►
From YouTube: IETF108-MBONED-20200731-1410
Description
MBONED meeting session at IETF108
2020/07/31 1410
https://datatracker.ietf.org/meeting/108/proceedings/
A
A
A
B
All
right,
okay,.
C
A
C
A
B
A
B
We
can't
fly
the
plane
without,
without
a
me,
take
a
note.
B
A
A
A
B
We'll
we'll
we'll
we'll
be
yes,
we'll
be
done.
We
we
should
be
done
in
30
to
40
minutes.
We
should
see
the.
B
B
A
B
A
No,
we
see
nothing
right
now,
all
right.
Let
me
re
retry
that
yeah
my
my
screen
sharing
dropped
a
couple
times
on
wednesday
morning,
so
just
be
aware
of
I'm
like
talking
along
and
suddenly
when
nobody
saw
it.
B
Okay,
I
thought
I
had
something
up.
Okay,
I
am
publish
screen.
Yes,
I
want
to
really
publish
screen.
B
B
Here's
our
agenda,
it's
pretty
light,
I'm
just
gonna
go
through
updates
on
the
current
drafts
and
current
current
working
group,
docs
and
jake
is
going
to
go
through
his
drafts,
his
multicast,
his
three
multicast
to
the
browser
drafts.
B
A
B
Okay,
all
right,
then
I'll
start
from
scratch.
So
everyone
must
see
the
note
well
and
note
it
well
now
the
agenda
cool
any
any.
Anybody
have
anything
to
add
to
the
agenda.
A
Let
me
just
say
something
lenny,
because
this
came
up,
hopefully
there's
a
lot
of
overlap
between
people
here
and
those
taking
place
in
writing
area
as
well.
So
you
know
beer
is
some
aggressive
work,
moving
forwards
on
a
whole
different
replicating
model
for
the
for
networks
as
whole,
so
internet
and
in
the
charter.
We
have
a
may
pursue
a
v6
native
option.
What
we've
really
discovered
is
there
is
no
real
native
way
to
do
this.
A
The
question
is:
do
we
need
an
end
cap
in
six
and
we've
got
a
requirements
draft
right
now?
That
is
what
the
working
group
is
focused
on.
Looking
at
what?
What
are
we
trying
to
do?
That's
potentially
unique
in
six
or
is
there
anything
unique?
We
need
to
do
in
six
to
provide
transport
of
beer,
so
please
just
pay
attention.
Look
at
that.
A
You
know
operator
input
would
be
fantastic,
so
I
can
send
the
draft
to
the
list
if
you
want,
so
you
guys
know
what
we're
looking
at
and
chime
in
and
bring
your
colleagues
if
you
can
thanks.
B
Great
yeah
just
greg,
keep
us
updated
on
anything.
You
think
that
is
pertinent
to
us.
Where
are
we
in
terms
of
operational
concerns
for
beer
networks?
Are
we
approaching?
Are
we
are
we
there
yet,
or
is
it
still
a
ways.
A
Oh
yeah,
I
wouldn't
say
a
ways
off
I.
I
know
there
is
hardware
in
labs
right
now
that
operators
are
evaluating,
so
my
expectation
is.
We
will
probably
hear
about
some
real
live
deployments,
possibly
this
year,
but
certainly
next
year
from
there.
Hopefully
we
get
some
bcps.
You
know
some
pressure
put
in
this
group.
Go
hey,
there's
some
things.
We
need
to
do
to
make
sure
we
either.
A
You
know,
deploy
correctly
transition
correctly
interface
with
existing
services
correctly,
whatever
made
to
be,
but
I'm
hoping
we'll
look
to
this
group
to
provide
some
operator
guidance.
B
Great
okay,
moving
right
along
here
are
the
active
working
group
docs.
Actually,
the
dryad
draft
is
since
last
meeting
is
now
rfc
8777
congrats
jake,
forgetting
that
moved
along
pretty
quickly.
The
deprecate
asm
draft
is
at
the
rfc
editor
queue.
It's
been
there
a
little
while,
hopefully
we're
getting
close,
but
that
should
be
hopefully
by
ietf
109.
That
will
be
an
rfc
as
well
the
multicast
problems
draft.
B
That
is
still.
We
are
eagerly
awaiting
revisions.
I
think
two
ietfs
ago,
maybe
was
it.
Singapore
there
was
mike
met
with.
