►
From YouTube: IETF114-RTGAREA-20220725-1730
Description
RTGAREA meeting session at IETF114
2022/07/25 1730
https://datatracker.ietf.org/meeting/114/proceedings/
A
A
A
A
That's
andrew,
this
is
john,
and
this
is
of
course
me
alvaro.
Today
we
have
a
one
hour
meeting.
It
is
gonna,
be
a
short
meeting.
We
have
some
items
in
the
agenda,
we're
going
to
go
through
in
a
second
I'm
happy
that
we're
having
this
meeting
at
the
beginning
of
the
week,
which
is
unusual
because
we
usually
have
it
in
the
middle
or
towards
the
end.
So
it
gives
us
a
chance
to
when
we
get
to
open
mic
to
talk
about
any.
A
B
A
A
A
So
this
is
the
note.
Well,
please
be
aware
of
what
it
talks
about
not
just
about
intellectual
property,
but
also
about
conduct.
The
code
of
conduct
is
one
of
the
items
listed
there
and
the
anti-harassment
policy
as
well.
Next
slide.
The
code
of
conduct
includes
things,
for
example,
be
nice
to
other
people
is
basically
what
it
summarizes
to
treat
everyone
with
respect
speak
slowly.
A
A
A
A
A
A
This
is
an
important
time
because
we
are
in
the
middle
of
the
year
time
frame
of
the
ietf,
which
means
that
the
nom-com
is
starting
to
gear
up.
I
know
some
of
you
already
have
volunteered
to
be
part
of
the
nomicon,
which
is
great.
Please
consider
volunteering
to
serve
in
one
of
the
open
positions
as
well
this
year,
john
and
I's
period
ends
in
march.
A
So
if
you
are
interested
in
being
an
ad,
please
consider
running
if
you
know
of
other
people
who
you
think
would
be
good
at
ads,
please
consider
nominating
them
as
well.
If
any
of
you
want
to
talk
to
any
of
us
about,
you
know
what
what
type
of
job?
What
do
we
do
every
day?
What
what
those
are
they
in
life
would
maybe
look
like
please
let
us
know
and
we'll
be
very
happy
to
to
talk
about
that
and
and
tell
you
everything
you
need
to
know
about
the
a.d
job.
A
Another
request
that
we
always
ask
is:
please
please,
please
review
documents.
This
is
important,
because
a
big
part
of
the
mission
of
the
atf
is
to
produce
documents
that
people
can
use
to
interoperate.
These
documents
need
to
be
clear.
They
need
to
be
readable
by
everyone,
not
just
by
the
person
who
actually
wrote
them.
A
It
is
great
that
in
many
cases
you
trust
all
your
fellow
itf
participants
and
trust
that
their
documents
are
clear.
It
would
be
better
if
you
actually
reviewed
the
documents
and
we
all
made
sure
that
they
are
the
way
this
should
be
so
please
review
them
as
part
of
the
whole
working
group
process.
Adoption
working
on
the
documents
working
with
last
call
ietf
call
at
any
time
that
you
want,
or
you
can
please
review
documents.
A
A
Then
we're
going
to
talk
a
little
bit
about
the
routing
discussion,
mailing
list
and
a
couple
of
changes
that
we're
making
there
then
we're
gonna
have
peng
talk
about
or
give
us
an
update
on
the
computer
aware
networking
work
that
has
been
going
on.
They
had
a
buff
at
the
last
ietf
and
then
we're
gonna
have
some
time
at
the
end
for
open
discussion.
C
D
C
F
G
G
A
Sure
so
this
slide
is
just
the
distribution
of
the
working
groups
between
the
three
of
us.
In
case
you
don't
know
who
the
corresponding
idea
is
just
come
in
and
look
for
us,
but
yeah.
You
can
talk
to
any
of
us
about
any
anything,
especially
if
there
are
issues
in
the
working
groups
next
slide.
So
now
we're
on
to
the
running
directorate
report.
