27.965 --> 00:39:48.635
Has carried out over the last 10 years or so, um, including professor Chris
priest, who I believe may, or may not be in the audience and Paul, and I've
seen his name in the audience, but we work with many of you here. So this is
it's really fantastic to meet.

175
Daniel Schien 00:39:48.755 --> 00:40:09.785
To meet the people we already work with and meet so many new faces and names
and the email thread that we had going between the participants is already
showing that this is a fantastic forum. There's an enormous amount of knowledge
that has been exchanged already. And we look.

176
Daniel Schien 00:40:09.789 --> 00:40:30.784
Forward to hearing what you have to say critically about the ideas that we're
sharing. So, in this talk, I am coming from the perspective of understanding
the environmental footprint for digital services for media services
specifically and.

177
Daniel Schien 00:40:30.994 --> 00:40:47.434
And we will see that this offers a slightly different lens to thinking about
the carbon footprint of and of networks specifically. And, and I'll be talking
about the state of the arts. So to say, and then we're proposing.

178
Daniel Schien 00:40:48.065 --> 00:41:02.195
Uh, in addition to the current methods for estimating the carbon footprint and
and in particular to that, I look forward to to a discussion with you. Okay.

179
Daniel Schien 00:41:03.905 --> 00:41:04.325
Um.

180
Daniel Schien 00:41:09.154 --> 00:41:26.764
So, what I, what I often nowadays tend to start with is showing this picture
here, which categorizes the way that in general is related to environmental
impact. It's a slightly busy, um.

181
Daniel Schien 00:41:27.454 --> 00:41:48.454
But it really helps to focus the conversation to provide some terminology. What
we see here are 2 columns with the red column describing all of the ways in,
which has a negative effect on the environmental impact and in the green, the
positive effect and.

182
Daniel Schien 00:41:48.515 --> 00:42:02.375
And and then horizontally, there are 3 layers. The 1st, is the direct layer and
this is exclusively associated with what we call the direct impact of.

183
Daniel Schien 00:42:02.644 --> 00:42:23.494
Um, meaning these are activities in the economy where, where there's a real
release of greenhouse gases into the atmosphere that is directly associated and
measurable with the use of. So, for example, from the production of Silicon and.

184
Daniel Schien 00:42:23.765 --> 00:42:44.825
Assembly from transport, and then the use case where we, where we consume
electricity that is based on fossil fuel combustion, and then the disposal
phase and this is, this is what our work in Bristol has mainly focused on, but
increasingly.

185
Daniel Schien 00:42:44.915 --> 00:43:05.915
Um, we're also looking at, um, positive effects and how the tool may or may not
balance. Um, but but I think it's useful to know that the, the terminology and
I think in this in this workshop here so far, we're really focusing on the
direct impact through those lifecycle.

186
Daniel Schien 00:43:06.994 --> 00:43:27.184
And the electricity consumption is, uh, mainly relevant as part of the use
phase, although it is really, really important more so than ever to think about
the life cycle phases that proceed the use face and what we often called the
embodied impact. Okay. So this.

187
Daniel Schien 00:43:27.189 --> 00:43:48.334
It's just some terminology in Bristol, I already said we're specifically
focused on understanding digital services in 2011. we started to work with the
Guardian newspaper who at that time, found themselves at a point of transition
from paper to digital and they had a very good.

188
Daniel Schien 00:43:48.339 --> 00:44:09.004
Understanding of the environmental impact of paper, and we work with them to
understand how that compares to reading the news online. And and then 2015 we
worked with the BBC, um, another media service trying to help them to.

189
Daniel Schien 00:44:09.515 --> 00:44:30.185
Understand how a move away from terrestrial broadcast, um, to video on demand
would affect the net environmental impact. We are computer scientists, but we,
we are basing our work on on an environmental assessment methodology called
lifecycle assessment.

190
Daniel Schien 00:44:30.665 --> 00:44:35.285
And 1 of the key components here is to take into account.

191
Daniel Schien 00:44:36.484 --> 00:44:57.394
All of the lifecycle phases now with the caveat that, um, there's a, there's a
strong focus on usefulness, electricity consumption for various reasons. But in
terms of the system that we want to take into account, um, we, we try to be
inclusive and.

192
Daniel Schien 00:44:57.430 --> 00:45:18.365
Data centers, networks, wired and cellular networks, and then the user devices,
and that's important in order to prevent burden shifting, for example, and to
provide a complete and consistent understanding of the environmental impacts of
media services. Great. Here's an example of of the.

193
Daniel Schien 00:45:18.814 --> 00:45:39.424
Of the process, such an assessment takes into account all of the, all of the
processes that are involved and there can be many and such an assessment
usually starts with with creating such a map. And then for each of the boxes
identify how the underlying infrastructure.

194
Daniel Schien 00:45:39.814 --> 00:46:00.874
Um, consumes electricity in proportion to the service and then there are some
technical steps that that are required when processes are consuming
electricity, or, uh, or other environmental flow, environmentally relevant flow
for, for more than.

195
Daniel Schien 00:46:00.880 --> 00:46:21.845
In 1 service, and then suddenly we need to start talking about allocation and
that's a really important topic that this that I'm going to spend more time
talking about in the moment. Okay. We also work with an organization called
capstone to construct a tool.

196
Daniel Schien 00:46:22.144 --> 00:46:42.874
Um, that is more self service so the work with the BBC, and the Guardian
earlier was very bespoke and now we're living in an age where all media
organizations are, all organizations in the economy are concerned about their
carbon footprint and there's a big need to, to carry out baseline assessments.

197
Daniel Schien 00:46:43.445 --> 00:46:59.885
And then set top for decarbonization, and for the media sector, the impact tool
is, is probably the most popular at the moment with many organizations, many
more than are listed here using this tool to, to calculate the carbon emissions
for their services.

198
Daniel Schien 00:47:01.744 --> 00:47:22.654
Um, there has been some exchange of emails already around the carbon footprint
of streaming. I believe. I'm, I'm very convinced that at the moment, the most
robust that is out there is the 1 that is based on the impact model published
in 2000.

199
Daniel Schien 00:47:22.684 --> 00:47:43.774
21 by the common trust and against, for example, he was informing this as well
which sorry I moved my, my gallery here, which found that for for a large
global video.

200
Daniel Schien 00:47:43.834 --> 00:47:57.094
Demand streaming service, the average footprint, uh, per hour, um, for the
entire audience with a mix of different viewing devices is in the order of 55
grams per hour.

201
Daniel Schien 00:48:00.904 --> 00:48:20.464
Excuse me um, so this is a, this is a useful data point. Um, and it clears up
with sadly sensational values. Um, now we understand why, um, the.

202
Daniel Schien 00:48:20.674 --> 00:48:41.374
Picks up on dramatic values, but that has been pointed out. It's, it's
dangerous to go out with wrong estimates now. And the crucial element that is
worth spending more time talking about is the influence of decarbonization of
the electricity mix.

203
Daniel Schien 00:48:41.675 --> 00:49:02.435
Um, we can see here that very clearly, if we're just looking at the, uh, the
use phase, which is, um, the, the carbon intensity of electricity results in a
proportional variation of the overall carbon footprint. Of course.

204
Daniel Schien 00:49:03.034 --> 00:49:23.764
Um, the the decarbonization of the electricity will also results in a reduction
of the embodied impact. So, so this is a 1 key lever that we need to take into
account when we're thinking about initiatives to reduce the environmental
impact. Okay. So, um, so these are some results.

205
Daniel Schien 00:49:23.945 --> 00:49:45.065
Um, gives us, um, a perspective, um, this service based perspective, I think
it's useful for a variety of ways. Um, it is useful for the organizations,
providing those services to to think about footprint. So, um, so the users of
the impact of the, for example, they get this perspective for the.

206
Daniel Schien 00:49:45.069 --> 00:50:06.184
Entire audience for the entire, the entire service that they provide, and then
they can set targets, um, on these larger, uh, system boundaries. Um, but for
consumer, it's also relevant to put into perspective. What is the footprint of
a certain activity if you're thinking about your footprint? So here.

207
Daniel Schien 00:50:06.219 --> 00:50:27.034
For example, I'm in the UK, the carbon footprint per hour on average. So this
would be a use of maybe 70% televisions and 20% laptops and 10% other devices.
Um, we've got a carbon footprint of less than 50 grand per hour.

208
Daniel Schien 00:50:27.695 --> 00:50:48.515
Uh, in a, in, in, in the Guardian article, um, last year, I said, you need to
compare this to, um, to food, for example, and this is 1 hour of activity, and
the bowl of alone has a footprint of 100 grams of carbon. So, I'm not saying
that.

209
Daniel Schien 00:50:48.544 --> 00:50:52.144
Because it's small in the duration, maybe.

210
Daniel Schien 00:50:52.894 --> 00:51:13.894
It's not worth thinking about, but that scale is is important. And there was
Colin was making the point that we need to. We should be spending some efforts
in thinking about what data we need. The question of what data we need entirely
depends on what we want to.

211
Daniel Schien 00:51:13.924 --> 00:51:35.044
To do, and what we want to do is the question of what is the political will
maybe social political that is of the society. And so this is we need to take
that into account coming back to what we, as a community of engineers can do.

212
Daniel Schien 00:51:35.435 --> 00:51:56.195
I think the most important thing that we, the role that we have to play is
provide data to the debate to the consumers to the organizations, providing the
services to the legislators regulate, et cetera and that data depends on again.
We start with this understanding of what the infrastructure looks like. Um,
this is an old.