I
believe
there
were
some
security
area
concerns
and
mike
was
working
on
revisions
and
getting
out
that
final
revision
based
on
the
final
comments
and
what
was
agreed
to
mike.
I'm
are
you
here.
Are
you
out
there
mike
I'm,
not
seeing
you
in
the
list
yeah
I'm
seeing
the
list
either
warren.
I
believe
you
are
a
co-author
of
that
draft.
B
E
A
E
We
do
something
with
it
soon.
I
believe
that
yeah,
I
don't
know-
hopefully
we'll
get
it
done
in
the
next
couple
of
weeks,
but
we've
been
hoping
to
do
that
for
a
long
time
and
kept
getting
bankrupt.
E
B
Yeah,
it
seemed
like
mike
had
the
cookie
on
that
one.
So
if
you
want
to
encourage
him
to
get
what
get
it
across
the
finish
line
so
that
we
can
start
working
on
the
solutions.
E
B
Okay,
next
was
the
data
center
deploy
draft.
Also
mike
looked
like
mike
was
on
the
lead
on
that
one
and
as
well.
I
guess
we
don't
have
mike
to
talk
about
that
one,
but
basically
that
went
through
last
call
last
year,
not
quite
enough
support
in
the
first
last
call.
We
had
a
second
last
call
and
from
that
jake
a
couple
months
ago,
gave
some
good
some
good
reviews
and
jordan.
B
Jordan
had
provided
some
additional
reviews
as
well
that
were
pretty
much
pretty
similar
to
what
jake
had
to
say,
which
was
essentially.
This
is
a
valuable
draft.
There's
some
good
stuff
in
it,
but
there's
lots
of
room
for
editorial
improvements.
B
Kind
of
wordy
could
be
much
the
the
text
could
be
improved
significantly
so
and
and
last
jury
last
ietf
greg.
Did
you
get
a
chance
to
read
this
one
as
well
that
greg
and
dino
were
yeah.
B
B
To
maybe
you
know
rewrite,
I
I
I
I
believe
if
my
recollection
is
correct,
this
was
written
for
a
different
purpose
and
then
it
started
out
as
a
survey
and
then
it
evolved
into
an
informational
draft
and
I
think
it
probably
has
remnants
of
its
earlier
purpose
in
it,
and
maybe
just
a
clean
slate
and
starting
from
scratch
on
this
draft
might
be
the
best
way,
but
maybe
we'll
get
we'll
take
that
to
the
list
and
get
feedback
from
the
authors.
B
The
last
one
any
other
thoughts
on
this
draft.
Anybody
else
have
anything
to
say
about
this
draft.
B
B
And
then
the
last
draft
is
the
yang
models
sandy
any
any
updates.
This
has
been
out
for
a
little
while
is
sandy
still
there.
D
F
B
B
Jake,
do
you
want
to
drive
since
it's
or
should
I
bring
up
your
slides
and
drive
for
you.
A
B
E
B
B
I
want
to
go
in
inception
mode
here.
I
tell
you
what
jake,
unfortunately,
I
launched.
B
D
I
can
try,
I
haven't
done
it
yet.
Let's
see
the
good.
D
Okay,
so
let's
see
is
my
video
and
screen
coming
in,
I
don't
mind
doing
inception
mode.
That's
all
fine!.
A
D
E
D
Okay,
so
hi,
I'm
jake
I'll,
be
talking
about
the
progress
on
the
multicast
of
the
browser
work
that
we're
doing
the
project
that
we're
doing
this
is
built
on
our
on
the
three
drafts
we
have
active
in
the
working
group.
Now,
primarily
the
dorm
see
back
in
ambi
I'll,
be
touching
on
those,
but
most
of
my
time
will
be
spent
on
the
sort
of
overall
project
status
and
and
where
things
are
so,
I'm
gonna
be
going
over.
D
What
we've
done
since
the
last
presentation
of
the
2020
interim,
where
we
stand
on
the
implementation
work,
we're
doing
and
what
we've
done
in
terms
of
outreach
I'll
be
going
over
some
of
the
feedback
that
we
got
from
operators
in
particular,
and
I'm
probably
going
to
spend
most
of
my
time
on
the
two
slides
that
are
the
that
are
the
the
current
sort
of
thoughts
I
have
on
responding
to
the
feedback.
D
There's
a
couple
of
issues
that
have
been
raised
and
I
think
they're
pretty
addressable
by
a
you,
know
reasonably
easy
to
to
hack
together
approach.