H
I
H
We
don't
have
a
member
changed
from
last
ietf.
The
purpose
of
this
team
is
to
give
the
review
of
the
routing
and
other
related
drops
from
other
areas,
and
we
also
offer
the
early
review
of
any
routing
area
document
before
the
last
cross.
So
all
the
working
group
chairs
are
welcome
to
request
this
kind
of
review
to
the
doctorate
and
the
wiki
page
is
available
and
you're
welcome
to
visit
that
next
week.
Please-
and
here
are
some
very
brief
statistics
from
last
itf.
H
Actually,
we
just
trimmed
this
to
restart
the
data
from
2021
and
from
2021.
We
are
totally
having
21
early
review
and
53
last
call
reviews,
and
since
last
itf
there
are
five
early
review
and
the
14
last
call
review.
So
basically,
this
is
a
very
close
to
the
average
number.
Given
the
past
few
meetings
and
the
top
working
groups
are
idr
best
and
the
rsr.
We
also
give
quite
many
review
on
the
other
areas
say
basically
on
all
psa
and
side
drops.
If
I
remember
correctly,
and
the
next.
H
H
This
would
be
almost
a
half
50
percent
of
them
getting
ready
and
there
there
are
a
lot
of
the
works
need
to
be
follow
up
after
the
review
as
well
and
basically,
another
statistic
for
the
non-routing
draft
is
on
the
right-hand
side
for
people's
information,
and
I
think
that's
it,
and
we
will
keep
this
updated
every
time
and
report
it
to
the
routing
area.
Thank
you.
I
A
A
The
purpose
of
this
list
has
been
to
have
general
discussions
around
topics
related
to
the
routing
area,
so
we
have
updated
the
description
of
this
list
and
we
wanted
to
talk
a
little
bit
about
about
that
with
you
just
so
that
you're
aware
you
can
now
visit
in
the
next
slide.
There's
a
link
to
the
mailman
list
information.
You
can
visit
that
and
see
the
whole
description
of
everything
that
we
put
there.
A
Basically,
the
purpose
of
the
list
doesn't
change
we're
just
making
the
description
a
little
bit
more,
more
complete
to
include
a
little
bit
of
scoping,
calling
out
things
that
are
okay
to
be
posted
there.
We
never
really
had
you
know,
issues
with
with
inappropriate
postings,
and
also
we
are
talking
about
a
little
bit
more
specific
about
the
the
moderator
role.
What
a
moderator
is
we're
assigning
a
moderator
to
the
list,
ross
white
who's
sitting
over
there
he's
waving.
A
One
of
the
reasons
that
we're
doing
this
is
because
there
have
been
topics
that
have
come
up
in
different
places
in
the
rowing
area
or
even
outside
the
routing
area.
In
some
cases
that
are
maybe
not
ready
enough,
be
a
a
working
group
or
a
boss
or
have
a
specific
proposal.
Those
things
should
already
follow.
The
normal
process
maybe
go
to
our
twg,
for
example,
but
they
are
important
to
discuss
last
time.
In
our
twg
example,
there
was
a
presentation
around
the
importance
of
thinking
about
the
evolution
of
routing
in
the
internet.
A
Yes,
that's
an
important
topic.
It
is
within
the
purview
of
the
routing
area.
Our
mission
remember,
is
to
keep
the
routing
system
operational
and
to
make
sure
that
the
internet
works
pretty
much.
So
you
know,
topics
like
that
are
within
the
purview,
but
they
are
not
specific
enough
right
to
be
considered
to
be
a
working
group
to
just
go,
discuss
routing
the
future
of
routing,
but
we
want
to
track
those
discussions,
so
we
already
have
this
list,
so
the
basic
addition
is
that
extra
function
of
the
moderators.
A
A
A
A
We
also
include
here
well,
the
first
bullet.
