►
From YouTube: IAB workshop on Environmental Impact of Internet Applications and Systems 2: What do we know
Description
IAB workshop on Environmental Impact of Internet Applications and Systems 2: What do we know?
Workshop webpage: https://datatracker.ietf.org/group/eimpactws/about/
Papers in GitHub: https://github.com/intarchboard/e-impact-workshop-public
Session 1 (The Big Picture): https://youtu.be/90GxlL34rQ4
Session 2 (What Do We Know?): https://youtu.be/EaNgREHLXRg
Session 3 (Improvements): https://youtu.be/jjEZwuuChZc
Session 4 (Next Steps): https://youtu.be/Pc_XY5sDR58
A
All
right,
so
the
the
rates
at
which
people
are
joining
seems
to
have
slowed
down
so
I
I
guess
we
should
get
started.
Hopefully
you
can
all
hear
me
fine.
This
is
the
the
second
session
in
the
internet
architecture
boards
workshop
on
the
environmental
impacts
of
Internet
applications
of
the
systems
and
I'm
Colin
Perkins,
and
one
of
the
the
group
who's
organizing
this
Workshop,
along
with
Yari
and
and
and
above
others,.
B
A
This
meeting
please
try
to
be
polite.
Please
try
to
explain
your
Viewpoint
in
an
understandable
manner.
Please
try
to
learn
from
other
people's
viewpoints.
We've
got
quite
a
diverse
group
of
people
here,
people
with
a
quite
a
wide
range
of
backgrounds,
some
people
from
the
academic
Community,
some
from
industry,
some
who
have
been
heavily
involved
in
the
ATF,
but
for
many
years
some
who
haven't
some
from
it
I
think
perhaps
a
wider
range
of
backgrounds
than
that
there's
a
risk
that
we
talk
past
each
other.
A
A
A
What
we're
focusing
on
today
is
what
we
know
for
transit
to
learn
what
we
understand
and
what
we
don't
understand
about
the
environment,
environmental
impact
of
the
network,
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
and
also
we're
here
to
try
and
think
about?
A
You
know
who
we
might
need
to
talk
to
who
we
might
need
to
connect
with
that?
We're
not
already
talking
to
to
get
access
to
information,
that's
needed,
and
if
there
are
areas
where
we
we
don't
necessarily
have
the
the
expertise
we
need
and
where
we
need
to,
as
I
say,
form
connections
and
try
and
enhance
our
understanding.
A
This
will
feed
into
the
session
tomorrow.
At
the
same
time
tomorrow,
which
is
looking
at
protocol
improvements,
implementation
improvements
and
concrete
actions
we
can
be
taking
in
the
session
on
Monday
next
week,
which
is
starting
to
look
at
the
future
developments
and
the
plans
and
what
are
the
next
steps?
A
Foreign,
the
the
goals
for
today
as
I
say,
are
to
understand
what
we
we
do
and
do
not
understand
about
the
situation
and
I
I
guess
the
phrase
known
unknowns
and
known
moons
and
so
on
will
come
into
this
discussion
at
some
point,
one
one
thing
I
did
want
to
highlight
from
vesnar's
talk
in
on
Monday.
There
was
a
suggestion
to
put
efforts
into
the
actual
actions.
A
You
know
it's
easy
to
talk
about
what
what
data
we
would
like
to
know
what
measurements
we
would
like
to
take
and
it's
it's
easy
to
continually
focus
on,
measuring
and
understanding
and
improving
that
understanding,
and
that's
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
right,
let's,
let's
focus
on
how
to
achieve
concrete
action.
It's
not
just
how
to
improve
our
understanding
of
the
situation.
A
A
C
All
right,
thank
you.
So
this
is
about
the
internet.
Energy
usage,
not
all
of
ict's
energy
is
used,
which
is
supposedly
a
bit
smaller,
but
maybe
well
depends
and
a
little
bit
about
misinformation.
People
have
already
done
some
spoilering
actually
on
the
mailing
list
or
regarding
some
things.
I
was
going
to
talk
about
here.
I'm
doing
my
you
know,
making
an
effort
to
give
a
very,
very
precise
overview
of
magnitude
relationships
exactly
like
that.
You're
that
tiny
spot
your
own
footprint.
This
could
be
a
city.
C
This
could
be
a
country,
this
could
be
the
internet
and
that's
the
internet
or
all
of
ICT,
depending
on
stupid
claims.
Now
we're
gonna
get
back
to
these
relationships,
but
actually
the
point
is
that
these
are
maybe
not
completely
off
we'll
see
so
I'll
begin
with
nonsense
just
to
get
it
done
to
get
it
out
of
the
way.
C
So
Mike
bonusly
the
brother
of
Tim
bernersley.
He
has
been
talking
to
the
Press
apparently
about
this
issue.
Quite
a
bit
and
I
just
came
across
these
press
statements
some
time
ago.
Actually,
as
for
a
presentation
I
gave
for
students
where
I
was
trying
to
make
the
point
that
there
is
just
lots
and
lots
and
lots
of
nonsense
in
the
press
that
we
don't
need
to
believe.
You
know
giving
very
concrete,
precise
numbers
about
an
email
that,
to
begin
with
is
maybe
idiotic,
just
as
an
idea,
I
would
say.
C
C
C
I
guess
the
main
point
of
this
slide
is
that
there
are
many
URLs
here
right.
You
see
all
these
statements
coming
from
various
sources
and
it
goes
on
here.
It
says
the
typical
office
worker
sends
and
receives
so
many
emails
per
day.
You
know
what's
a
typical
number
anyway
here,
but
over
the
course
of
the
Year.
C
This
is
as
much
as
flying
from
London
to
Bruges,
where
I
don't
even
think,
there's
a
direct
flight
to
begin
with,
but
it's
watching
955
movies,
all
that
kind
of
stuff
really
simply
put
you,
can
try
and
calculate
these
things
and
compare
them,
but
trust
me
I
did
actually
try,
and
there
is
just
nothing
here
that
makes
any
sense
at
all.
All
these
numbers
don't
match
so
I'll.
Just
nonsense.
C
There's
also
this
article
saying
that
climate
change
is
going
to
make
us
more
stupid,
I
just
leave
that
here
without
say
more
about
it.