213
Daniel Schien 00:51:56.224 --> 00:52:17.344
But, in principle, I think it still is representative of how the networks work.
You've got an electronic routing layer, then you've got an optical transport
layer underneath and we need to take into account the efficiency of, of sending
data across the system. With this understanding. We can then produce estimates.

214
Daniel Schien 00:52:17.644 --> 00:52:37.984
Of, um, of energy intensity and sees this. I know you will think oh, this is
too simple. And, of course, um, I fully support that point, and I'll provide
something that is a bit more complicated and nuance in the moment. But that is
the state of the art, what we've got here and.

215
Daniel Schien 00:52:38.499 --> 00:52:59.044
Like that, um, I turned I V, here for intensity of data volume and multiplied
with an amount of data transported over network. We're producing a footprint.
We are getting a number.

216
Daniel Schien 00:52:59.764 --> 00:53:20.794
That is supposed to represent some sort of responsibility for for the energy
consumption, and hence carbon emission of the network associated to this data
volume. Um, so this is this is, um, the state of the art. Excuse me? What I've
got here is this.

217
Daniel Schien 00:53:21.005 --> 00:53:41.945
Um, exponentially decreasing, um, line at very consistently these, um, average
energy intensities for electricity have have have have been.

218
Daniel Schien 00:53:41.974 --> 00:54:03.094
Used over time, this is from the paper from 2018, which the current standard
for service carbon assessments, the greenhouse gas protocol sector guidance
points to. So, this is this is very relevant.

219
Daniel Schien 00:54:03.604 --> 00:54:13.324
And then on the 1, and then, on the other hand, we've got, we've got data that
has been broken down from from news corps QC here in the audience as well. So
they're.

220
Daniel Schien 00:54:14.855 --> 00:54:34.745
There are different approaches to these, the numbers are slightly varying
budget in, sort of, in the order of magnitude. There are comparable. Um, the
problem is that they are not very good to evaluate change. They're, they're, um.

221
Daniel Schien 00:54:35.674 --> 00:54:56.674
Uh, and and here is where again comes in, we 1st, start with the observation
that energy consumption of networking equipment in particular, but of all other
devices as well is has low energy proportionality just for completeness.

222
Daniel Schien 00:54:56.680 --> 00:55:17.825
Energy proportionality is defined as the degree to which the power consumption
varies, depending on utilization. Ideally, if a device is not used, we would
like the power consumption to be 0 that would be perfect energy.
Proportionality. That is certainly not the case for energy.

223
Daniel Schien 00:55:17.854 --> 00:55:38.014
Device for network devices, energy, baseline, energy consumption so the power
consumption when the device is not used isn't the order of 80% there's some
variation um, depending on what network device 1 looks at, but for wired
networking devices, I don't think that's heavily contested.

224
Daniel Schien 00:55:39.154 --> 00:55:59.764
The episode of that is that the energy consumption for the network is very
insensitive to traffic and I like to point to this news article from have been
criticized for it but I think doesn't doesn't pick up on the irony of it. So,
in the article.

225
Daniel Schien 00:56:00.130 --> 00:56:21.095
They are promoting that the networks network operators didn't see a change in
the energy consumption of the networks despite an increase in traffic. It means
that the overheads in their network are so large that they didn't have to.

226
Daniel Schien 00:56:21.484 --> 00:56:42.424
Um, they didn't have to do anything to the networks to accommodate that extra
traffic. And that means that before they were wasting a lot of electricity and
the other consequences that this formula cannot be used to evaluate changes to
user.

227
Daniel Schien 00:56:43.055 --> 00:57:03.305
Um, so there's an article here, um, I think, um, call me and are criticizing
this in the jewel paper. If not, they, they would have agreed with me where
this is just 1 case of many and I, and our own work with a phone.

228
Daniel Schien 00:57:03.579 --> 00:57:24.724
Victim of this reductionist simplistic assumption about the way that the
networks work. But here they're saying, if we would turn off our video cameras
in calls, we would be able to reduce the carbon footprint from 9.4 kilogram of
per month to 377 grams. And.

229
Daniel Schien 00:57:24.754 --> 00:57:45.874
Sally, that is not the case in particular. It won't result in an instantaneous
energy consumption change in the network. It does affect the power consumption
of my laptop. Um, whenever I'm in the call, the fan just kicks in and I've got
a new camera, which is just so bad in terms of the overhead.

230
Daniel Schien 00:57:45.905 --> 00:58:07.025
Power consumption that it induces in on the user device, but the, the immediate
impact on the network is very small. Okay. So I mentioned that, uh, has
contributed to this and I don't want to take all of your powder. However, what
I want to point out.

231
Daniel Schien 00:58:07.054 --> 00:58:27.694
Is that, um, the updated model that yens provides? Which factors which, which
takes this uh, here this exactly this observation that the power consumption,
the total power consumption is the sum of some base power consumption and then
some factor, some dynamic portion that is proportional to use.

232
Daniel Schien 00:58:28.745 --> 00:58:34.145
And he and he just provides coefficients for this, um.

233
Daniel Schien 00:58:35.254 --> 00:58:56.374
And then multiply the base power consumption over time. Now that the problem
with that is that it, it ignores a fundamental the fundamental dynamic of how
infrastructure is operated. Now. I'm, I'm, I'm not.

234
Daniel Schien 00:58:56.405 --> 00:59:16.895
Creating the systems myself, but from conversations with good colleagues and
project partners, I understand that infrastructure capacity is there to enable
services. And as it increases the organizations that provide the services, they
are making use of this capacity and they are implementing.

235
Daniel Schien 00:59:17.734 --> 00:59:38.614
User quality, better user experience and users looking looking for this, they
want to have HD video calls, and they want to have 8 K television probably. And
this has been going on as far as long as we are measuring the energy
consumption and the data.

236
Daniel Schien 00:59:38.705 --> 00:59:59.795
You mean, the Internet and and the volume of traffic in the network has very
consistently doubled every 2 years and that drives the infrastructure capacity.
Um, and this is just a given. It's extremely constant. Now, if we.

237
Daniel Schien 00:59:59.829 --> 01:00:20.974
Find the 2 observations that capacity increases in order to meet demand. We
know that there are these step changes here where the energy consumption in the
network changes. Now, this is not to scale and actually, as, um, as Paul is
pointed out today. So.

238
Daniel Schien 01:00:20.979 --> 01:00:22.714
Is all hot off the press.

239
Daniel Schien 01:00:23.825 --> 01:00:44.945
The baseline power consumption here can actually go down as well, but in
practice um, and this is again, it's open for debate the energy consumption of
new kit. The capacity of the network increases and if the.

240
Daniel Schien 01:00:46.084 --> 01:01:06.064
Total demand grows faster than the energy efficiency improvements of the
devices. There is no other way in the medium or long term that the energy
consumption of the network is increasing. And this is what we're looking at at
the moment, the energy efficiency improvement.

241
Daniel Schien 01:01:06.099 --> 01:01:27.244
The equipment at the moment is, we're thinking around in the order of 2025%,
but we know that they demand is increasing in the order of 50% per year. So,
just from looking from combining those 2 factors, unless we're missing
something.

242
Daniel Schien 01:01:27.274 --> 01:01:48.394
It's impossible that the power consumption of the network will remain constant.
Now. Um, so this, this is, but we are not representing this mechanism in the
way that we're thinking about energy intensity. This blue line here shows us.

243
Daniel Schien 01:01:48.544 --> 01:02:09.544
This IV, this energy intensity of data transport this is what we're using in
order to calculate those footprints where we're saying your use of the network
is responsible. We are, we're assigning to you responsibility for a part of the
electricity consumption and we don't care what.

244
Daniel Schien 01:02:09.635 --> 01:02:16.865
Time of the day or the year, or when you're using this. Um, but really, we have
just said that.

245
Daniel Schien 01:02:18.604 --> 01:02:39.154
The, um, it's the, the peak, it's, it's the, it's the, it's, it's when we're
coming close to the use of the capacity of the network that really drives the
increase of the, um, the, the change of the of the network. And so.

246
Daniel Schien 01:02:39.185 --> 01:03:00.305
We argue that actually it is, it is more representative of what is happening
causally to burden traffic that uses the system at peak time. And and that's
exactly what we're proposing. We're proposing a transformation of the.

247
Daniel Schien 01:03:00.334 --> 01:03:21.454
The intensity factor that takes takes into account the degree to which a
service contributes to peak demand and and here. So there are many different
ways how to do this. The version that we present here is just taking into
account the regular diagonal pattern that we observe.

248
Daniel Schien 01:03:21.484 --> 01:03:42.244
Um, this is from the, from the London Internet exchange, but similar patterns
can be found across the planet and, and we take this and we are moving just the
base power consumption and shifting it towards peak. And and that's a penalty
for traffic.

249
Daniel Schien 01:03:42.724 --> 01:03:54.244
The closer that it comes to peak time use. Okay that's the proposition here.
Here's the formula for it. And, um.

250
Daniel Schien 01:03:55.835 --> 01:04:13.325
We've seen this before, and then the next step on top of this is the
observation that carbon intensity of the electricity is also variable. And, um,
and now, if we are combining the tool, we are, um, we come to an even, um.

251
Daniel Schien 01:04:14.105 --> 01:04:34.595
More pronounced difference, um, of carbon intensity over the course of the day.
Um, again, there are lots of nuances in here that can be, um, that can that we
can talk about. I really look forward to discussion. Um, but, uh.