There
is
initial
discussions
of
open-ended
topics,
so
evolution
of
routing,
for
example,
is
an
open
in
the
topic
that
we
hope
that
by
discussing
it,
we
will
come
up
with
more
specific
proposals
that,
as
a
community,
we
can
discuss
things
and
agree
on
maybe
a
direction
that
I
don't
know
what
what
the
discussion
is
going
to
be
about.
A
Maybe
announcements
of
conferences
or
events
that
are
related
to
the
routing
area.
That's
fine
and
discussions
of
idea
of
routing
area,
specific
policy
or
direction.
Those
things
are
all
in
scope,
things
that
are
not
in
scope
that
you
can
see.
There
are,
of
course,
advertisements
or
you're
trying
to
hire
someone,
or
you
know
whatever
those
types
of
emails
are
not
not
good
anything
that
goes
against
the
code
of
conduct
or
or
is
at
odds
with
whatever.
The
note
well
talks
about
discussions
of
subjects
that
are
unrelated
to
the
routing
area.
A
Of
course,
we
don't
want
to
discuss
them
there,
there's
probably
other
working
group
other
lists,
maybe
working
groups,
maybe
other
non-working
lists,
go
discuss
them
in
announcements
about
conferences
that
are
not
related
to
the
routing
area.
So
you
know
there's
a
wide
range
of
things
you
can
post
and
discuss
on
the
list.
A
We're
just
making
it
a
little
bit
clearer
and
the
text
about
the
moderator,
there's
two
parts
of
that
text:
one
is
what
you
would
expect.
You
know
a
moderator
batteries
list.
You
know
there
are
tools
that
a
moderator
can
use
if
there
is
uncivil
behavior,
for
example,
to
to
restrict
postings
etc.
You
know
the
same
things
that
work
group
chairs
have
for
their
own
lists.
A
Hopefully,
in
a
meeting
like
this
in
the
routing
area
open
meeting,
there
may
be
a
discussion
and
request
for
maybe
time
enter
a
meeting.
Maybe
you
know
some
other
time
to
go,
discuss
some
of
the
topics
that
are
being
discussed
on
the
list.
This
is
again
intended
to
catch
important
discussions
around
writing
technology
around
the
routing
area
that
currently
don't
have
a
specific
home
that
are
maybe
broad
general
that
are
important
to
the
area
and
that
are
important
for
us
to
as
a
community.
Consider.
A
A
It
would
be
great
if
you
all
went
and
subscribed
to
the
rather
discussion
list
as
yeah.
We
hope
again
that
it's
going
to
be
more
of
a
general
discussion
forum
for
routing
specific
topics
and
since
all
of
you
are
sitting
here,
I'm
assuming
all
of
you
are
routing
specific
topics.
It's
going
to
be
a
great
list
for
you
to
be
part
of.
A
K
E
C
All
right
so
just
to
briefly
tee
this
up
just
to
briefly
tee
this
up.
You
may
remember
that
at
the
at
113
we
had
a
computing
aware
networking
buff
and
we
there's
been
sort
of
a
lot
of
action
on
the
list
since
then
so
there's
a
short
readout
prepared
for
today
and
are
you
ready
to
speak?
Actually,
I
don't
know
is
yeah
peng,
please
go
ahead
and
unmute
yourself
and
andrew.
Are
you
going
to
advance
the
slides
or
us?
Okay.
F
F
So
in
the
camp
of
the
use,
cases
showed
the
necessity
of
steering
traffic
considering
more
factors.
Besides
network
letters,
we
also
analyzed
the
gap
with
dynasty
and
the
load
balancer,
which
has
a
gap
of
design
for
dynamic
scenarios
of
bus
approach
with
additional
latency
and
should
learn
more
resource
status,
and
there
were
two
potential
solutions,
including
the
dynamic
anycast
and
enhanced
load
balancer.
F
F
The
first
category
is
what
the
relation
between
can
traffic
steering
and
service
deployment
discovery
and
the
upper
layer
protocol.