But
I'm
looking
basically
for
kind
of.
D
Maybe
some
of
you
guys
have
a
better
idea
on
some
of
this,
so
I
wanted
to
kind
of
give
you
a
hint
as
to
what
I'm
thinking
and
see
if
anybody
has
any
suggestions
and
and
sort
of
give
the
reasoning
on
on
where
that,
where
that
would
go
and
then
go
over
a
little
bit
of
the
what
I'm
planning
to
do
next,
what
we're
planning
to
do
next
and
where
we're
headed
with
us,
so
I
haven't
done
any
actual
updates
to
the
drafts.
D
Yet,
although
I
got
some
good
feedback
from
dino
and
I
have
a
few
changes
pending,
but
mostly
since
when
did
we
meet
april
or
march
or
so
it's
been,
it's
been
like
implementation,
outreach
work,
and
I
haven't
yet
gotten
around
to
updating
the
draft
text.
D
There's
a
few
known
things
to
put
in
there
and
still
seeking
feedback
from
people.
D
So
the
the
current
status
on
the
implementation
we've
issued
the
intent
to
prototype
for
chromium
we've
refactored
the
api
that
that
was
public
what
last
july,
or
so
like
a
year
ago,
now
to
to
use
readable
streams
in
the
api,
which
is
the
way
that
web
transport
does
it
and
it's
it's
guarded
with
a
command
line
flag,
which
is
one
of
the
requirements
for
doing
these
kinds
of
early
stage
experiments
for
where
we
are
now
hoping
to
try
to
check
that
in
before
too
long.
D
I
don't
know
how
the
code
review
is
going
to
go
when
that
gets
going,
but
you
know
ideally
that'll
be
that'll,
be
somewhere
before
too
long.
It's
possible
that
it
won't
be
checked
in
still
in
november,
but
one
can
hope
our
internal
poc.
I
guess
it
was
successful,
we're
going
ahead.
Obviously
the
we
had.
We
got
the
webassembly
build
to
run
inside
the
browser
and
produce
video
from
our
product
line.
That
does
video
with
multicast.
D
So
that's
all
we
are
go
on
that
we
kind
of
have
our
our
proof
of
concept
running.
None
of
this
is
using
ambi
or
dorms
or
cbac.
Yet
it's
just
doing
a
basic
native
receive,
but
but
the
sort
of
transport
path
is
there.
D
The
cpu
utilization
is
still
not
great,
but
it's
adequate
to
the
task.
For
now
there
will
presumably
be
some
updates
to
to
improve
on
that.
Hopefully,
at
some
point
I
I
would
say
that
that
carries
some
risk
at
this
point
like
we
don't
know
yet
where
the
issues
are,
but
we
think
that
they're
mostly
on
the
transport
between
the
render
and
browser
processes,
so
that
by
putting
that
in
a
shared
memory
transport,
it
should
be
okay.
D
So
there's
just
some
like
engineering
work
to
do
on
that.
But
but
it
looks
tractable.
I
think
I've
spun
up
a
dorm
server,
it's
running
and
discoverable
now,
so
you
should
be
able
to
to
do
the
lookup
as
described
in
the
dorms
draft
and
get
the
srv
record
to
see
where
to
connect
to
it,
and
it
has
cbac
data,
not
yet
any
ambi
config,
but
that
all
that's
using
the
cz
next
check
con
implementation
of
of
rest
conf
and
it's
using.
D
I
think
it's
it's
a
slightly
changed
yang
model
from
from.
What's
in
the
drafts
for
dorms
and
see
back.
D
I
think
it
was
only
to
ambi
actually
because
the
the
crypto
types
changed,
if
I
remember
right
but
that'll,
be
part
of
the
next
draft
update
we'll
include
the
updated
young
model
for
that,
but
that
all
runs
okay.
That
seems
like
it
does.
What
it's
supposed
to,
and
I
have
a
script
that'll.
You
know,
given
an
sg
will
say
what
c
back
says.
D
The
max
bandwidth
is
and
I'll
be
trying
to
integrate
that
into
the
into
the
multicast
received
platform
that
we're
going
to
be
using
in
the
trials
did
a
bunch
of
outreach.
The
operator
community
we've
had
some
30
ish
meetings
with
different
isps,
doing
an
architecture
walkthrough.