252
Daniel Schien 01:04:34.804 --> 01:04:44.674
Just want to close with, with, with 2 observations we start in this with when
we present this metric, we start with the observation of a causal relationship
here.

253
Daniel Schien 01:04:46.234 --> 01:04:54.874
Well, we said that traffic that contributes to to, to the use of the network
close to.

254
Daniel Schien 01:04:56.105 --> 01:05:16.925
With the fact that the capacity increases are necessary that we start with this
causal observation, but we end up with a metric that has a normative effect
because it incentivizes the users as well as the providers of services to think
about ways to carry a demand smoothing.

255
Daniel Schien 01:05:17.104 --> 01:05:36.274
Shifting traffic away from the peaks towards periods that are less heavily
penalized and, and so resulting in a manifest reduction of of peak demand. And
so we can delay the need for increasing network capacity, which is a good thing.

256
Daniel Schien 01:05:38.674 --> 01:05:41.674
Okay, that is it. Thank you very much.

257
Daniel Schien 01:05:46.235 --> 01:05:47.225
Closing this quickly.

258
Colin Perkins 01:05:48.965 --> 01:05:57.755
Thank you very much Daniel. Interesting. Talk. Do people have questions for
Daniel? We have. We have some time.

259
Colin Perkins 01:06:00.155 --> 01:06:02.735
A bunch of discussion in the chats, like.

260
Rob Wilton 01:06:04.835 --> 01:06:05.255
And.

261
Rob Wilton 01:06:06.485 --> 01:06:27.425
From the meeting, I was going to make a quick comment, I think sort of breaks
told us in the chat here. Is that looking at some of the sort of paragraphs
that shown earlier? It may be that what's in the core competency, in terms of
routing protocols in the core network and things like that would have very
little impact on the Internet usage. If a lot of how it's been consumed on the
other devices.

262
Rob Wilton 01:06:27.874 --> 01:06:42.034
So, I still think it's, it's clear that we need to understand, or have good
data for exactly where it's been used to make sure that when we try and
optimize anything we do here that we get a good, good, good value in terms of.

263
Rob Wilton 01:06:43.804 --> 01:06:51.994
Uh, cause you could optimize the core routing and find you actually make, like,
1% less overall difference. So we need to make sure we focus on.

264
Colin Perkins 01:06:55.955 --> 01:07:10.805
Yeah, that was certainly certainly an interesting point to discuss that the
energy proportionality that data makes some of this clear that there's only so
much scope we can change by changing the protocols. I think.

265
Colin Perkins 01:07:13.534 --> 01:07:13.864
A.

266
Pernilla Bergmark Ericsson 01:07:16.624 --> 01:07:33.844
Yeah, just a small question. So so, regarding your step graph, uh, Dan, uh, to
me, it seems like the X axis would rather be loaded than time because time
would make some assumptions on the development of, of the efficiency and so on
in place.

267
Pernilla Bergmark Ericsson 01:07:33.849 --> 01:07:36.784
Sadly, so.

268
Pernilla Bergmark Ericsson 01:07:38.194 --> 01:07:41.674
I am I missing something or or do you think that makes sense?

269
Daniel Schien 01:07:41.854 --> 01:07:49.384
Yeah, you can't directly swap the tour because because Lord is.

270
Daniel Schien 01:07:51.124 --> 01:07:58.894
Yeah, yeah, I I think I think that's that that wouldn't quite work because of
the low energy. Proportionality. I think the steps.

271
Pernilla Bergmark Ericsson 01:08:02.674 --> 01:08:21.784
So, that was just to clarify the questions to me, it seems like the steps are
more related to the growth of the load. So, at some point, you, you pass a
threshold then to time as such. Because time in this case would more be a
representation of the.

272
Pernilla Bergmark Ericsson 01:08:22.383 --> 01:08:24.064
Uh, but maybe I missed something.

273
Pernilla Bergmark Ericsson 01:08:25.054 --> 01:08:25.653
Way you think.

274
Daniel Schien 01:08:25.653 --> 01:08:42.934
Yes, yes sorry. I totally agree with that. That it is. Ultimately you're
interested in your time. Isn't isn't so important I think the 2 are related.
Yeah, because we've got that 2 x increase in demand over time. So.

275
Daniel Schien 01:08:42.963 --> 01:08:51.963
So so certainly the 2 are related, but ultimately, the main thing that you're
concerned with is is load. Yeah, I do agree with that.

276
Pernilla Bergmark Ericsson 01:08:52.383 --> 01:08:53.943
Okay, then I understand thanks.

277
Colin Perkins 01:08:58.024 --> 01:08:58.984
Did you have a question.

278
Dom Robinson 01:09:00.394 --> 01:09:03.064
It's not done excellent presentation. Of course.

279
Dom Robinson 01:09:04.234 --> 01:09:16.294
Now, Christian, that's a fantastic way to visualize the upgrade book. Um,
question for you on that. Um, did you include sort of application layer
services? Like, CDM?

280
Dom Robinson 01:09:16.863 --> 01:09:25.834
Or is that is this type already to specific network or infrastructure or have
you got cloud services application layer stuff.

281
Daniel Schien 01:09:26.044 --> 01:09:26.254
So.

282
Dom Robinson 01:09:26.403 --> 01:09:28.744
Factored into any particular work flow representative.

283
Daniel Schien 01:09:29.524 --> 01:09:37.923
So, um, this is purely conceptual. I think the principles apply to to, to, to
boxes.

284
Daniel Schien 01:09:37.955 --> 01:09:58.925
Or, to other other equipment, we did model this the demand curve comes from the
London Internet exchange from a typical work day. And so this is sort of real
data but the rest is it's just proportions.

285
Daniel Schien 01:09:59.944 --> 01:10:20.044
I think there's, there's more work to be done to, to look at how this would
plan play out with server equipment that is now more energy proportional. We're
talking maybe 40 based power consumption rather than 80%. The principle is the
same. Now, your, your.

286
Daniel Schien 01:10:20.229 --> 01:10:27.934
The gains in terms of the penalty for for peak usage they would, they would be
smaller.

287
Dom Robinson 01:10:28.594 --> 01:10:30.514
Yep, cool. Good.

288
Uri Cummings 01:10:31.234 --> 01:10:34.384
Hold on back in, they just, like, demanded going outside.

289
Daniel Schien 01:10:38.584 --> 01:10:39.934
Sorry, was that the question.

290
Colin Perkins 01:10:46.175 --> 01:10:46.895
Sure.

291
Colin Perkins 01:10:50.705 --> 01:11:02.285
Okay, uh, so I think we should maybe move on to, um, Genesis presentation. Uh,
do you feel do you know if the shut? The slide sharing is working now.

292
Jens Malmodin 01:11:03.125 --> 01:11:05.015
Uh, doesn't seem to work.

293
Colin Perkins 01:11:05.075 --> 01:11:07.475
Okay, I can try and share it from here.

294
Jens Malmodin 01:11:08.315 --> 01:11:09.395
So, I send you the.

295
Jens Malmodin 01:11:09.694 --> 01:11:10.774
So, I can.

296
Colin Perkins 01:11:11.854 --> 01:11:13.924
Let me see what I've got.

297
Jens Malmodin 01:11:14.074 --> 01:11:14.674
Yeah.

298
Colin Perkins 01:11:19.205 --> 01:11:19.745
Huh.

299
Jens Malmodin 01:11:19.895 --> 01:11:33.125
It was some kind of a set thing. I'm on a switch from PC to Apple and there was
some kind of private accepting teams work. I have only done team so far. So it,
it them sharing works.

300
Colin Perkins 01:11:34.295 --> 01:11:35.225
Yeah.

301
Colin Perkins 01:11:41.645 --> 01:11:43.085
Spend the right window to.

302
Jens Malmodin 01:11:43.625 --> 01:11:46.175
Yeah, oh.

303
Jari Arkko 01:11:46.535 --> 01:11:47.345
Seeing it now.

304
Jens Malmodin 01:11:48.035 --> 01:11:48.395
Yeah.

305
Colin Perkins 01:11:48.695 --> 01:11:53.195
I'm not sure I can get easily get rid of the thumbnails on the side, but that
is may be.

306
Jens Malmodin 01:11:53.195 --> 01:12:01.745
No, maybe you can just take if you can see all then it's fine by me. So a short
presentation for me yes.

307
Jens Malmodin 01:12:02.074 --> 01:12:22.894
Eric, so, uh, we talked a lot about the Internet with the sector and, uh, the
title I wrote here was it consumes less energy and it less carbon than what
management this goes along with what was said in the 1st presentation. Uh,
unfortunately you can find other information.

308
Jens Malmodin 01:12:23.014 --> 01:12:44.044
About the Internet on the Internet, so you can go to the next slide. Just some
examples from BBC Guardian. So usually they talk about the Internet anything
more common than the aviation sector stream. Media video is especially.

309
Jens Malmodin 01:12:44.345 --> 01:12:45.815
Consume lots of energy.

310
Jens Malmodin 01:12:47.135 --> 01:13:07.775
The sector groups, fast emissions, energy and so on and I I say, no, that's not
the case. That's all false. And if we start, if we go to the slide number 3
there, so this is the kind of comparison I would like to make, uh, if we take
the aviation sector.