The
reason
for
me
to
talk
about
this
is
that
I
do
agree
with
also
Jonathan
Comey,
who
has
said
that
this
is
a
harmful
thing.
People
go
out
saying
these
things
to
the
press.
It
gets
spread
and
it
may
actually
have
a
negative
effect
for
the
cost
that
we
are
having
here,
which
is
to
consider
the
environmental
impact
of
the
internet.
C
It
may
cause
some
people
not
to
take
these
whole.
This
whole
kind
of
effort,
serious,
and
especially
when
these
people
work
at
funding
bodies.
Then
this
is
not
a
good
thing.
We
can
also
consider
bundling
this
law
here,
which
is
also
called
the
asymmetry
principle.
Basically,
the
fact
that
it
it
takes
an
order
of
magnitude,
more
effort
to
refute
than
to
put
it
out
there.
C
Jonathan
Comey
has
done
a
good
job
at
quoting
or
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
I
think
it's
interesting
reading.
Indeed,
I'm
done
with
this
already.
Let's
look
at
reality.
C
My
take
is
that
if
we
can
come
with
some
conservative
estimates
that
are
possibly
reasonably
realistic
and
possibly
go
out
to
the
Press
with
them
with
them,
then
we
may
be
able
to.
You
know
counter
these
things
a
little
bit,
and
maybe
this
is
a
better
better
thing
than
having
these
crazy
exaggerations.
C
Now,
what
are
these
estimates?
I
have
made
a
start
with
this
table
that
our
IAB
submission
has
and
what
I
did
is.
I
didn't
really
try
to
well
try
to
gather
data
myself,
but
I
just
looked
at
reports
that
others
have
made
now
when
you
look
at
reports
actually,
you
know
we
have
already
discussed
this
I
think
to
some
degree
in
the
first
meeting
that
there
is
no
even
no
notion
of
the
internet.
What
is
the
internet?
C
You
know
it
considers
it
consists
of
of
devices
the
wired
part,
the
wireless
part,
devices
or
customer
premises,
equipment
user
equipment.
There
is
the
embodied
energy
of
devices
which
is
from
constructing
the
thing
that
we're
using
there
are
data
centers,
which
you
know
may
maybe
well
yeah
things
may
differ
quite
a.
D
A
C
Study
is
also
different
in
age,
so
what
I
did
is
I
just
tried
to
find
a
couple
of
sources
and
from
these
sources,
I
arrived
at
a
range
of
of
0.5
percent
to
1.17
percent
of
the
global
CO2
footprint.
Now
my
point
here
is
not
that
this
is
the
number.
My
point
is
that
it's
probably
not
10-
it's
probably
not
0.001
percent,
but
it's
probably
something
in
that
rough
magnitude.
C
One
possible
derivation,
just
to
make
this
a
bit
concrete,
is
that
you
can
take
the
smarter
2030
report.
That's
been
cited
a
lot
which
states
that
ICT
in
general
has
a
CO2
footprint
of
2.7
percent
of
global
emissions
in
2020.,
and
there
are
numbers
from
2012
that
say:
Telecom
electricity
is
IC
with
a
third
of
ICT.
C
C
Is
this
a
small
number
0.9
doesn't
exactly
sound,
very
big,
but
you
know
to
begin
with.
The
internet.
Energy
usage
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,
because
that
number
of
0.9
percent
is
actually
comparable
to
the
CO2
emissions
of
the
aviation
industry.
But
then
in
aviation
we
also
have
methan
and
other
greenhouse
gases
playing
roles.
C
If
you
take
those
into
account,
then
the
number
of
the
aviation
industry
is
significantly
higher
and
then
I've
recently
read
something
that
that
apparently
a
quite
recent
study
that
finds
that
the
influence
of
water
vapor
is
actually
quite
significant,
and
this
water
vapor
up
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
three
percent.
C
But
anyway,
you
know
to
get
an
idea
of
the
of
the
magnitude
here.
According
to
this
page,
the
20
to
21
emissions
of
the
UK
were
0.93.
So
that's
a
comparable
number
Norway
where
I
live,
has
0.11
percent
here
in
Norway,
that's
a
strong
push
to
for
everybody
to
reduce
the
personal
CO2
footprint.
There
are
people
going
to
Great
Lengths.
C
To
do
this.
There's
a
public
press
offering
calculator.
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
make
more
people
join
this,
but
I
think
it's
important
to
put
things
into
perspective
a
little
bit.
The
population
here
is
5.4
million,
so
roughly
the
per
person
contribution
is
0.902
percent.
C
It's
tiny!
Now
this!
These
are
the
people
here,
it's
also
a
small
fraction
of
the
world
population,
but
it
is
interesting
to
consider
that
if
we
could
with
a
standard
somehow
we
use
the
internet's
power
Powers
here
by
10
I
mean
even
if
you
think
of
one
percent
right,
maybe
that's
more
realistic.
Even
then
this
is
this
is
a
big
number
compared
to
you
know
a
couple
of
a
couple
of
people
now,
ten
percent.
That
would
be.
C
You
know
again
for
going
from
0.9
to
0.81,
which
is
4.5
million
Norwegians,
and
you
can
imagine
them
all.
You
know
varying
Viking
helmets
and
things
so
many
okay
I'm
stopping
here
it
does
get
a
bit
silly.
But
the
point
is
that
the
potential
of
a
standard
that
has
an
impact,
the
global
impact
on
the
Internet
is
truly
enormous,
and
it's
going
to
be
enormous.
You
know
whether
this
is
one
percent,
two
percent,
three
percent:
zero
point:
half
percent,
it's
very
big!
C
Getting
back
to
the
picture
from
the
beginning,
I,
don't
think
we're
so
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
and
the
urinal
snapped
on
I.
Don't
know
these
could
be
potential
savings
from
iitf
standards.
Who
knows
you
know
time
a
bit
smaller
than
that?
But
you
know
it's
about
this
General
General,
very
broad
relationships,
there's
already
everything
I
had
to
say:
that's
it.