The
object
is
that,
can
you
assume
the
service
has
been
deployed
and
discovered
in
multi-edge
sites
aiming
at
traffic
steering
can
be
coordinated
with
upper
layer
for
service
deployment
and
discovery.
F
Moreover,
the
commuting,
metrics
collection
income
will
also
benefit
them,
and
the
second
category
is:
what's
the
relation
between
can
auto
auto
solve
the
problem
of
service
instance
selection
as
an
office
solution
as
dinners
and
the
road
bands
replies
for
applications
or
service
before
traffic
delivery
might
not
be
optimal
or
valid
after
the
handover.
So
more
details
are
needed
of
also
including
some
extension
to
support
multi-deployment,
quick
interaction,
integrate
more
performance,
metric
information.
F
F
F
F
F
C
K
A
A
F
J
Chris,
I
yeah,
I
don't
I
tried
to
get
in
the
queue.
I
don't
know
it
didn't
work
but
yeah.
I
I
mean
I
just
to
echo
what
tony
just
said.
I
just
because
you
can-
or
this
isn't
echoing
tony,
but
just
because
you
can
doesn't
mean
you
should
I
I
just
don't
think
this
is
the
right
thing
to
do.
We
shouldn't
use
routing
to
do
this.
M
So
the
thing
is,
we
do
believe
you
know
this
work
is
significant
and
involve
lots
of
routing
related
work
and
then
the
techno
details
all
have
been
discussed
and
then
to
make
it
a
simple.
You
know
for
for
some
proposed
solutions
like
the
diecast
one,
or
they
even
mentioned
that
lb.
Those
type
of
things
you
know
for
for
the
cam
work
it's
going
to
include
the
both
the
compute
and
matrix.
M
The
thing
is
that
you
have
to
collect
those
information
through
some
mechanism
either
like
the
igp
bgp
and
they
use
those
information
at
based
on
the
architecture.
That's
going
to
be
proposed
or
grid
to
make
the
decision.
You
know
for
the
the
optimal
instance
and
then
about
instance,
some
joking
paths.
Those
type
of
things
will
be
like
a
proposed
discuss
all
kinds
of
things
so
from
the
actually
from
the
the
very
beginning,
the
site
meeting
that
was
held
the
last
march,
plus
the
buff
this
past
march.
M
D
Well,
I
I
think
we
we,
we
have
got
a
very
good
beginning
of
the
game,
walk
because
because
we
have
finished
buff
in
the
previous
meeting,
that
is
pretty
good
and
then
different
people
may
have
different
view
on
the
can
walk
that
that's
totally
fine
and
we
can
discuss
it
further
on
the
main
list
or
offline.
Yet,
thank
you.
N
So
without
that,
I
really
find
hard
entertain
any
arguments
on
extension
to
igp,
because
the
problem,
yes
might
be
neat
solved,
but
it
appears
at
least
from
what
I
understand
in
outer.
That
outer
is
reasonable
mechanism
and
already
exists
and
with
some
extensions
seems
to
be
quite
a
suitable
solution
to
the
problem,
which
would
be
just
one
use
case.
So
I
would
probably
encourage
I
would
encourage
proponents
to
analyze
outer
and
then,
if
they
can
demonstrate
that
yes,
alta
is
not
applicable
to
their
problem.
N
K
I
think
for
future
objective,
it's
necessary
to
make
it
the
scope
clear,
and
then
we
can
focus
in
some
specific
user
case
such
as
a
cloud
to
to
to
access
then
otherwise,
right
now,
the
some
area
is
not
very
clear
to
to
me
because
it's
kind
of
it's
a
fix
mixed
with
the
dynasty
and
also
we
mix
with
the
note
balance.
So
we
better
make
this
clear.
Then
what
is
the
difference
with
the
future
work
then
we'll
be
good.
Thank
you.
O
Hi,
I
hope
you
can
hear
me.