311
Jens Malmodin 01:13:09.484 --> 01:13:29.134
uh so this is results from uh uh another swedish uh researcher about the
innovation sector so in two thousand and nineteen about four percent of global
and as you can see to the left so it's not only see we have also

312
Jens Malmodin 01:13:29.404 --> 01:13:49.954
Emissions and effects, especially with the high attitude effects that they
knocks and so on and then we need to produce the and we should also look into.
Okay. What, what does it take to construct all the airports and aircraft and so
on? Unfortunately, no. But as done that research.

313
Jens Malmodin 01:13:50.464 --> 01:14:11.044
I think it's pretty large anyway, if we look at the sector and the typical
split and definition, I use you have the data centers, you have the networks
and data centers and networks down. It's the operation. That's most important.
They are on 24, 7 all year round.

314
Jens Malmodin 01:14:11.704 --> 01:14:32.614
But as you can see here, it's all the devices, all the phones, smartphones, PCs
and so on, uh, not to forget the home routers the, we, all have it fixed
broadband. And then, since these devices are not on all the time, just a few
hours every day, uh, on average.

315
Jens Malmodin 01:14:32.825 --> 01:14:53.735
then the uh production of these devices actually has it's nearly as much as
operation but if we add all this up we only get to around seven hundred million
pounds uh and i will show you how you arrive at such a number so i've been
doing these start this now since the first

316
Jens Malmodin 01:14:53.974 --> 01:14:58.144
We did was back in 2005 some more than 15 years ago.

317
Jens Malmodin 01:15:00.155 --> 01:15:20.525
And then if we look at that Asian sector, we actually see that. It's only only
1% of all people they do about half of our flying, and only slightly more than
10% fly a given year. So, 1% of all people there at travel. It's actually a bit
more than the entire sector.

318
Jens Malmodin 01:15:21.694 --> 01:15:40.354
Which is used by, I guess, we can say around 5Billion people daily for hours.
So, let's go to the next slide. I, this is how we, this is ongoing research,
but we have done start this a couple of times now, since 2005 about sector.

319
Jens Malmodin 01:15:41.764 --> 01:16:02.854
You have them bolded footprint for data center networks and use the devices and
operational top there. Uh, and when it comes to how to measure, uh, these
parts, I think we need to go to all the companies. And today, the good story is
that most companies actually.

320
Jens Malmodin 01:16:02.914 --> 01:16:23.734
Porter electronics consumption and their emissions and so on so, for data
centers, we have data from 4,646, large companies, whether it's Google,
Microsoft, and all the others about 60% operational data centers. Uh, but these
companies actually represent more than 90% of all data traffic.

321
Jens Malmodin 01:16:24.635 --> 01:16:45.155
for network operators sixty seven launch operators eighty percent of all the
subscriptions in the world but then we need to add an estimate for the
enterprise networks officers most airports and so for the onboarding part we
have done a separate start where we used data from

322
Jens Malmodin 01:16:45.245 --> 01:17:01.985
59 large manufacturers, so we have a lot of data and then with such large data
covering samples, it's you can accurately extrapolate to the total. So if you
go to the next slide.

323
Jens Malmodin 01:17:03.815 --> 01:17:24.365
Uh, we have the total effort sector, so about 700Million, we also do the same
kind of assessment for closely related sectors. So we have the sector
entertainment and media and today it's mainly made up of TVs. The, any
consumption of.

324
Jens Malmodin 01:17:24.815 --> 01:17:45.635
The boxes and TV networks are just small in all this and then other consumer
electronics have gone from being quite many cameras areas and so on. But today
it's most of that function is within smartphones that we don't buy those
products anymore. So that.

325
Jens Malmodin 01:17:47.314 --> 01:18:04.024
Reducing impact on the total paper media, uh, and then you have some new
cameras, uh, cryptocurrency, security cameras and so on, but still rather
small. Uh, you can go to the next slide.

326
Jens Malmodin 01:18:06.514 --> 01:18:25.144
So, we follow back to 2010 and compare to this new 2020 start that we see
actually a total reduction, uh, based on. It's actually less tedious and TVs,
pieces and TV. So, 2020, compared to 2010, 2011.

327
Jens Malmodin 01:18:26.795 --> 01:18:47.795
We have the smartphone impact. I mentioned a lot of functions. You need a
dedicated devices too, before as cameras are now killed in a smartphone, uh,
for PCs. We mainly use laptops, smaller, much, much more energy efficient on
both desktops.

328
Jens Malmodin 01:18:49.204 --> 01:19:04.504
we watch tv and today's tvs are actually a lot more energy efficient even
though they are larger and then there's a set paper media reduced by about
thirty since uh before two thousand and ten

329
Jens Malmodin 01:19:06.305 --> 01:19:27.215
Okay, we can go to the next slide. Maybe I shouldn't go so much into details,
but this is kind of how we go about this data, this an example of measured
electricity consumption by these companies that run data centers. You have the,
you have the Google, Microsoft.

330
Jens Malmodin 01:19:27.334 --> 01:19:48.454
Facebook all reporting good data Amazon we still have to estimate them, but
they are reporting now, uh, their whole organization, but then they have 1 and
a half 1Million employees for retail and warehousing and other businesses. So
is probably the largest consumer. We don't.

331
Jens Malmodin 01:19:48.484 --> 01:20:08.944
Know, exactly how to this, but I think that estimate is pretty good. Uh, and as
you can see the last decade number of new servers in the world, it's been very
stable outside China, but then China has, uh, built out their data center
capacity. So, most of the.

332
Jens Malmodin 01:20:10.444 --> 01:20:30.364
Increase in, in China, uh, you can step 2 networks, same for networks. And here
we, I can point to this interesting stuff is where followed European operators
for 10 years and they've seen that electricity consumption is more or less.

333
Jens Malmodin 01:20:31.205 --> 01:20:35.285
Exactly, the same very stable during the last 10 years.

334
Jens Malmodin 01:20:36.544 --> 01:20:52.774
If you look to data traffic, it has increased by 20 around 20 times. Uh, so we
don't see any relation data energy so more capacity, but it's the same
electricity consumption.

335
Jens Malmodin 01:20:56.855 --> 01:20:59.315
Yeah, we can step to go to the next.

336
Jens Malmodin 01:21:01.385 --> 01:21:13.535
For the manufacturing there we see in the manufacturer semiconductors and this
place is really important. And then you have a large assembly count.

337
Jens Malmodin 01:21:13.925 --> 01:21:21.515
Focused on there too and this is the start that we have submitted. It's under
review and I think it.

338
Jens Malmodin 01:21:21.519 --> 01:21:42.274
It's very interesting for everyone that is done lifecycle assessment on
devices, like PCs and smartphones. So we have based on 20 new 2020 data. So we
have done an allocation for each device that that, uh.

339
Jens Malmodin 01:21:45.424 --> 01:22:03.814
Uh, okay, a little bit we can go to the next slide. If you look on results for
data centers and networks over time, this is what we have seen, uh, for data
centers. Ourselves is very similar to what.

340
Jens Malmodin 01:22:03.819 --> 01:22:24.064
To me, and in the U. S. has published over the years, uh, not surprised we're
using the same kind of model, but we are adding here is also this, uh, data
center company data, that kind of verifies what has done over the years.

341
Jens Malmodin 01:22:25.804 --> 01:22:45.874
Data centers, 2029% of global electricity networks a little bit more as grown
was similar to data centers. It's not loud. Yeah, we basically, because we
built out networks are all around the world and if you step 1.

342
Jens Malmodin 01:22:46.355 --> 01:23:06.575
1 further we look at networks and I'll look at number of subscriptions in the
network. We see. Yeah the network electricity the built up the networks very
proportional to number of subscriptions. It's actually so that it's 3 times as
much subscriptions, but has.

343
Jens Malmodin 01:23:07.294 --> 01:23:14.704
Increased by about 2 types, so we are a bit more efficient per user, but then
as you can see data.

344
Jens Malmodin 01:23:16.324 --> 01:23:29.974
51100timeswhatcan we expect to 20, 30, maybe 10 times I think maybe 8 times,
like 3 doubling, according to more slow a little bit slower. It's moving a
little bit slower today.

345
Jens Malmodin 01:23:33.035 --> 01:23:39.665
And then we can go to the next slide. So if we reconnect to, um.

346
Jens Malmodin 01:23:41.164 --> 01:23:53.194
To, um, what's been said in media, where does these figures come from? Uh,
there's many other startups out there doing, uh.

347
Jens Malmodin 01:23:55.534 --> 01:24:09.934
And and when you look at them, it pretty much looks like this, that it is
assumed that energy goes up as data goes up. And then you will see this
exponential increase reactions in the future.

348
Jens Malmodin 01:24:12.155 --> 01:24:15.155
However, if you if you step to the next slide.

349
Jens Malmodin 01:24:16.924 --> 01:24:30.814
I say this, this is based on the wrong assumption, that exponential increase in
data, at least an exponential increase of energy. And according to all the
measurements we have done over this past 10 years.

350
Jens Malmodin 01:24:31.925 --> 01:24:52.955
That is not the case. So typically you change 1 piece of equipment to a new
piece of equipment. The typically can do 10 times more data, but often it's
really to a lower energy consumption. So it improved further which the network.

351
Jens Malmodin 01:24:53.584 --> 01:24:54.604
Kind of, um.

352
Jens Malmodin 01:24:56.284 --> 01:24:56.704
Data.

353
Jens Malmodin 01:25:01.114 --> 01:25:01.954
And then.

354
Jens Malmodin 01:25:05.014 --> 01:25:25.204
Yes, that was it unfortunate. I have some interesting slides I could have shown
you about when you take the figures down to per subscription, which I think is,
is what we need to do. Uh.