D
C
A
A
A
D
A
E
F
C
Simply
you
know,
if
you
could
just
imagine
reducing
I
mean
now
we
usually
send
approximately
an
act
for
every
other
packet,
so
I've
been
I've
been
just
wondering
you
know.
If
we
would
take
this
down
to
an
act
for
every
quarter,
I
mean
every
fourth
packet
you
know:
would
that
have
a
significant
influence
on
the
whole
internet?
C
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
multicast.
I
find
this
very
interesting.
This
paper
that
was
in
there
I
think
your
own
paper
had
interesting
suggestions
about
routing
along
well,
not
just
energy
saving,
but
depending
on
carbon
efficiency.
F
Different,
it's
not
yellow
anymore.
It's
confusing
me
how
it
is
anyway,
so
one
sort
of
piece
of
feedback-
maybe
I
mean
you.
You
showed
these
nonsense,
grab,
metrics
or
measurements,
whether
we
reported
and
like
many
others
that
I've
seen
they
all
they
they
throw
together.
They
bundle
up
like
the
networking
stuff
and
the
application,
and
they
say
you
know
it's
per
email
right
and
I
actually
find
that
a
bit
confusing
right
and
it
sort
of
muddles
the
waters.
F
Because
if
you
look
at
the
data
center
or
you
look
at
your
Leo
but
the
stuff
that
burns
energy
right
is
the
stuff
that
gets
hot
and
that's
usually
the
CPUs
and
the
gpus,
it's
not
so
much
the
network
right
and
so
obviously,
that's
also
well,
not
obviously,
but
I
I
assume.
That
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.
F
But
it's
in
it's
getting
more
efficient,
since
the
switch
shifts
are
getting
better
and
so
I
wonder.
If
there's
anybody
who
sort
of
tries
to
tease
that
apart
and
tries
to
talk
about,
you
know
Power
users
CO2
emissions,
the
different
layers,
if
you
will
right
at
the
application
layer
and
then
for
the
communication
part
and
for
other
aspects
or
is
it
all
always
reported?
As
you
know,
this
big
lump
thing.
B
C
I
mean
this,
of
course.
Obviously
these
nonsense,
Studies
have
lumped
everything
together.
There
are
ways
to
get
CPU
and
memory
with
our
APL.
I
was
wondering
if
there
are
apis.
That
may
give
something
like
that
in
software,
because
that
will
be
actually
especially
interesting
for
for
NYX
I,
don't
know
of
any.
A
H
So
one
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,
we
had
some
kind
of
pdu
right
like
from
the
data
center
like
so
what
you
can
do
is
we
can
go
to
the
Reddit
and
pdu
and
see
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've
at
least
done.
Something
like
that.
C
D
Something
worth
noting
is
that
all
Network
equipment
is
on
the
order
of
one
tenth
of
All
Electronics
energy
use.
All
Electronics
are
not
internet
connected,
but
an
increasing
fraction
is
so
just
that's
a
worth.
A
point
worth
taking
and
look
at
electronics
are
about
14
well.
This
was
from
eight
years
ago
of
all
buildings,
energy
use,
so
there's
roughing
order
of
magnitude
from
Network
equipment
to
electronics
and
roughly
an
order
of
magnitude
from
Electronics
to
all
electricity
and
buildings.
A
A
B
Yeah,
it
was
you,
you
said
something,
and
thank
you
for
your
presentation
around
the
opportunity
of
of
less
frequent
signaling,
let's
say
and
of
course,
the
experience
from
from
the
5G
development,
where
signaling
has
been
significantly
less
frequent
and
what
it
enables
in
terms
of
sleep
mode
is
something
that
is
worth
considering.
I
also
think
from
for
other
systems
and
so
on,
and
then
also
in
terms
of
standardization.
B
You
mentioned
the
so
I'm
actually
in
itu,
what
kind
of
Standards
we
would
need
so
I
think
in
terms
of
specification
how
to
calculate
the
ICT
footprint.
That
is
a
standard
in
place,
which
is
quite
detailed,
but
then,
of
course,
some
studies
look
at
ICT.
Some
look
at
digital
Technologies,
which
is
not
so
well
defined.
Some
look
at
the
internet
and
how
do
we
Define
Internet?
Is
it
from
an
end-to-end
perspective?
B
A
D
A
I
B
D
J
J
So
my
name
is
Daniel
Sheen
I
work
at
the
University
of
Bristol
in
the
UK
and
present
work
that
our
group
here
has
has
carried
out
over
the
last
10
years
or
so,
including
Professor
Chris
priest
who
I
believe
may
or
may
not
be
in
the
audience
and
Paul
shabbaji
and
I've
seen
his
name
in
the
audience.
But
we
work
with
many
of
you
here.
J
So
this
is
it's
really
fantastic
to
meet
to
meet
the
people
we
already
work
with
and
meet
so
many
new
faces
and
names
and
the
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
yeah,
and
we
look
forward
to
hearing
what
you
have
to
say
critically
about
the
ideas
that
that
we're
sharing.
J
J
J
J
Now,
with
the
caveat
that
there's
a
there's,
a
strong
focus
on
news
phase,
electricity
consumption
for
various
reasons,
but
in
terms
of
the
system
that
we
want
to
take
into
account,
we
we
try
to
be
inclusive
and
include
data,
centers
networks,
wired
and
cellular
networks,
and
then
the
user
devices
and
that's
important
in
order
to
to
prevent
burden
shifting,
for
example,
and
to
provide
a
complete
and
consistent
understanding
of
the
environmental
impacts
of
of
Media.
Services.
J
Great
here
is
an
example
of
of
the
of
the
process.
Such
an
assessment
takes
into
account
all
of
the
all
of
the
processes
that
are
involved
and,
and
there
can
be
many
and
and
such
an
assessment
usually
starts
with
with
creating
such
a
map
and
then
for
each
of
the
boxes,
identify
how
the
underlying
infrastructure
consumes
electricity
in
proportion
to
the
service.
And
then
there
are
some
technical
steps
that
that
are
required
when
processes
are.