D
That
was
a
little
bit
well
that
wasn't
very
much
through
the
ietf
effort,
but
but
with
some
advice
you
know
from
from
a
couple
of
you
guys,
I
went
to
nana
and
presented
there
that
also
led
to
the
apnic
blog
post
that
we
put
up
the
response
on
this
has
been
pretty
good,
so
I
might
continue
doing
a
little
more
on
this
front
just
to
answer
all
the
questions
people
have,
but
that
was
that
was
pretty
successful.
People
are
pretty
interested.
D
One
thing
to
mention
is
that
people
are
recognizing
the
software
and
game
delivery
as
a
sort
of
key
use
case.
Now
it's
not
just
the
the
live
video
use
case.
We're
we're
expecting
to
do
some
trials.
D
Let's
start
this
year,
those
are
not
nailed
down
yet,
but
but
we're
still,
you
know,
negotiating
about
that,
and
and
people
have
broadly
been
supportive
of
the
general
direction,
although
there
is
some
feedback
that
I'll
get
into
we've
also
been
doing
outreach
to
content
owners
and
making
sure
that
that
kind
of
you
know
that
there's
not
major
objections
on
that
front
as
well.
D
This
this
includes
so
most
of
the
people.
We've
talked
to
so
far
are
video
content
owners,
but
but
we're
also
reaching
out
to
to
some
of
the
game
and
software
delivery
folks
to
see
if
we
can
get
feedback
and
and
some
engagement
with
them
as
well.
D
One
of
our
stretch
goals
we're
not
sure
if
it'll
happen,
but
we
would
love
to
be
able
to
try
a
production
delivery
of
some
of
some
actual
traffic
during
an
av
video
test
using
our
our
video
streaming
product,
but
delivering
it
over
an
amt
ingest
path
into
a
willing
network
with
some
willing
content
owner
trying
it
that's
you
know,
that's
we
consider
that
plausible,
but
there
there
are
several
hoops
to
get
through
to
see.
If
we
can
do
that.
D
Most
of
the
time
people
are
on
the
content
owner
side.
You
know
they
see
the
utility
of
this
so
far.
Everyone
we've
talked
to
so
you
know
the
fact
that
we
have
an
sdk
right
now.
Some
people
would
consider
that
they're
not
gonna
do
that,
but
most
of
the
time
they
just
look
at
that
as
a
as
a
minus,
but
one
that
they
could
live
with.
D
If
it
meant
that
we
started
getting
progress
down
this
path
and-
and
you
know
the
the
intent
I
mean
once
the
delivery
is
there,
then
then
it
opens
up
options,
for
you
know
for
more
open
transports
of
it
as
well,
but
we
haven't
made
any.
We
haven't
moved
anywhere
on
on
trying
to
open
our
sdk
yet
or
write
up
the
protocol.
D
That
would
do
it
so
there
there
might
be
some
transport
layer
work
to
to
get
to
before
this
is
really
ubiquitous,
but
but
we
think
that
that
we're
aiming
toward
the
right
first
steps,
hopefully
on
this,
so
we
did
get
some
notes
from
operators
about
places
where
they've
got
concerns
on
on
our
approach,
although
in
general
everyone
we've
spoken
to
has
hit
with,
I
think
one
exception
which,
by
the
way
everyone
was
spoken
to
is,
of
course,
a
biased
sample,
because
those
that
we
haven't
spoken
to
are
more
likely
to
to
not
care,
but
of
those
we
did
speak
to
people
generally
are
are
supportive.
D
They
want
this
to
work,
they
recognize
the
need
and
they
think
that
there's
there's
approaches
that
are
viable.
Probably
that
said,
there
are
some
challenges,
maybe
just
transitional
challenges,
but
but
like
the
way
that
people
use
it
today
and
what
they,
what
their
networks
are
prepared
to
accept,
does
not
necessarily
fit
very
well
with
this
vision
of
of
using
the
global
sg.
D
D
D
I
was
surprised
to
hear
this,
I'm
looking
for
more
details
on
it,
but
but
they're
uncomfortable
with
just
having
it
sort
of
attempt
to
forward
whoever
came
in
first
from
the
you
know,
from
the
homes
that
are
trying
to
join
some
kind
of
global
sg
in
in
some
cases
there's
you
know
existing
tv
services
that
are
using
multicast,
but
what
they're
using
is
like
these
statically
pinned
groups,
so
there's
no
dynamic
joining
of
of
groups
anywhere
in
the
network.