355
Jens Malmodin 01:25:26.554 --> 01:25:45.334
But I can show them on a later state to submit all up. So, in contrast them to
what can be seen in media the sectors, carbon footprint is stable. It is not
growing, uh, and 1% of all people on earth more when flying.

356
Jens Malmodin 01:25:47.825 --> 01:26:06.245
Uh, compared to 60%, um, or people using daily for hours, and then the time it
to me, and the sector TVs are the consumer electronics and paper media is
actually being reduced by 30% or 30%, lower emissions versus 10 years ago.

357
Jens Malmodin 01:26:07.774 --> 01:26:27.814
Uh, the sector, the data center companies and manual the operators, or the
leading investors now, in renewable electricity in the world. Uh, there's a
start there by international Energy agency that shows this and we also see this
in the data. Uh.

358
Jens Malmodin 01:26:28.924 --> 01:26:50.044
Mentioned that, but you still saw all these green figures about the electricity
consumption in the background data, and that was renewable electricity. So some
of these companies have already switched 100 to renewables building their own
solar forms and wind farms.

359
Jens Malmodin 01:26:50.049 --> 01:26:56.074
Important more data in the future do not mean more energy. So, uh.

360
Jens Malmodin 01:26:59.135 --> 01:27:19.355
We can keep on surfing the web stream music and with you and downloading games
and play for fun. It will not consume more energy probably less energy. Uh, 1
of the slide. I had this background if time allowed me to, but I can just
mentioned that, but.

361
Jens Malmodin 01:27:20.315 --> 01:27:40.325
My new fiber connection that I dig down this summer, the connection consumes
around 2.5 watts, uh, at, uh, fixed network site, uh, in my old house, the
phone line.

362
Jens Malmodin 01:27:41.254 --> 01:27:47.764
Consume tablets total, 10 megabit per 2nd. Now I have.

363
Jens Malmodin 01:27:48.154 --> 01:28:09.004
Megabit per, 2nd, so at 100 times more data but oh, it's 4 times less energy
uh, the route, the home router and fiber more than consumed 12 watts. Uh, back
in the days you need the modem, and the router consumed 18 months. And then you
have the core, so called quarter.

364
Jens Malmodin 01:28:09.395 --> 01:28:29.585
Nobody has like, phones anymore, fixed phones and that consumes another few
watts. So, near the halt consumption for, for port by 100 times more data and
in the future, who knows? Maybe in 10 years uh, we can all get 10.

365
Jens Malmodin 01:28:30.215 --> 01:28:41.945
Gigabits per. 2nd, what to do with all the data I have no idea if we're not
cost more energy. That is 1 thing. I know though. Uh, okay. Thanks for me.

366
Jens Malmodin 01:28:44.525 --> 01:28:55.895
Sorry about the slide. I thought this was going like running water. Uh, but
apparently the Webex and Apple don't talk to each other.

367
Colin Perkins 01:28:56.885 --> 01:29:05.405
Eh, I think no matter how much data we have the technology is always going to
defeat us at some point. Uh, thank you for the.

368
Colin Perkins 01:29:06.335 --> 01:29:13.745
Some interesting data some interesting perspectives. I, I'm sure someone has
some questions and some comments.

369
Colin Perkins 01:29:16.714 --> 01:29:17.554
Go ahead Danny.

370
Daniel Schien 01:29:19.144 --> 01:29:37.174
Thank you very much. And I will open the stage with a provocative point. You
were saying that we should, we should not worry and keep on consuming because.

371
Daniel Schien 01:29:38.284 --> 01:29:52.984
Say the energy consumption will remain the same if the rest of the economy does
deliver and aggressively the compromise sector will look very bad. If we don't.

372
Jens Malmodin 01:29:55.624 --> 01:29:58.984
Yeah, but I, I, I'm very well.

373
Jens Malmodin 01:29:59.045 --> 01:30:20.135
I mean, looking to what is happening now, I think that in a few years, the data
center companies and the network companies will probably be the ones that task,
which to renewable, which is the fastest lander other sector can do, uh, for
data centers. It's relatively easy. They are not spread around.

374
Jens Malmodin 01:30:20.945 --> 01:30:41.195
Already located in countries like Sweden, uh, with access to renewable
electricity for networks is a bit harder. Especially if you, if your own
account, where there's little access to renewable electricity. But hey, look at
the India, they have more than 100,000 sites with renewable electricity.

375
Jens Malmodin 01:30:41.314 --> 01:30:45.364
Production, um.

376
Jens Malmodin 01:30:46.985 --> 01:31:08.075
So, I'm not worried and then I would, I would challenge you a bit there when
you say that, when you install new things. Uh, and it goes up what we actually
saw the opposite in European when when, when data went up they had to actually
modernize. But what they had to do was take out the oldest equipment, they
couldn't cope with the data traffic.

377
Jens Malmodin 01:31:08.374 --> 01:31:21.214
And when they install new equipment, it consumed less. So we actually saw they
weren't the opposite direction, much more data. But then then you just start to
actually trickle down a little bit.

378
Daniel Schien 01:31:22.114 --> 01:31:29.224
So so I don't want to hijack the entire conversation, but I think this is
another point that needs drilling into.

379
Daniel Schien 01:31:29.284 --> 01:31:49.564
Um, I, I totally get the point that if you retire legacy equipment, you can
realize substantial savings. Um, but if you're if you're just looking at the,
the trend data and I think there's little the, the assumption that the
equipment on.

380
Daniel Schien 01:31:50.405 --> 01:32:10.955
Energy efficiency with those 20 to 25% on average. Um, but the demand for data
for capacity increases by 50%. Um, you just can't combine those 2, those 2
trends to end up with something that remains the same.

381
Jens Malmodin 01:32:16.384 --> 01:32:24.364
But what how do you mean? I mean, more more basically, we're starting from
we're still on the very low level, right?

382
Daniel Schien 01:32:24.394 --> 01:32:32.464
Yeah, so you do provide Erico perspective you say we are we have, um, we've got
data reaching back 20 years.

383
Daniel Schien 01:32:32.705 --> 01:32:53.825
Data shows that, despite those enormous increases exponential increases in data
volume into consumption remains stable. And I, I take that point. The question
is, um, I'll be nearing the end and underneath. I know when we walked through
the data center.

384
Daniel Schien 01:32:53.854 --> 01:33:14.974
Stockholm you showed me rows of equipment that has turned as being turned off.
The question is if there's enough legacy, still there, enough margin margin
that can be trimmed to keep us on that trajectory for longer or if, or if we
are seeing the end.

385
Daniel Schien 01:33:16.234 --> 01:33:17.674
That's the question and.

386
Jens Malmodin 01:33:18.544 --> 01:33:36.124
I think to begin with, I would say that is only proportional to number of human
users that that is what did not not their usage, but actually nodes the
connection points. So, uh, I mean, the, the fiber.

387
Jens Malmodin 01:33:36.130 --> 01:33:37.505
I got this summer.

388
Jens Malmodin 01:33:39.034 --> 01:34:00.094
It's a little piece of equipment that they put in the port, uh, consumes what?
And then the, like, in other words so, towards, uh, when I started to work or
maybe a few years before, then you would need then you have the.

389
Jens Malmodin 01:34:00.100 --> 01:34:20.765
The same capacity going from 1, big city to another and you would need a whole
room full of equipment, consuming several kilowatts. It had my 5 minutes, the
same capacity, but it only consumed like a 0.1% of the.

390
Jens Malmodin 01:34:21.664 --> 01:34:42.304
And, uh, I'm, I'm only using, like, 1% or below 1% on average or the capacity.
Uh, so there's no, the only way to sort of and this, I mean, that sure we can
have sleep mode and try that and so on but in there and it's.

391
Jens Malmodin 01:34:42.399 --> 01:35:03.544
It's, uh, it's only a few words that can say per user and even if I use 10
times more data, it wouldn't change the voltage of my connection or my router.
And so it will, uh, W, when I load my router with the 500 megabits per. 2nd, it
goes.

392
Jens Malmodin 01:35:03.784 --> 01:35:12.514
Oh, 3, watts, if I'm on the lawn port, if I own the WI Fi, it can go up with
more than it works.

393
Colin Perkins 01:35:15.034 --> 01:35:34.024
Clearly other factors in whom, uh, energy usage, which are affecting this, um,
you know, the WI, Fi, the PC that's attached to it and so on. That's yeah. Um,
uh, I see there are other questions in the chat. I see Dom has his hand up to
ask a question as well.

394
Dom Robinson 01:35:36.004 --> 01:35:56.374
Thanks Kevin. Yeah. Um, uh, really interesting presentation, particularly
improved by the take down of the aviation, uh, measurements that was really,
um, engaging because I've been I've been up with just, I think, using that
comparison, without fact checking it. And that seemed like a fairly robust fact
checking. So, I think for the, um, with the, with respect to your.

395
Dom Robinson 01:35:57.575 --> 01:36:18.275
The, the, um, your views about the network scaling up in terms of usage, but
the energy consumption, uh, essentially staying flat or even potentially
reducing. Um, do you measures include, um, uh, it's almost the same question. I
asked, Dan, did they include.

396
Dom Robinson 01:36:19.204 --> 01:36:39.154
Is there any thought that actually a lot of the energy use might be staying
flat on the networks because the traffic, the heavy duty traffic has been
offloaded to, which are outside of your measurement scope. Because I work with
a lot of in streaming, and we are only just at the beginning of measuring these
things in any practical terms. So it's, it's.