J
Okay,
we
also
worked
with
an
organization
called
constone
to
construct
a
tool
that
is
more
self-service,
so
they
work
with
the
BBC
and
the
guardian
earlier
was
very
bespoke
and
now
we're
living
in
an
age
where
all
media
organizations
or
all
organizations
in
the
economy
are
concerned
about
their
carbon
footprint
and
there's
a
big
need
to
to
carry
out
Baseline
assessments
and
then
set
targets
for
decarbonization
and
for
the
media
sector.
The
dimpec
tool
is,
is
probably
the
most
popular
at
the
moment
with
many
organizations.
J
There's
been
some
exchange
of
emails
already
around
the
carbon
footprint
of
streaming.
I
believe
I'm.
Very
convinced
that
at
the
moment,
the
most
robust
that
is
out
there
is
the
one
that
is
based,
partly
on
the
on
the
dimpec
model
published
in
2021
by
the
carbon
trust
and
against
my
modin,
for
example.
J
J
We
can
see
here
that
very
clearly,
if
we're
just
looking
at
the
the
use
phase
which,
which
is
the
the
the
carbon
intensity
of
electricity
results
in
in
a
proportional
variation
of
the
overall
carbon
footprint,
of
course,
but
the
the
decarbonization
of
the
electricity
will
also
result
in
a
reduction
of
the
embodied
impact.
So
so
this
is
a
one
key
lever
that
we
need
to
take
into
account
when
we're
thinking
about
initiatives
to
to
reduce
environmental
impact.
J
Okay,
so
so
these
are,
some
results,
gives
us
a
perspective.
This
service
based
perspective
I
think
it's
useful
for
a
variety
of
ways.
It
is
useful
for
the
organizations
providing
those
services
to
to
think
about
Footprints,
or
so
the
users
of
the
impact
of
the
dimpack
tool.
For
example,
they
get
this
perspective
for
the
entire
audience,
for
the
enter
the
entire
service
that
they
provide,
and
then
they
can
set
targets
on
these
larger
system
boundaries,
but
for
a
consumer,
it's
also
relevant
to
put
into
perspective.
J
J
J
We
start
with
this
understanding
of
what
the
infrastructure
looks
like.
This
is
an
old
slide,
but
in
principle
I
think
it's
still
is
representative
of
how
networks
work.
You've
got
an
electronic
routing
layer,
then
you've
got
an
optical
transport
layer
underneath
and-
and
we
need
to
take
into
account
the
efficiency
of
of
sending
data
across
the
system.
J
With
this
understanding,
we
can
then
produce
estimates
of
of
energy
intensity
and
when
Yen
sees
this
I
know
he'll
think
oh,
this
is
too
simple
and,
of
course,
I'm.
I'm
I
fully
support
that
point
and
then
I'll
provide
something
that
is
a
bit
more
complicated
and
nuanced
in
the
moment.
But
that
is
the
state
of
the
art.
What
we've
got
here
and
a
metric
that
I've
termed
IV
here
for
intensity
of
of
data
volume
and
and
multiplied
with
an
amount
of
of
data
transported
over
a
network.
J
We
are
producing
a
footprint.
We
are
getting
a
number
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.
So
this
is
this:
is
the
state
of
the
art?
Excuse
me
what
I've
got
here
is
this
exponentially
decreasing
line
where
how
very
consistently
these
average
energy
intensities
for
electricity
have
have
have
have
been
produced
over
time?
J
So
this
is
this
is
very
relevant
and
then
on
the
one
and
then,
on
the
other
hand,
we've
got
we've
got
data
that
has
been
broken
down
from
from
Lewis
cook.
Who
is
here
in
the
audience
as
well,
so
there
there
are
different
approaches
to
these.
The
numbers
are
slightly
varying,
but
in
in
sort
of
in
the
order
of
magnitudes
they're
comparable.
J
J
J
The
upshot
of
that
is
that
the
energy
consumption
for
the
network
is
very
insensitive
to
to
traffic
and
I
like
to
point
to
this
news.
Article
from
gsma
I've
been
criticized
for
it,
but
I
think
gsma
doesn't
doesn't
pick
up
on
the
irony
of
it.
So
in
the
article
They
are
promoting
that
the
networks,
the
network
operators
didn't
see
a
change
in
the
energy
consumption
of
the
networks.
J
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
Behavior.
So
there's
an
article
here,
I
think
Kumi
and
masane
are
criticizing
this
in
the
in
the
jewel
paper.
If
not
they,
they
would
have
agreed
with
me
where
this
is
just
one
case
of
many
and
I
and
our
own
work.
J
We've
Fallen
victim
of
this
reductionist,
a
simplistic
assumption
about
the
the
way
that
the
networks
work,
but
here
they're
saying
if
we
would
turn
off
our
video
cameras
in
in
calls,
we
would
be
able
to
reduce
the
carbon
footprint
from
9.4
kilogram
of
CO2
per
month
to
377
grams,
and
sadly
that
is
not
the
case.
In
particular,
it
won't
result
in
an
instantaneous
energy
consumption
change
in
the
network.
J
J
However,
what
we
and
then
multiply
the
base
power
consumption
over
time
now
that
the
the
problem
with
that
is
that
it,
it's
ignores
a
fundamental,
the
fundamental
Dynamic
of
how
infrastructure
is
operated
now,
I'm
I'm
I'm,
not
operating
these
systems
myself,
but
from
conversations
with
good
colleagues
and
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
better
user
quality,
better
user
experience
and
users.
Looking
are
looking
for
this.
J
They
want
to
have
HD
video
calls
and
they
want
to
have
8K
television,
probably
and-
and
this
has
been
going
on
as
far
as
long
as
we
are
measuring
the
energy
consumption-
and
you
know
the
data
volume
in
the
internet
and
and
the
the
volume
of
traffic
in
the
network
has
very
consistently
doubled
every
two
years
and
that
drives
the
infrastructure
capacity,
and
this
is
just
a
given.
It's
extremely
constant.
J
J
The
Baseline
power
consumption
here
can
actually
go
down
as
well,
but
in
practice-
and
this
is
again
open
for
debate-
the
energy
consumption
of
new
kit-
it
the
the
capacity
of
the
network
increases
and
if
the
total
demand
grows
faster,
then
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.
J
The
Energy
Efficiency
Improvement
of
the
equipment
at
the
moment
is
what
we're
thinking
around
in
the
order
of
20
to
25
percent,
but
we
know
that
the
demand
is
increasing
in
the
order
of
50
per
year.