Yes,
I
just
wanted
to
both
reply
to
greg,
but
also
in
parts
to
tony.
We
did
do
updates
on
to
the
drafts,
also
from
some
of
the
updates
try
to
clarify
the
positioning
of
doncaster's
non-path
selection
mechanism.
So
from
that
perspective
it
is
an
aspect
of
load
balancing
pointed
out
that
it's
an
onpath
selection
rather
than
off
our
selection,
which
is
also
the
differentiation
we're
using
when
looking
at
alto,
plus,
dns
and
and
other
of
our
solutions.
O
So,
although
it's
covered
in
the
gap
analysis
part
now,
one
can
and
maybe
should
expand
that
I
do
agree
and
we're
working
on
that,
and
we
have
some
work
that
we
would
like
to
integrate
where
we
looked
at
a
comparison
between
one
parts
and
off
parts
which
may
answer
some
of
the
questions
that
greg
had
had
as
to
why
half-past
mechanisms
like
alto
do
not
suffice
for
the
use
case
that
we
have
in
mind.
So
the
the
the
work
has
been
progressing.
O
That's
the
message
I
wanted
to
give
there's
more
work
to
be
done,
but
we're
not
ignoring
our.
I
mean.
I
also
want
to
make
that
clear,
although
has
actually
been,
even
in
the
version:
zero
already
of
the
gap
analysis
back
in
march
or
even
last
year
already.
I
think
it
was
listed
already,
so
it's
not
that
we're
ignoring
it
we're
working
on
making
it
clearer
and
and
also
adding
potentially
references
to
quantifications
that
come
from
work.
That
has
been
done.
O
C
F
The
auto
issues
we
address
this
is
in
the
analysis
chart
and
we
do
have
some
discussion
into
auto
rpg
and
at
this
stage
I
think
this
is
might
also
be
open,
because
auto
also
need
to
do
more
work
than
to
to
see
if
it
can
meet
the
requirements
or
can
and
reply
to
used,
link,
hands
problem
that
maybe
load.
Balance
in
fact
is
a
different
aspect
with
traffic
steering
of
can.
C
C
D
Yeah,
thank
you.
Thank
you,
john,
like
it's
really
good
for
us
that
we,
we
still
have
a
lot
of
work
to
do
so.
It's
really
good
and
we
do.
We
do
have
done
some
work
of
the
gap,
analysis
and
we
we
do
propose
a
new
draft
and
then,
if
possible,
please
read
the
draft
and,
and
we
we
do,
we
will
be
very
happy
to
see
guys
to
reply
to
the
email
that
we
are
answering
the
issues
that
would
that
would
make
the
work
more
effectively.
I
Yeah,
hello,
hello,
my
name
is
george
carl
jones.
I'm
working
for
huawei
just
a
comment.
You
know.
I
heard
some
opinions
that
during
the
113
that
it
was
proposed,
you
know
that
can't
should
be
based
on
a
load
balancing
solution.
But
that's
not
the
case
from
what
I
remember.
There
was
an
open
discussion
there.
There
were
two
actually
proposals
that
could
actually
work
towards
solving
this
issue.
C
P
This
is
linda
dumber
from
futureway.
Since
john
you
mentioned
that
we
can
start
other
broader
discussion.
I
have
a
question,
so
I
heard
people
talking
about
low
balance
right
so,
but
if
you
look
at
load
balance
there's
there
are
three
load
balance
there:
layer,
four
load
balance,
there's
seven
load
balance
right
in
routing
domain.
We
always
do
ecmp
right,
that's
kind
of
low
balance,
your
balance
between
multiple
paths.
P
So,
if
my
question
is
really
can
routing
area
take
that
kind
of
work
like
front
end,
multiple
paths,
balancing
balancing
among
multiple
paths
would
that
be
part
of
routing
area,
because
you
do
need
the
routing
information
and
then
the
distance
matrix
and
the
destination
matrix.
That's
my
question
to
them
to
the
group.