397
Dom Robinson 01:36:39.785 --> 01:36:47.555
It it's important, we get you that data, but if you've got other sources of
that, I'm intrigued because I see it only energy bills going up.

398
Jens Malmodin 01:36:50.974 --> 01:37:10.354
Yeah, I mean, uh, our company, we have our own little media company that have
both streaming. Uh, and then typically what you would you be using. Like, I'm
also wants big network. I connect to, uh, Daniel, you mentioned.

399
Jens Malmodin 01:37:10.714 --> 01:37:13.444
Work with the requirement, trust and Netflix and others.

400
Jens Malmodin 01:37:15.065 --> 01:37:33.395
And in that work, that there is a, uh, uh, qualification or their entire work
network, including all the C down. And so, and that part was rather small
compared to the rest. And in our data, we have.

401
Jens Malmodin 01:37:34.954 --> 01:37:55.564
Big big players all the big players are there like lumen and, uh, they have
others, uh, adding all the ports. Um, you have Amazon the other 1 uh, we
actually ask our providers, uh.

402
Jens Malmodin 01:37:55.594 --> 01:38:01.414
For the data, so to speak and get it, I cannot share it. Unfortunate. Of
course. But.

403
Jens Malmodin 01:38:02.915 --> 01:38:23.765
I think that when it comes to networks is the last mile, that always has been
the number 1 thing. And as soon it's like, as soon as I mean, I see it in the
cabinet with around 100 dollars. And on the back side, it's it's 110 Gigabit
port out, which will share and have a hard time.

404
Jens Malmodin 01:38:23.794 --> 01:38:44.134
Feel it, and it consumes nothing compared to all the access that has to go out
to everyone else. So I'm not so worried about the core network. I also know we,
we did some we did measurements, uh, 1 of the Atlantic cables and so that yeah.

405
Jens Malmodin 01:38:45.124 --> 01:38:50.674
Very small consumption compared to the rest uh.

406
Dom Robinson 01:38:52.054 --> 01:38:54.094
I'd like to volunteer some data in the spring.

407
Dom Robinson 01:38:57.214 --> 01:38:58.774
So, uh, I'll connect with.

408
Colin Perkins 01:38:59.164 --> 01:39:09.664
That would certainly be interesting. I see some discussion in the chat so I was
going to start column E. he was just stuck a hand up. And then, uh, Louise, do
you want to jump in after as well?

409
Eve Schooler 01:39:11.764 --> 01:39:32.854
Hey, great talk, thank you so much. It's so thought provoking. It's very
perplexing. Of course why the numbers stay the same. And so that's really the
gist of my question is you've done this amazing analysis for so long. I've read
so many of your publications. It's great to hear this talk. I would love to.

410
Eve Schooler 01:39:32.884 --> 01:39:54.004
Understand more of the why? And I, um, why the numbers remain the same and it
sounds like there are some things behind the scenes, uh, including
understanding this definition of what we include in the Internet and even
trying to understand, you know, if we understood the factors for why things are
remaining.

411
Eve Schooler 01:39:54.035 --> 01:40:05.675
The same could we under do we also understand the percentage of the impact that
they're having for? Why things remain the same? So dumb talked about.

412
Eve Schooler 01:40:06.699 --> 01:40:27.094
The previous talk, you know, Dan's talk was about, um, you know, uh, the, the
load on the network. Um, and so I wonder if you have sort of a catalog of all
the things that are sort of working in tandem to offset the, the data growth.

413
Eve Schooler 01:40:27.850 --> 01:40:36.665
And if we understood that better, I mean, maybe you understand that, because
it's all in your head. But here, I would love to hear your thoughts on that.

414
Jens Malmodin 01:40:37.205 --> 01:40:44.855
I will 1st, I, I would switch around the data that okay. There's more data I
would say.

415
Jens Malmodin 01:40:46.180 --> 01:41:03.065
What happens 1st, is that we make the technology improvement and we always have
this more slower. Uh, it's it's not the low. It's like it's an observation and
it's still going on maybe not as fast and you can order if it's still, but even
though.

416
Jens Malmodin 01:41:04.655 --> 01:41:10.265
It's, it's slows down the summer, then we take up on other areas. Um, and over
time.

417
Jens Malmodin 01:41:12.694 --> 01:41:32.704
The the laptop I'm, I'm using right now, which consumes less than less than
tablets and it's about a check to when I was starting at school. It's like 100
times. Every each permit is like, 100 times more. Mm hmm. Orange memory speed,
everything data.

418
Jens Malmodin 01:41:32.734 --> 01:41:53.584
So, but it's only 110th of the way to 110th of the energy and actually everyone
can sees this. But the same thing happens in the data centers. Now, it's the
same semiconductors. That is the basis for all this and you, you, you see the
same. I mean, like the example.

419
Jens Malmodin 01:41:55.654 --> 01:42:06.874
Back in the 80 s, that you need the whole room doing the same thing that my
little port can do now, uh, that I connect to my fiber.

420
Jens Malmodin 01:42:10.085 --> 01:42:30.485
Another example, for mobile, it's like, when I started, you needed, we had this
huge cabinets, 500 kilos for GSM, uh, 1 base station and we were thinking it
would be good if we could make it much smaller. So, we can put it up in Montana
to avoid this a half of them.

421
Jens Malmodin 01:42:30.604 --> 01:42:51.634
You got lost in the cables up between the but today since a few years ago, the
radius are like, now, if 40 kilo radio, those 3 times the work that this 5 and
a kilo did when I started, but much higher capacity.

422
Jens Malmodin 01:42:51.669 --> 01:42:54.514
In terms of data, so it's not comparable. Let's just.

423
Jens Malmodin 01:42:56.164 --> 01:42:58.384
It's like a 1Million times a really.

424
Eve Schooler 01:42:58.984 --> 01:43:11.434
So, it sounds like your assessment is that just the, the sheer improvements
that are happening in the technology, have have allowed us to continue along
this path.

425
Jens Malmodin 01:43:11.884 --> 01:43:16.954
Yeah, yeah, I forgot to say that that 1st come to the technology.

426
Jens Malmodin 01:43:17.104 --> 01:43:38.194
Allow more data and then we have more data. If the technology, all of a sudden
stop believing of this. They will not be exponential more data. Because that
would then start to cost exponentially more. And the things would get
exponentially, larger because of exponential more energy consumption. Which is.

427
Jens Malmodin 01:43:38.254 --> 01:43:54.994
Physically impossible if we really want that. Yeah. Who would pay the cost? I
don't think anyone will do it, but the data amount down would be such high that
we would probably say. Yeah. Hey, it's all right. We don't have enough.

428
Jens Malmodin 01:43:58.204 --> 01:44:14.644
If if if air travel had improved as much as a electronics, the time I have been
living, we would be traveling at the speed of light 1Million times faster
1Billion kilometers an hour.

429
Jens Malmodin 01:44:17.494 --> 01:44:27.934
Improvement but that you can, of course, say that yeah. It's still you still
have this 1 piece your only writing documents yeah. In a way.

430
Colin Perkins 01:44:28.959 --> 01:44:41.134
Unconscious conscious of time that we have other things to discuss. Um, I saw
Louise, um, had some some comments in the chats. Is there anything you want to
say in the microphone?

431
Louise Krug 01:44:42.454 --> 01:44:50.104
Yeah, I mean, I just, I think maybe we should have some discussions offline.
Um, if I could share some of the stuff we have.

432
Louise Krug 01:44:50.224 --> 01:45:10.804
See, what's happening in the networks and how things changing and why things
are changing, um, models, like to dig into your assumptions about embodied
energy versus in use energy. Because again, I think what we're seeing within
our network are somewhat different.

433
Louise Krug 01:45:11.405 --> 01:45:28.385
To what you're seeing, um, and it could well, just be perspective, you know,
you're looking at the global taking into, which takes into account the sort of
social economic stuff that goes on as well as just what's happening in 1
country and 1 network. But, um, in the interest of time, I think we'll just
move along. Is that okay?

434
Colin Perkins 01:45:29.944 --> 01:45:40.834
I think that was very, very useful. Thank you. And yes, I think if we can make
some connections, then I think that would that would be helpful to dig more
into the details.

435
Colin Perkins 01:45:43.055 --> 01:45:47.855
All right so is this okay. Okay. So, um.

436
Colin Perkins 01:45:49.354 --> 01:46:10.474
I would like to try and do to finish up is the dig a little into, um, W, we
spent a lot of time talking about what we know. Um, what, what are the things?
We don't know. Wh, what are the things where we, um, WH, what are the areas
where we know our expertise is is not sufficient.

437
Colin Perkins 01:46:10.504 --> 01:46:13.324
And, um, what are the areas where we know we.

438
Colin Perkins 01:46:14.164 --> 01:46:14.854
Data.

439
Colin Perkins 01:46:16.324 --> 01:46:33.334
And, uh, I, I sort of asked this question, uh, on the, on the meeting list
ahead of time. And, uh, I've sort of put prompts for a number that the, the
email thread. So, so, um, does anyone have opinions on this? W, W, what what
what don't we know what are the nodes.

440
Colin Perkins 01:46:45.605 --> 01:46:54.425
Uh, yeah, yeah, you, you asked about the, um, for what, what, where is the
traffic in the trade offs? Do you wanna say a little about that? Maybe to get
things.