So
just
from
looking
from
combining
those
two
factors
it
unless
we're
missing
something
it's
impossible,
that
the
power
consumption
of
the
network
will
remain
constant,
yeah
now.
J
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
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
where
we
don't
care
what
time
of
the
day
or
the
year
or
when
you're
using
this.
J
J
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
energy
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
I'm,
just
taking
into
account
the
regular
diural
pattern
that
we
observe.
J
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
and
and
there's
so
thus
a
penalty
for
traffic,
the
closer
that
it
comes
to
peak
time
use.
Okay,
that's
the
proposition
here.
J
J
But
we
end
up
with
a
metric
that
has
a
normative
effect,
because
it
is
incentivizes
the
users,
as
well
as
the
providers
of
services,
to
think
about
ways
to
carry
out
demand,
smoothing
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.
A
K
I
need
to
drop
for
another
meeting,
but
I
was
going
to
make
a
quick
comment.
I
think
sort
of
brakes
told
us
in
the
chat
here
is
that
looking
at
some
of
the
paragraph
shown
earlier,
it
may
be
that,
what's
in
the
eye
test
core
competency
in
terms
of
routine
protocols
in
the
core
Network
and
things
like
that-
would
have
very
little
impact
on
the
opal
internet
usage.
K
If
a
lot
of
the
power
has
been
consumed
on
the
other
devices
and
in
the
home,
so
I
still
think
it's
it's
clear
that
we
need
to
understand
or
have
good
data
for
exactly
words.
Power
has
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
optimizations.
We
make
because
you
could
optimize
the
core
routing
and
find
that
you
actually
make
like
one
percent
less
oval
difference.
A
B
K
B
J
That's
my
way
of
thinking.
Yes,
yes,
sorry
I
totally
agree
with
that
that
that
it
is
ultimately
you're
interested
time,
isn't
isn't
so
important.
I
think
the
two
are
related
yeah,
because
we've
got
that
2x
increase
in
demand
over
time.
So
so
certainly
the
two
are
related,
but
ultimately
the
main
thing
that
you're
concerned
with
is
is
is
load
yeah
I
do
I
do
agree
with
that.
Okay,.
L
Yeah
high
fives
are
done
excellent
presentation.
Of
course,
I
I
love
the
now
Chris
and
step
graph.
That's
a
fantastic
way
to
visualize
the
upgrade
both
question
for
you
on
that
do.
Did
you
include
sort
of
application
layer
services
like
CDN
in
there
or
is
that
is?
Is
this
tight
already
to
specific
Network
infrastructure
or
have
you
got
cloud
services
application,
layer
stuff?
It's
attracted
into
any
particular
workflow
representative.
J
J
I
think
there's,
there's
more
work
to
be
done
to
to
look
at
how
this
would
plan
play
out
with
a
server
equipment
that
it
is
they're
they're,
now
more
energy
proportional
we're
talking.
Maybe
40
percent
base
power
consumption
rather
than
80.
The
principle
is
the
same
yeah.
Now
your
the
your
the
gains
in
terms
of
the
the
the
penalty
for
for
Peak
usage.
They
would,
they
would
be
smaller,
yeah
cool.
A
A
I
I
Enough,
oh
maybe
you
can
just
take
that
if
you
can
see
all
done,
it's
fine
by
me
so
short
presentation
now
for
me:
yes,
Mom
Dean
Erickson,
we
talked
a
lot
about
the
internet
or
the
ICT
sector,
and
the
title
I
wrote
here
was
internet,
consumes
less
energy
and
emit
less
carbon
than
what
many
think
this
goes
along
with
what
was
said
in
the
first
presentation
and
unfortunately,
you
can
find
other
information
about
the
internet
on
the
internet.
So
you
can
go
to
the
next
slide,
just
some
examples
from
BBC
Guardian.
I
So
usually
they
talk
about
internet
emitting,
more
common,
the
aviation
sector
stream
media
video
is
especially
consuming
a
lot
of
energy.
The
sector
grows
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
three
there,
so
this
is
the
kind
of
comparison
I
would
like
to
make
if
we
take
the
aviation
sector.
I
So
this
is
results
from
another
Swedish
researcher,
about
the
aviation
sector,
so
in
2019
about
four
percent
of
global
GSK
and
as
you
can
see
to
the
left
there,
so
it's
not
only
a
CO2.
We
have
also
other
emissions
and
effects,
especially
the
High
Altitude
effects,
methane
knocks
and
so
on,
and
then
we
need
to
produce
the
yet
fuel
and
we
should
also
look
into
okay.
What
what
does
it
take
to
construct
all
the
airports
and
aircraft
and
so
on?
Unfortunately,
nobody
has
done
that
research,
but
I
think
it's
pretty
large.
I
I
I
And
then,
if
we
look
at
Aviation
sector,
we
actually
see
that
it's
only
one
percent
of
all
people
they
do
about
half
of
all
flying
and
only
slightly
more
than
10
fly
a
given
year.
So
one
percent
of
all
people,
their
air
travel-
is
actually
emit
more
CO2
than
the
entire
ICT
sector,
which
is
used
by
I,
guess
we
can
say
around
5
billion
people
daily
for
hours.
I
I
You
have
the
embolded
footprint
for
data,
set
the
networks
and
used
devices
and
operation
on
top
there,
and
when
it
comes
to
how
to
measure
these
parts,
I
think
we
need
to
go
to
all
the
companies
and
today
the
good
story
is
that
most
companies
actually
report
their
electricity
consumption
and
their
emissions,
and
so
on.
So
for
data
centers.
I
We
have
data
from
46
46,
large
companies,
whether
it's
Google
Microsoft
and
all
the
others
about
60
operational
data
centers,
but
these
companies
actually
represent
more
than
90
of
all
data
traffic
for
Network
operators,
67
large
operators,
eighty
percent
of
all
the
subscriptions
in
the
world,
but
then
we
need
to
add
an
estimate
for
the
Enterprise
Networks
officers,
Smalls,
airports
and
so
on.
For
the
important
part
we
have
done
a
separate
start
where
we
used
data
from
59
large
manufacturers.