C
This
is
a
question
that
anybody
can
answer.
I
don't
think
it's
just
addressed
to
the
people
sitting
up
front
and
I
don't
have
like
I'm
not
sure
I
I
think
the
answer
is.
It
depends
right
because,
like
you
said,
there's
there's
load
balancers
that
exist
pretty
much
up
and
down
the
stack
and
things
that
are
close
to
routing
probably
belong
in
routing
and
things
that
that
aren't
don't.
P
So
so,
actually,
that's
a
that's
a
good
answer
in
a
way
that
load
balancing.
If
you
look
at
traditional
load,
balancing
balancing
the
service
instances
and
they
have
instance
right
behind
the
load-
balancer
balancing
among
them-
that's
definitely
not
part
of
the
routing
work
right.
So,
but
if
you
talk
about
network
like
network
wise
balancing
among
multiple
paths,
you
need
protocol
to
exchange
information,
maybe
that's
more
closer
to
the
routing
area
than
the
other
areas.
So
it's.
Where
can
we
start
this
kind
of
discussion
and
then
debate.
A
So
so,
just
to
really
quick,
so
there's
the
related
topic
related
to
can,
of
course,
in
the
then
cast
main
list
as
well
right
whether
this
question
is
appropriate.
If
there's
already
a
list,
please
go
there
and
the
other
thing
I
mean
to
add
to
what
john
said
about
it
depends
sure,
is
wrongly
technology
fit
in
the
routing
area?
Yes
does
routing
related
technology?
Yes,
what
is
the
specific
proposal
right?
We
can't
answer
a
definite
anything
else
besides,
maybe
unless
there
is
a
specific
proposal
right.
M
Next
kenji
chandler
mobile
again,
I
can
give
you
a
real
example
about
the
lb
load
balancing
not
going
to
work.
If
you
look
at
the
latest
5g
architecture,
and
then
you
can
consider
like
the
the
upf
and
then
there
are
a
trend
to
lower
down
the
upf
to
be
d
upf
toward
the
edge
side.
So
if
you
consider
this
type
of
things,
they
are
going
to
have
the
compute.
M
As
mentioned
by
my
colleague
pen,
the
compute
can
be
distributed
across
different
areas
like
in
local
data
network,
in
dn,
owned
by
by
mobile
operators
or
even
in
public
cloud
side.
So
they
are
also
a
trend
in
3gpp
to
put
some
to
the
compute
in
some
area,
even
in
the
the
core
side,
the
core
network
location
side.
M
N
Mic
and
anything
that
seems
to
related
to
routing
area.
We
had
this
interim
meeting
in
rtg
on
context
addressing,
and
I
found
very
appropriately
noted
that
work
on
oem
usually
often
lags
behind
other
documents
and
often
the
documents
and
technologies
news
introduced
without
defining
their
tools
or
extensions
to
existing
tools,
preferably
that
help
operators
to
work
with
these
technologies.
N
So
that
might
be
a
good
start,
something
to
look
at,
so
I
don't
think
that
it
needs
to
be
a
mandatory
section,
but
I
think
that
some
optional
section
would
be
helpful,
at
least
for
us
not
to
lose
this
and
not
to
really
get
to
the
situation
when
we're
about
to
close
a
group
and
a
group
never
produced
any
oem
related
document.
Thank
you.
G
Thanks
greg,
I
think
that,
at
least
from
my
perspective,
I
definitely
acknowledge
the
problem,
because
I've
certainly
seen
it
in
some
of
the
working
groups
that
are
under
me.
I
think
that
the
solution
to
this
is
something
that
I
think
alvaro
and
john
and
myself
are
going
to
have
to
take
away
and
have
a
bit
of
a
discussion
with
and
then
come
back
to
you
on
on
that,
because
I
do
think,
though,
that
it
is
something
worth
having
a
really
good
look
at
and
thank
you
for
the
feedback.
It's
that's
much
appreciated
and.