441
Jari Arkko 01:46:57.005 --> 01:47:16.685
Yeah, yeah, so, I mean, I, I was just approaching this as a sort of an yeah.
Coming into this us as non expert. So, if we want to make an improvement, I do
think that we actually do make need to make an improvement even if, even if the
energy levels are more or less the same that means that we could save.

442
Jari Arkko 01:47:16.985 --> 01:47:37.715
Some energy or be otherwise more efficient or useless money in various
enterprises. Um, energy is very costly. Uh, recently, if you've noticed. So, I
think we do need to make improvements and it's worthwhile for us to make
improvements. But, but in order to do that, we have to figure out, like.

443
Jari Arkko 01:47:38.014 --> 01:47:52.294
Effective to make improvements and, like, starting from, like, you know, not
from who is guilty. But, like, where is like, what is the proportion of, um,
environmental impacts coming from this part or that part? And, and so on.

444
Jari Arkko 01:47:53.410 --> 01:48:14.555
I guess W, W, W, we saw some of those numbers today. Um, another thing that it
would be very useful to understand, and I think that there also exists, um, at,
at least to some extent. Um, like, which part of the traffic is responsible for
what and I guess, uh, I don't know Peter on the Internet is Peter. And
streaming is.

445
Jari Arkko 01:48:14.614 --> 01:48:35.614
It's a huge fraction of that and then we could perhaps try and understand like,
you know, what what out of these things are the worthwhile things to do. I
mean, maybe the answer is that, you know, just keep optimizing, um, streaming
and video and compression and all of that. And then that's that's the, that
that's the.

446
Jari Arkko 01:48:35.764 --> 01:48:56.674
Useful thing we could do and everything else is noise. I'm not sure if that's
entirely true. But, um, the other thing that I wanted to say is that, um, I
guess, um, uh, again, this toggle sort of rely on on the improvements that had
been going on in equipment and it didn't come for free.

447
Jari Arkko 01:48:56.884 --> 01:49:18.004
It basically meant that lots of engineers and lots of standardizing people got
together and to design new stuff and new generations, and so on and chip
manufacturers came up with better process and so on. So a lot of effort has has
went into that. So, um, in order for us to do something, I think we need to.

448
Jari Arkko 01:49:18.035 --> 01:49:22.145
Again, what what the right places for us to do that.

449
Jari Arkko 01:49:24.094 --> 01:49:44.734
Um, and also, um, typically, and and you see this in, in some of the papers
that that that we have are basically all of the papers that we proposed to do
X, but X may have some trade offs or costs. And we actually want to understand
that too and not just like, you know, this, this thing here today is costly. We
propose an improvement.

450
Jari Arkko 01:49:45.124 --> 01:50:01.984
We actually have to understand what that implies also, you know, maybe you can
reduce the cost of your box or the cost of your box. But what about other boxes
in the network? Is that imply something for them? So, those sort of sort of
questions were on. My mind at least.

451
Colin Perkins 01:50:05.254 --> 01:50:09.094
Yeah, thank you. Certainly don't have opinions on these questions.

452
Suresh Krishnan 01:50:12.244 --> 01:50:21.454
Thanks, uh, like, I think that's a really good point but, um, 1, more thing I
kind of wanted to talk about is, like, you know, we are like, in the.

453
Suresh Krishnan 01:50:21.514 --> 01:50:42.604
Here right, like, and we know a lot about network so we're spending a lot of
time looking at how the networks are. Right? Like, you know, utilizing energy
and so on. I think we have a big blind spot on the end user devices. Like, you
know, how they are and the applications themselves right? And and, um, this is
something I started looking at, like, probably like a few months ago. Right?
And there's like a huge variability in, in how.

454
Suresh Krishnan 01:50:42.634 --> 01:51:03.754
Energy is consumed under devices, so we talked about like, media streaming or
like, video calls right? Like, so there's a huge difference in energy between,
let's say, a Webex call and a zoom call, right? Or a team's call right? And how
efficiently the clients have written the applications have written and there's
a big difference. If I use my Max built in the cam, versus like a Logitech cam,
that's got 164. so I think we don't.

455
Suresh Krishnan 01:51:03.785 --> 01:51:24.905
Have a really good grip on this. So you're kind of putting this all in the same
bucket and, um, I, I don't know the good answer to this, but I think this is
something we need to look at, like, uh, for sure. I think like, point, I was
like, oh, like whatever we do, it's gonna stay the same, but we assume that
things are gonna get improving on the consumer side, right on the edge device
side. But I think, um, that's something that.

456
Suresh Krishnan 01:51:24.935 --> 01:51:41.585
Needs to get better because, and we are not doing this, because it's a very
hard problem. So there's a lot more measurement points and there are like, a
lot more variations in the in the points themselves and I just want to throw
this up. I don't have a solution. I just want to see, like, you know, how we
kind of addresses, uh, going forward.

457
Colin Perkins 01:51:43.414 --> 01:51:58.054
I mean, it's certainly clear that each, irrespective of how much data it, you
know, how much energy it costs to transmit a bit there are costs in producing
the bits. Somehow. Um, they vary significantly, depending on what those bits
are and how they generated.

458
Jens Malmodin 01:52:03.935 --> 01:52:21.185
I mean, as the sorry, that's the networks or now, it's, it's they are consuming
this more or less the same all the time. Uh, usually we are on a low, low part,
or the, the, we are using just a few percent.

459
Jens Malmodin 01:52:21.994 --> 01:52:29.614
But still, let's consume me. So, uh, 1 thing is that yeah please use it more
than, uh.

460
Louise Krug 01:52:29.824 --> 01:52:30.244
What.

461
Jens Malmodin 01:52:31.024 --> 01:52:32.194
Make more use of it.

462
Louise Krug 01:52:37.235 --> 01:52:54.965
It's been it's very hard work to keep that profile and usage will make that
much harder to keep when we were trying to reduce that. Um, it it is a struggle
and that we can see what's happening. Now, we can also see that when we.

463
Louise Krug 01:52:55.204 --> 01:53:14.794
Completed some actions you can only do once, and you can see that behind that,
but basically we're trying to reduce stuff at the same time as other stuff is
growing fast. And so as a business, we're, we're keeping it in a balance. And I
suspect that's what's going on in certainly a number of the other columns. So,
if you could just keep throwing data directly.

464
Jens Malmodin 01:53:18.904 --> 01:53:24.484
Becomes, uh, gross when we allow it to grow. So when this whole new better.

465
Louise Krug 01:53:24.574 --> 01:53:27.904
There's a social economic aspect, which as well yes.

466
Jens Malmodin 01:53:28.774 --> 01:53:37.444
What I was going to say is that, I mean, when we, uh, have this online meeting,
if we compare the energy for this, it's in.

467
Jens Malmodin 01:53:37.474 --> 01:53:43.474
If you can compare it to if we would have traveled, I mean, to the meeting,
right?

468
Louise Krug 01:53:44.555 --> 01:53:45.815
Yes, that is a therapist.

469
Jens Malmodin 01:53:46.295 --> 01:54:02.825
And I mentioned it, I mean, I have a measurement device all the time on my
piece, and it goes up with the background for what's when I have a team's
conference I don't know about Webex. I have to unplug my PC so it's not
measuring it. Right now.

470
Jens Malmodin 01:54:04.414 --> 01:54:22.534
So, it's up 10%, so, from 40 to 44, what's the laptop on the screen and camera
and everything? So it's not that much. Um, uh, when I watch YouTube, it's
there's no in it's the same 40. what's like, doing a ordinary office work.

471
Colin Perkins 01:54:23.434 --> 01:54:25.234
It it it.

472
Colin Perkins 01:54:26.165 --> 01:54:45.965
The comments about, uh, understanding that this issue technical socio economic
drivers here is important, though, because, uh, obviously this affects, you
know, the number of users using the network and the number of devices that they
have, and that they bring that are being pulled in. So, yeah, put analysis.

473
Pernilla Bergmark Ericsson 01:54:47.284 --> 01:55:07.684
Okay, yeah, I was just trying to listen and at the same time reading in the
chat and I think it feels like we are discussing a couple of things at the same
time. So 1 is about what is the footprint of our city and I think when we and
especially when are talking and says that it's much smaller.

474
Pernilla Bergmark Ericsson 01:55:07.924 --> 01:55:28.834
Uh, and it is, it, it and and it's not likely to develop in line with some of
this, uh, more optimistic, um, uh, scenarios that is about understanding where
we are. And I think it's important to, to try to get a good assessment of that.
Then when it comes to comparing.

475
Pernilla Bergmark Ericsson 01:55:28.864 --> 01:55:49.984
With other activities, as aviation someone said in the chat is not important. I
think it's very important, because not for us as a industry, because we have to
look at our thing within our system boundaries, but it's also important for
society to have knowledge to understand which activities are the.

476
Pernilla Bergmark Ericsson 01:55:50.524 --> 01:56:11.134
Harming, and which are less resource intensive and then there is, I, I mean,
it's a great difference if a person is using a smartphone, or if they are going
on a cross Atlantic aviation trip, for instance. Uh, so, so it's.

477
Pernilla Bergmark Ericsson 01:56:11.195 --> 01:56:32.285
To understand those things, but that, that said, I think, of course, as the
sector, we have the responsibility for our emissions. So, although we need to
have a reasonable estimate of them and could say that some seem very that
doesn't take away the.

478
Pernilla Bergmark Ericsson 01:56:32.314 --> 01:56:53.434
Importance of optimizing and reducing the emissions which, and the energy
usage, which is associated with our sector and from the perspective of, uh,
electricity W. W. W hopefully going renewable and renewable is important. It is
the main strategy for decarbonization and we already.