I
I
But
today
it's
most
of
that
function
is
within
smartphones
that
we
don't
buy
those
products
anymore
so
that
that's
actually
how
the
reducing
impact
on
the
total
footprint
paper
media,
and
then
you
have
some
newcomers,
cryptocurrency
security
cameras
and
so
on,
but
still
rather
small.
You
can
go
to
the
next
slide.
I
So
if
we
look
back
to
2010
and
compare
to
this
new
2020
style
that
we
see
actually
a
total
reduction
based
on
it's
actually
less
TVs
and
TVs
pieces
and
TV
so
2020
compared
to
2010
2011.,
we
have
the
smartphone
impact.
I
mentioned
a
lot
of
functions.
You
need
the
dedicated
devices
to
before
us.
Cameras
are
now
included
in
a
smartphone
for
pieces.
We
mainly
use
laptops
smaller
much
much
more
energy
efficient
than
old
desktops.
I
We
watch
tv
less
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
30
percent.
Since
before
2010.,
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
these
data.
This
is
an
example
of
measured
electricity
consumption
by
these
companies
that
run
data
centers.
You
have
the
you:
have
the
Google
Microsoft
Facebook
all
reporting
good
data
Amazon.
I
We
still
have
to
estimate
them,
but
they
are
reporting
now
their
whole
organization,
but
then
they
have
one
and
a
half
million
employees
for
retail
and
warehousing
and
other
businesses.
So
AVS
is
probably
the
largest
electric
consumer.
We
don't
know
exactly
how
large
it
is,
but
I
think
that
estimate
is
pretty
good
and,
as
you
can
see,
the
last
decade
number
of
new
servers
in
the
world
has
been
very
stable
outside
China,
but
then
China's
built
out
their
Data
Center
capacity.
So
most
of
the
increases
in
China.
I
I
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
for
data
centers.
Our
results
is
very
similar
to
what
Kumi
and
masternet
in
the
US
has
published
over
the
years,
not
a
surprise
we're
using
the
same
kind
of
model.
What
we
are
adding
here
is
also
this
data
center
company
data,
that
kind
of
verifies
what
Google
Masonic
has
done
over
the
years:
data
centers
2020,
0.9
percent
of
global
electricity
networks.
I
I
If
we
look
at
networks
and
then
look
at
number
of
subscriptions
in
the
network,
we
see
yeah
the
network
electricity,
the
build
data
networks
very
proportional
to
number
of
subscriptions.
It's
actually
so
that
it's
three
times
as
much
subscriptions,
but
energy
has
increased
by
about
two
times.
So
we
are
a
bit
more
efficient
per
user.
I
I
I
Yes,
that
was
it
unfortunately,
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,
but
I
can
show
them
on
a
later
stage.
To
sum
it
all
up,
so
in
contrast,
then,
to
what
can
be
seen
in
media
the
ICT
sectors,
carbon
footprint
is
stable.
It
is
not
growing
and
one
presentable
people
on
Earth
emit
more
when
flying.
I
Other
consumer
electronics
and
paper
media
has
actually
been
reduced
by
30
or
30
low
remissions
versus
10
years
ago,
and
the
ICT
sector,
the
data
center
companies
and
many
of
The
Operators
or
the
leading
investors
now
in
renewable
electricity
in
the
world,
there's
a
start
there
by
International
Energy
agency.
That
shows
this,
and
we
also
see
this
in
the
data.
I
didn't
mention
that,
but
you
still
saw
all
this
green
figures
about
electricity
consumption
in
the
background
data,
and
that
was
renewable
electricity.
I
I
I
One
of
the
slides
I
had
as
background
if
time
allowed
me
to,
but
I
can
just
mention
that
that
my
new
fiber
connection
that
I
dig
down
this
summer.
The
connection
consumes
around
2.5
Watts
at
the
fixed
Network
site
in
my
old
house.
The
phone
line
and
ads
cell
consumed
handles
total
10
megabit
per
second
now.
I
have
1000
Giga
megabit
per
second,
so
I
have
100
times
more
data,
but
only
it's
four
times
less
energy,
the
route,
the
home,
router
and
fiber
model
consumes
12
Watts
back
in
the
days.
I
You
need
the
modem
and
the
router
consumed
80
modes,
and
and
then
you
have
the
core
so-called
cordless
phone.
Nobody
has
like
phones,
anymore,
fixed
phones,
and
that
consumes
another
few
Watts,
so
nearly
hold
consumption
for,
for
my
CP
part,
100
times
more
data
and
in
the
future,
who
knows
maybe
in
10
years
we
can
all
get
10
gigabit
per
second.
What
to
do
with
all
this
data?
I
have
no
idea.
I
I
A
A
J
Thank
you
very
much
and
I
will
I
will
open
the
stage
with
a
sadly
provocative
points.
You
were
saying
that
we
should.
We
should
not
worry
and
keep
on
consuming,
because
your
projects
can
say
the
energy
consumption
will
remain
the
same
if
the
rest
of
the
economy
does
deliver
and
aggressively
decarbonize
the
ICT
sector
will
look
very
bad
if
we
don't
follow.
I
Yeah
but
I
I,
I'm,
very
well
I
mean
looking
to
what
is
happening
now,
I
think
the
in
a
few
years
the
data
center
companies
and
the
network
companies
would
probably
be
the
ones
that
has
switched
to
renewable.
It's
just
the
faster.
The
Lander
other
sector
can
do
for
data
sentences
It's
relatively
easy.
They
are
not
spread
around
already
located
in
countries
like
Sweden,
with
access
to
renewable
electricity
for
networks.
It's
a
bit
harder,
especially
if
you're.
If
you
are
an
encounter
where
there's
little
access
to
renewable
electricity,
but
hey
look
at
India.
A
I
Then
I
would
I
would
challenge
you
a
bit
there
when
you
say
that
when
you
install
new
things
the
and
it
goes
up
what
we
actually
saw
the
opposite
in
Europe
in
20,
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
and
when
they
installed
new
equipment,
it
consumed
less.
I
J
J
J
J
So
you
do
provide
an
empirical
perspective.
You
say
we
are.