D
It's
just
it's
almost
more
like
a
broadcast
sort
of
setup
with
these
statically
assigned
groups
that
have
specific
meanings
and
those
groups.
You
know
at
the
at
the
receiver
side
yeah
you
join
it
or
don't
join
it,
but
in
the
network
part
there's
nothing.
D
That's
that's
changing
about
the
group
join
status
as
part
of
the
forwarding
and
this
sort
of
helps
make
sure
that
the
network
is
is
sufficiently
stable
and
then
one
of
the
other
cases
we
encountered
is
that
they're
happy
to
give
us
access
to
a
specific
building
and
that
specific
building
has
a
bunch
of
machines
that
has
I
you
know
that
have
each
machine
has
an
ip,
they
would
assign
us
one
of
the
ips.
D
We
could
have
dynamic,
multicast
groups,
you
know
ssm
groups,
but
they,
but
all
the
all
the
traffic
has
to
come
from
an
ip
that
goes
into
that
building
or
they're
not
going
to.
D
You
know,
they're
not
going
to
propagate
it.
To
that
to
that
building
we
asked
whether
this
is
a
sort
of
business
driven
constraint
or
a
technical
constraint,
and
it
was
business
driven.
So
it's
not
like
fixing
the
technical
side
would
address
it.
This
is
like
we're
going
to
have
to
send
it
from
sources
that
are
specific
to
that
network
in
order
to
get
it
transported
across
that
network,
although
they
are
willing
to
transport
the
traffic.
D
So,
having
taken
some
of
these
comments,
I
had
some
sorry
did
somebody
try
to
jump
in
there.
B
No
actual
jake,
my
question
is
so:
are
they
basically
saying
they
want
to
do
walled
garden
stuff?
They
don't
want
to
do
inner
domain.
D
Well,
they're
not
opposed
to
having
the
source
come
from
outside
is
what
they're
saying
it
sounds
like
they
just
built
a
walled
garden
right
now.
They've
got
a
wild
garden
right
now,
and
they
don't
want
to
change
that.
So
let
me
get
to
what
I'm
thinking
as
a
solution.
Now
I
haven't
written
this
up
as
a
draft,
yet
sorry
about
that,
but
I
wanted
to
get
early
feedback
if
anybody
has
comments
on
it.
D
So
that's
why
I'm
showing
it
here
so
the
the
solution
I
have
in
mind
here
is
essentially
to
let
them
use
their
own.
You
know
let
them
map
it
to
the
the
sg
that
they
want
to
use
within
their
network
right.
So
you
just
build
sort
of
a
translation
layer
at
the
ingest
and
the
at
the
ingress
and
egress,
and
in
this
case,
what
what
pins
it
together
is.
This
I'm
calling
it
a
group
network
address
translation
service
which
is
nats,
and
so
this
would
be
an
http.
D
You
know
just
an
http
api
and
what
you'd
have
is
at
the
access
point.
You
know
so
the
receiver
is
going
to
know
and
at
the
app
layer
discover
the
sg
it'll
issue
its
sg
subscribe.
D
You
know
as
as
usual
at
the
sort
of
you
know,
somewhere
along
the
chain,
either
upstream
of
the
access
point
or
in
the
you
know,
in
that
home,
wi-fi,
router
or
or
actually
in
the
app
I've
got
on
the
next
slide.
D
It's
you
know
just
change
it
around,
but
the
the
idea
is
that
when
you're
trying
to
to
join
this
sg,
you
know
that,
oh
because
I'm
in
this
network,
I'm
gonna
talk
to
this
network
associated
nat
service,
the
nat
service
will
tell
me
how
to
map
the
global
sg
to
the
local
sg
that
I
should
use
within
this
network
I'll
issue.
You
know
what
I'll
propagate
as
rpf
is
this
local
join
while
it's
inside
the
network,
whatever
you're
sort
of
topology,
looks
like
through
that
network?
That's
all
unchanged.
D
It'll
propagate
this
sort
of
locally
subscribed
sg
at
the
ingest
point.
Once
you
receive
the
the
you
know,
join
and
leave
for
the
sg's
that
are
within
that
network.
You'll.
Ask
that
same
nat
service.
D
What's
the
what's
the
mapping
of
the
local
network
into
that
global
space
and
then
you'll
issue
the
the
ingest
outside
into
the
global
subscribe
over?