479
Pernilla Bergmark Ericsson 01:56:53.464 --> 01:57:14.374
Side player, uh, even if we, we do that, of course, electricity will still be
important for a lot of different reasons but it's also important from the
perspective that, I mean, also the renewable resources. They are not endless.
So we have to keep down our.

480
Pernilla Bergmark Ericsson 01:57:14.674 --> 01:57:35.734
tricity consumption as much as possible, so we don't waste that entity. So I
think there are a lot of reasons to to look within our own boundaries. So I
think these are 2 different discussions. W. W. W. W. W. how large are we, uh,
but we, that should not be confused with, uh, discussions, whether, or not to,
to, to review.

481
Pernilla Bergmark Ericsson 01:57:35.739 --> 01:57:37.744
These are emissions because that we must do.

482
Colin Perkins 01:57:39.995 --> 01:57:41.465
Absolutely, absolutely.

483
Colin Perkins 01:57:43.504 --> 01:57:45.934
All right, so we, we in a few minutes left, um.

484
Colin Perkins 01:57:47.134 --> 01:57:58.504
To finish up, um, firstly a couple of minutes on what, what are we missing? W.
W. W what are the areas where people think we're missing something fundamental?
What are the risks and concerns.

485
Marisol Palmero Amador 01:58:01.564 --> 01:58:20.524
Calling it's here, I just wanted to, uh, make a note, right as I have not seen,
um, probably I have been using it as well but, um, I have not seen, uh, the the
point being raised and it is something that we don't know yet and it's coming
our way for.

486
Marisol Palmero Amador 01:58:20.914 --> 01:58:42.004
Organization under the European Commission, we are, uh, there is a conversation
going through the digital passport, and the passport will, um, affect any
product that any vendor might put outside. Right? And it will consider not only.

487
Marisol Palmero Amador 01:58:42.035 --> 01:59:02.855
That book has been built something that, eh, manufacturing should be labeling 1
way or another, but as well, eh, the lifecycle of the right? They all at least
related to scope 3 to the use of the approach. Um, that should be updated. Eh
I'm not. Sure, if realtime or not.

488
Marisol Palmero Amador 01:59:03.455 --> 01:59:23.105
Things needs to be defined, but something, I guess this community can help, uh,
to, uh, provide a more insight information to the European Commission on any
governmental, eh organization, right as this will be, uh, impulse by countries.

489
Marisol Palmero Amador 01:59:24.309 --> 01:59:25.324
That's my understanding.

490
Marisol Palmero Amador 01:59:27.215 --> 01:59:31.475
It is something that we should be looking at as well how to influence some of
their.

491
Colin Perkins 01:59:33.124 --> 01:59:44.974
Yeah, it's it's not an area I know a whole lot about, but I agree. It sounds
like something we should be thinking about. Um, can you maybe send something to
the to the list with some more ideas and more information.

492
Marisol Palmero Amador 01:59:46.144 --> 01:59:47.284
Sure.

493
Colin Perkins 01:59:48.094 --> 01:59:53.164
Okay, thank you. And if anyone has any other inputs on that, that would be
useful.

494
Colin Perkins 02:00:02.884 --> 02:00:06.664
Okay, any other risks and concerns people to raise.

495
Colin Perkins 02:00:14.194 --> 02:00:16.504
In that case yes, Daniel. Sorry.

496
Daniel Schien 02:00:16.954 --> 02:00:17.344
Um.

497
Daniel Schien 02:00:17.975 --> 02:00:21.575
Coming back to the initial point, Colin, you made.

498
Daniel Schien 02:00:24.515 --> 02:00:40.715
The bonds for more data, um, that we'd like to have, uh, there may be less
constructive dialogue than demands for the data that we need. So, asking for
justification, why we need. Uh, so I think he's interested here in your comment.

499
Daniel Schien 02:00:42.064 --> 02:01:02.974
I wanted to say that, um, at the moment. So I believe that if there was a
political bill to decarbonise all the sectors in the economy, um, without any
compromises yeah. Then a lot more could be done.

500
Daniel Schien 02:01:03.035 --> 02:01:24.065
We are currently even considering the conversation here is all around
efficiency gains that could be realized without impacting user experience and
we're also making some implicit assumptions about cost of activity. So that's
where we are at the moment.

501
Daniel Schien 02:01:24.455 --> 02:01:44.525
We are, uh, in a process of gathering data that is then provided to the
discourse that we are not currently engaged them. Um, that could then make, uh,
uh, calls about sector, reducing an absolute.

502
Daniel Schien 02:01:45.544 --> 02:01:47.554
Lots of conversation that we're having at the moment.

503
Colin Perkins 02:01:48.784 --> 02:01:48.994
Yeah.

504
Daniel Schien 02:01:49.564 --> 02:02:06.454
Um, but it, but I think this, um, yeah, so so so from that, I think it's very
difficult to know what is the data that we need, because we don't currently
know what, what's our, the, the interventions.

505
Daniel Schien 02:02:06.844 --> 02:02:09.124
That might be on the table um.

506
Daniel Schien 02:02:11.134 --> 02:02:11.974
7 years.

507
Colin Perkins 02:02:14.375 --> 02:02:34.655
Yeah, so sounds like 1 of the, I mean, the, the last slide was communication
and outreach. It sounds like 1 of the things we should be doing is having a
broader conversation about whether we are trying to optimize for keeping the
same set of services or how are we changing the way people use.

508
Colin Perkins 02:02:34.685 --> 02:02:43.115
The network, and how it regulates is changing the people that the way people
use the network to try and force down usage and how we can be part of that
conversation.

509
Colin Perkins 02:02:51.424 --> 02:02:59.764
All right, uh, I'm conscious that way of getting very short on time. Um, does
anyone have any concluding remarks? I can find things to say.

510
Eve Schooler 02:03:02.284 --> 02:03:12.904
To hope that someone is capturing the chat windows and we'll share the chat
window comments and pointers because they're it's a wonderful conversation
going on on the sidelines.

511
Colin Perkins 02:03:14.854 --> 02:03:18.814
Certainly true Cynthia. We were recording the chats.

512
Suresh Krishnan 02:03:18.814 --> 02:03:21.364
I think Cindy sent it for last time. So I think she is.

513
Eve Schooler 02:03:22.715 --> 02:03:24.215
I missed that 1. okay. Thank you.

514
Cedric Westphal 02:03:25.419 --> 02:03:29.134
Didn't see it for the last time I saw the transcript of the session, but in the
chat.

515
Cindy Morgan 02:03:30.664 --> 02:03:38.974
I'm not actually sure whether the chat is captured in the video recording, but
I can make a copy of the chat before I close the meeting.

516
Colin Perkins 02:03:41.824 --> 02:03:42.244
Thank you.

517
Colin Perkins 02:03:45.275 --> 02:03:50.015
All right, well, that I think we're about done for the day. Uh, Eric, do you
have anything to finish up with.

518
Jari Arkko 02:03:52.234 --> 02:04:13.354
No, I, I, I, thank you all, it's been very active discussion, different
viewpoints and lots of data, but also, I guess, um, some missing data. Um, and
I think that's that's 1 of the things that we need to talk about. Also,
tomorrow, if we can improve on that aspect as well, not just, uh, 1st,
understand what needs to be improved.

519
Jari Arkko 02:04:13.384 --> 02:04:34.384
And then, and, um, figure out the improvements and yes. Um, I was trying to
write some high level, very high level conclusions from the session. And that I
came up with. We have some understanding of the environmental impacts at least
some bracket thing. Like, this is roughly what we're talking about and and most
people seem to be selling at least from that perspective.

520
Jari Arkko 02:04:34.565 --> 02:04:55.265
A similar story, um, also seem to have agreement that some most alarming
stories may not actually be factual. Um, but then, of course, still and
Internet, uh, uses, you know, fair amount of energy and any savings. There
would actually be material like, um.

521
Jari Arkko 02:04:56.225 --> 02:05:16.055
Size of Norway even so we should probably pursue that. And there's, there's
probably if we needed to even sort of maintain the status quo, because the
engineers had to work hard to, you know, keep the levels where they are today.
And and we need to do that in the future as well.

522
Jari Arkko 02:05:16.894 --> 02:05:28.294
In part on us, and also, um, it may actually be even hard in the future if more
flow it's, it's more difficult to follow just Moore's law and and, um, rely on
that.

523
Jari Arkko 02:05:29.525 --> 02:05:30.635
And that's that's what I.

524
Colin Perkins 02:05:30.965 --> 02:05:36.905
And it's really not clear that sustaining at the current that keeping things at
the current usage is sustainable.

525
Colin Perkins 02:05:38.404 --> 02:05:59.104
Yep. All right. Thank you. Everyone this, this has been a fascinating
discussion. Um, we reconvene at the same time tomorrow. Uh, and, uh, what is
the discussion tomorrow is, um, improvements what's called improvements,
implementation, improvements and incentives. So I hope so.

526
Colin Perkins 02:05:59.134 --> 02:06:03.004
Again, tomorrow, thank you everyone.

527
Jari Arkko 02:06:03.004 --> 02:06:03.334
You.

528
Suresh Krishnan 02:06:05.375 --> 02:06:05.675
So.

529
Pernilla Bergmark Ericsson 02:06:05.705 --> 02:06:07.055
It's all very interesting.

530
Cedric Westphal 02:06:09.124 --> 02:06:09.544
Or is it.