We
have
we've
got
data
reaching
back
20
years
and
that
data
shows
that,
despite
those
enormous
increases
exponential
increases
in
data
volume,
energy
consumption
remains
stable
and
I.
I.
Take
that
that
point
the
question
is:
are
we
Nearing
and
underneath
I
know
when
we
walk
through
the
data
center
and
Stockholm
you
should
you
showed
me
rows
of
pstn
equipment
that
has
turned
has
been
turned
off.
J
I
I
It's
a
little
piece
of
equipment
that
you
put
in
the
port
consumes
a
what
and
then
like,
in
other
words,
so
two
Watts
when
I
started
the
work
or
maybe
a
few
years
before,
then
you
would
need.
Then
you
had
the
same
capacity
going
from
one
big
city
to
another
and
you
would
need
a
whole
room
full
of
equipment
consuming
several
kilowatts.
I
I
But
in
the
end,
it's
it's
it's
only
a
few
Watts
that
can
be
saved
per
user
and
even
if
I
use
10
times
more
data,
it
wouldn't
change
the
wattage
of
my
connection
or
my
router,
and
so
on.
It
will,
when
I
load
my
route
with
the
500
megabit
per
second.
It
goes
up
0.3
Watts,
if
I'm
on
a
long
port,
if
I
own
the
Wi-Fi,
it
can
go
up
with
more
than
I
want.
I
A
L
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
cdns,
which
are
outside
of
your
measurement
scope,
because
I
work
with
a
lot
of
cdns
in
green
and
streaming,
and
we
are
only
just
at
the
beginning
of
measuring
these
things
in
any
practical
terms.
So
it's
it's!
It's
it.
It's
important!
We
get
you
that
data,
but
if
you've
got
other
sources
that
I'm
intrigued
because
I
see
you
only
see
CDN
energy
Bills
going
up.
I
I
You
mentioned
the
work
with
the
requirement,
trust
and
Netflix
and
others,
and
in
that
work
there
is
quantification
of
the
their
entire
work,
Network,
including
all
the
sedan
and
so
on,
and
that
part
was
rather
small
compared
to
the
rest
and
in
our
data
we
have
big
big
players,
all
the
big
yeah
see
them
players
or
they
are
like
aluminum
and
they
have
others
adding
all
the
pots.
You
have
Amazon
the
others
one.
We
actually
ask
our
providers
of
data
for
the
data
so
to
speak
and
get
it.
I
I
cannot
share
it.
Unfortunately,
of
course,
but
I
think
that
when
it
comes
to
networks,
it's
the
last
Model
that
always
has
been
the
number
one
thing,
and
as
soon
it's
like
as
soon
as
I
mean
I
sit
in
a
cabinet
with
around
100
others
and
on
the
back
side.
It's
what
it's
one
10
gigabit
portal,
which
will
share
and
have
a
hard
time
to
fill
it,
and
it
consumes
nothing
compared
to
all
the
access
that
has
to
go
out
to
everyone
else.
I
L
A
E
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
understand
more
of
the.
Why
and
I.
E
You
know
why
the
numbers
remain
the
same
and
it
sounds
like
there
are
some
things
behind
the
scenes,
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
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
you
know
Dom
talked
about
cdns.
E
The
previous
talk
you
know,
Dan's
talk
was
about.
You
know
the
the
load
on
the
network
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
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
your
thoughts
on
that.
I
I
would
first
I
would
switch
around
the
data
that
okay,
there's
more
date,
I
would
say
what
happens
first
is
that
we
make
a
technology
Improvement
and
we
always
had
this
more
slower.
It's
it's,
not
the
law.
It's
like
it's
an
observation
and
it's
still
going
on.
Maybe
not
this
fast
and
you
can
order
it
if
it's
still,
but
even
though
it's
it's,
it
slows
down
the
Summers.
Then
we
take
up
on
other
areas
and
over
time.
I
If
I
look
at
like
my
the
the
laptop
I'm
I'm
using
right
now,
it
consumes
less
than
less
than
10
watts,
and
it's
about
a
check
to
when
I
was
starting
at
school.
It's
like
100
times
every
each
parameter
is
like
100
times
more
or
Rich
memory,
speed,
everything,
data
and
so
on,
but
it's
only
one
tenth
of
the
weight
and
one
tenth
of
the
energy,
and
actually
everyone
can
seize
this.
I
But
the
same
thing
happens
in
the
data
centers
Network,
it's
the
same
semiconductors
that
that
is
the
basis
for
all
this
and
and
you
you
see
the
same,
I
mean
it
like
the
example,
the
back
in
the
80s
that
you
need.
The
whole
room
doing
the
same
thing
that
my
little
Port
can
do
now.
What
that
I
connect
to
my
fiber
another
example
for
mobile,
it's
like
when
I
started,
you
needed.
I
I
E
I
I
If,
if
air
travel
had
improved
as
much
as
Electronics
the
time
I
have
been
living,
we
would
be
traveling
at
the
speed
of
light
one
million
times
faster,
1
billion
kilometers
an
hour.
It
has
been
a
lot
but
that
you
can
of
course
say
that
yeah.
It's
still,
you
still
have
this
one
PC
you're
only
writing
documents,
yeah.
A
M
I'd
also
like
to
maybe
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
to
to
what
you're
seeing
and
it
could
well
just
be
perspected.
You
know
you're
looking
at
the
global
taking
into
it,
which
takes
into
account
the
sort
of
socioeconomic
stuff
that
goes
on
as
well
as
just
what's
happening
in
in
one
country
and
one
network,
but
in
the
interest
of
time
I
think
we'll
just
move
along
is
that
okay.
A
A
A
What
are
the
areas
where
we
know
our
expertise
is,
is
not
sufficient?
What
are
the
areas
where
we
know
we
need
more
data
and
I
sort
of
asked
this
question
on
on
the
on
the
mailing
list
ahead
of
time
and
I've
sort
of
put
prompts
for
another.
The
the
email
threads
I
saw
does,
and
anyone
have
opinions
on
this.
What
what?
What?
What
don't
we
know?
What
are
the
known
unknowns.