You
know,
dryad
or
if
you've
got
some
other
kind
of
connection,
that
that's
all
fine.
D
So
the
idea
is
just
to
sort
of
wrap
this
up
within
the
network
and
do
that
that
sort
of
at
the
boundary
translate
so
that
the
network
can
do
whatever
they
want
inside
and
they
don't
have
to
change
what
they're
doing
and
and
just
let
that
happen
so
most
likely
I'll
try
to
write
this
up.
Assuming
this,
this
addresses
the
issues
that
they've
raised
and
that
it
becomes
necessary
to
reach
some
networks.
D
This
might
you
know
not
have
to
remain
this
way
permanently,
but
but
it's
it's
probably
okay.
If
it
does,
I
think
so.
B
D
Yeah,
it's
an
interesting
question.
The
I'm
not
sure
I
fully
understand
it,
but
I
do
know
that
the
same
the
same
question
actually
came
up.
I
spent
a
little
bit
of
time
with
dino
trying
to
to
see
his
his
lisp
approach
to
doing
ingest
and-
and
if
you
remember,
I
think
it
was
maybe
singapore.
He
talked
about
pulling
amt
into
into
a
network
over
using
lis.
I
forget
what
it
was
exactly,
but
it
ended
up
re-keying
the
sg
so
that
the
source
was
the
the
was
the.
D
I
forgot
the
term
now
the
you
know
his
yeah
yeah
right,
the
xtr
that
was
doing
the
forwarding,
because
you
need
the
rpf
to
find
that
thing
in
order
for
it
to
propagate
from
there
right.
A
A
C
Understanding,
one
of
the
one
of
the
reasons
why
some
providers
operators
do
not
want
to
have
to
to
use
global
sources
network
is
relatively
easy.
If
you
have
a
six
to
four
backbone,
where
you
only
have
a
bgp
free
core,
you
don't
want
to
ingest
foreign
ip
addresses
on
your
lsr
sites.
D
Yes,
that
was
nils
right,
yes,
yeah,
who
knows
hi?
Yes,
thank
you.
C
B
C
C
When,
when
you
do
that
via
unicast
and
you
transport
it
via
mpls,
you
will
have
an
lsp
pass
for
it,
but
when
you
are
doing
native
multicast
in
a
six
to
four
backbone,
you
only
have
ip
native
ip
routing
on
the
lsr
and
you
typically
do
not
re-transmission
all
your
bgp
routes
into
your
igp.
B
D
D
Just
one
other
point
I
wanted
to
make
is
that
if
there's,
if
there's
trouble,
deploying
the
sort
of
translation
layer
at
the
edge
where
they're
trying
to
propagate
the
the
join,
it's
also
prop
possible
through
dns
sd,
to
to
have
the
end
application,
or
you
know
an
agent
on
the
on
the
end
device
sort
of
do
that
translation,
so
that
the
only
thing
coming
out
is
a
join
for
the
local
network
as
well,
and
that
can
be
forwarded
without
without
having
to
deploy
something
on
the
end.
D
So
that's
kind
of
my
thinking
on
how
I'm
looking
to
change
this
there's
some
complications.
I've
left
out
for,
like
you
know,
I
think,
there's
some
see
back
related
sort
of
when
you
run
out
of
groups.
You
know
how
are
you
going
to
do?
The
the
group
assignment
and
you'll
have
to
be
able
to
say?
D
Well,
no,
you
don't
have
any
local
assigned
group
for
this
and
and
sort
of
handle
that,
but
I
think
that
this
this
approach
is
relatively
simple
and
I'll
probably
be
doing
it
in
at
least
one
of
the
trials
we're
hoping
to
run
so
so
I
wanted
to
give
a
little
heads
up
the
immediate
problems
that
this
can
address
that
I
know
about.
Are
you
know
the
the
their
specific
local
ip
assignments
within
a
network?
There's,
maybe
asm
only
networks.
D
It
occurred
to
me
that
this
also
could
work
to
handle.
If
there's
a
network
that
is
only
transporting
v6,
or
that's
only
transporting
v4
and
you're
trying
to
join
the
opposite
kind
of
group,
then
you
can
do
that
that
translation
at
this
at
the
same
layer
there's
no
reason
that
the
translation
can't
map
between
v6
and
v4,
and
that
the
nat
service
can
maintain
a
population
count
and
perform
the
sort
of
group
assignment
and
and
make
decisions
about
yes
or
no.