A
G
G
That
means
that
that
we
could
save
some
some
energy
or
be
otherwise
more
efficient
or
use
less
money
in
various
Enterprises
energy
is
very
costly
reasonably
if
you've
not
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
where
is
it
cost
effective
to
make
improvements
and
like
starting
from,
like
you
know,
not
from
who
is
guilty
but
like?
Where
is
like?
G
Another
thing
that
it
would
be
very
useful
to
understand
and
I
think
that
data
also
exists
at
least
to
some
extent
like
which
part
of
the
traffic
is
responsible.
For
what
and
I
guess,
I
don't
know.
Video
on
the
internet
is
video
and
streaming
is
it's
a
it's
a
huge
fraction
of
that
and
and
then
we
could
perhaps
try
and
understand.
Like
you
know
what
what
out
of
these
things
are,
are
the
worthwhile
things
to
do
them?
G
So
a
lot
of
effort
has
has
went
into
that.
So
in
order
for
us
to
do
something,
I
think
we
need
to
figure
out
again
what
the
right
place
is
for
us
to
do
that,
and,
and
also
typically-
and
you
see
this
in
in
some
of
the
papers
that
that
we
have-
or
basically
all
of
the
papers
that
you
know
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.
G
This
thing
here
today
is
supposed
to
be
proportional
Improvement.
We
actually
have
to
understand
what
that
implies.
Also,
you
know,
maybe
you
can
reduce
the
cost
of
your
box
or
the
energy
cost
of
your
box,
but
what
about
other
boxes
in
the
network?
Does
that
imply
something
for
them?
So
those
sort
of
sort
of
questions
were,
on
my
mind
at
least.
A
H
So
we
talk
about
like
video
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
attend.
The
applications
have
written
and
there's
a
big
difference.
If
I
use
my
Max
built-in
webcam
versus
like
a
Logitech
cam,
that's
got
h264,
so
I
think
we
don't
have
a
really
good
grip
on
this.
H
So
if
you're
kind
of
putting
this
all
in
the
same
bucket
and
I
I,
don't
know
the
good
answer
to
this.
But
I
think
this
is
something
we
need
to
look
at
like
for
sure,
I.
Think,
like
Yen's
point,
I
was
like
oh,
like
whatever
we
do,
it's
going
to
stay
the
same,
but
we
assume
that
things
are
going
to
get
improving
on
the
consumer
side
right
on
the
edge
device
side,
but
I
think
that's
something
that
needs
to
get
better
because
and
we're
not
doing
this
because
it's
a
very
hard
problem.
H
A
I
I
mean
as
the
I'm
sorry,
as
the
networks
are
now
it's
it's,
they
are
consuming
this
more
or
less
the
same.
All
the
time,
usually,
we
are
on
a
low
low
part
of
the
the
we're
using
just
a
few
percent,
but
still
it's
consumed
me.
So
one
thing
is
that
yeah,
please
use
it
more
now
well
make
make
more
use
of.
It
is.
I
I
I
B
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
one
is
about
what
is
the
footprint
of
ICT
and
I?
Think
when
we,
and
especially
when
Jens,
are
talking
and
says
that
it's
much
smaller
than
it
is
it
it
and-
and
it's
not
likely
to
develop
in
line
with
some
of
this
more
alarmistic
scenarios.
B
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
with
other
activities
as
Aviation.
Someone
said
in
the
chat
is
not
important.
I
think
it's
very
important
because
not
for
us
as
the
ICT
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
most
harming
ones
and
which
are
less
resource
intensive.
B
B
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
alarmist,
that
doesn't
take
away
the
importance
of
optimizing
and
reducing
the
emissions
which
and
the
energy
usage
which
is
associated
with
our
sector
and
from
the
perspective
of
electricity.
B
Hopefully,
going
renewable
and
renewable
is
important,
it
is
the
main
strategy
for
decarbonization
and
we
are
a
demand
side
player.
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
so
we
have
to
keep
down
our
electricity
consumption
as
much
as
possible,
so
we
don't
waste
that
entity
so
I
think
there
are.
There
are
a
lot
of
reasons
to
to
look
within
our
own
boundaries.
B
A
N
The
life
cycle
of
the
products
right,
the
all
that
is
related
to
a
scope
3
to
the
use
of
of
the
products,
and
that
should
be
updated,
I'm,
not
sure
if
real
time
or
not
still,
things
need
to
to
be
defined,
but
something
I
guess
this
community
can
help
to
provide
more
inside
information
to
the
European
Commission
on
any
governmental
organization.
Right
as
this
will
be
imposed
by
a
countries
and
that's
my
understanding
but
yeah.
A
A
N
M
J
J
J
K
A
Yeah,
so
it
sounds
like
one
of
the
I
mean
the
last
slide
was
communication
and
Outreach.
It
sounds
like
one
of
the
things
we
should
be
doing
is
having
a
broader
conversation
about
whether
we
are
trying
to
optimize
for
for
keeping
the
same
set
of
services
or
how
are
we
changing
the
way
people
use
the
network
and
how
it
Regulators
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.
A
E
I
A
K
A
G
No
I
I
I.
Thank
you
all.
It's
been
a
very
active
discussion,
different
viewpoints
and
lots
of
data,
but
also
I,
guess
some
missing
data
and
I
think
that's
that's
one
of
the
things
that
we
need
to
talk
about
also
tomorrow.
If
we
can
improve
on
that
aspect
as
well,
not
just
and
first
understand
what
needs
to
be
improved
and
then
and
figure
out
the
improvements
and
I
guess.
I
was
trying
to
write
some
high
level
very
high
level
conclusions
from
the
session
and
that
I
came
up
with.
G
We
have
some
understanding
of
the
environmental
impacts.
At
least
some
bracketing
like
this
is
roughly
what
we're
talking
about
and
and
most
people
seem
to
be
selling.
At
least
from
that
perspective.
A
similar
story
also
seem
to
have
agreement
that
some
most
alarming
stories
may
not
actually
be
factual,
but
then,
of
course,
still
ICT
and
internet
uses.
G
You
know
fair
amount
of
energy
and
any
savings
there
would
actually
be
material
like
size
of
Norway.
Even
so,
we
should
probably
pursue
that,
and
this
probably
effort
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.