D
Should
this
one
be
joined
without
having
you
know,
it
can
have
that
population
count
information
without
having
to
propagate
it
through
the
the
network,
because
it's
it's
operating
locally.
D
There
might
also
be
some
applicability
to
things
like
beer
signaling.
D
I
don't
see
any
reason
that
you
couldn't
do
a
beer
encapsulation
this
way
as
well
or
to
do
it
to
integrate
with
the
the
layer
one
sort
of
rf
channels
for
this.
So
if
you're
gonna
be
sending
something
you
know,
I
think
that
a
lot
of
these
devices
have
some
support
for
mapping
specific
groups
into
an
rf
channel
already.
D
But
I
don't
know
if
there's
going
to
be
any
applicability
to
being
able
to
assign
a
sort
of
you
know
couple
that
inside
the
nats
service,
but
I
think
that's
all
like
tbd
and-
and
I
would
hope
only
to
leave
that
open
as
an
extension
path
not
to
put
that
into
the
first
version
of
the
of
the
service.
So.
B
D
No,
I
was
hoping
that
was
there
somewhere.
I
hadn't
gotten
around
to
reading
it.
B
I'm
wondering
if
I
mean
one
that
can
help
with
the
bgp
free
core,
maybe
because
you
can
get,
though
it
wouldn't
help
your
rpf
problems.
Underneath
you
know,
if
the
next
hop
isn't
it
was
an
lsp
though
there
are
apple.
There
are
implementations
that
do
support
that
I
mean
there's
a
so
there's,
there's
a
there's,
a
thing
called
gtm
which
enables
you
to
do.
D
D
Yeah,
that's
certainly
possible
yeah,
I
mean
I,
I
think
that
the
you
know
if
it's
buggy,
then
you
end
up
not
getting
a
stream
or
maybe
you
end
up
getting
a
wrong
stream,
but
then
the
authentication
fails
if
you're
authenticating.
B
That
was
a
that
was
a
joke.
Gnats.
D
Fortunately,
we're
all
engineers,
so
if
you've
got
a
suggestion
on
a
different
name,
that's
fine
I'd
love
to
have
it,
but
the
the
yeah
any
any
feedback
you've
got.
Please
please!
You
know
think
it
over.
Send
it
to
me,
because
I'm
probably
going
to
write
this.
If
I
don't
have
a
better
idea
just
to
make
it
so
that
you
know
all
the
operators
that
are
interested
have
a
good
path
to
to
participate.
B
Maybe
being
a
little
more
serious,
maybe
mnat
would
be
more
of
yeah
just
because
it's
everybody
knows
what
nat
is,
and
you
know
gnat
sounds
like
you
know,
I
mean
regular
nat
is
unicast
nat
that
we
all
know
and
love
is
global
in
some
dimension,
so
maybe
multicast
nat.
B
D
I'll
give
it
some
thought,
like
I
said
it's
not
written,
so
it's
easy
to
change
names
at
this
point
and
you're
right.
We
can
like
show
down
the
name
all
day,
but
yeah
send
me
suggestions
if
you
think
something
would
be
better
any
any
other
questions.
D
Great,
so
so
my
next
steps
we
are
aiming
to
boil
down
our
our
interested
operators
to
three
to
six
of
them
and
to
try
to
do
some
trials
to
start
this
year.
We
are,
we've
started,
sending
out
our
proposal
for
how
those
trials
would
go
and
we're
aiming
to
get
some
some
buy-in
from
people.
D
The
rough
plan
is
to
use
the
multicast
ingest
platform.
That's
the
same
one
I
presented
back
when
I
started
working
on
the
dryad
stuff
or
right
before
that,
I
think
and
to
add
a
see
back
capability
to
it
so
that
it
would
be
using
the
the
global
you
know,
the
globally
discoverable
dorm
server
and
making
making
choices
about
bandwidth
allocation.
D
At
the
ingest
point
we
would
like
to
to
integrate
with
some
of
the
actual
operator
devices
as
well,
so
that
you
know
they'll,
ideally
we're
hoping
to
get
somebody
who's
willing
to
kind
of
take
that
that
seeback
code
and
port
it
into
driving
their
routers
api
to
to.
We
think
it's
possible
to
do
this
by
just
managing
the
acl
in
the
router,
so
that
see
back
can
cut
off
if
you're
over
can
cut
things
off.
D
If
you're
oversubscribed,
just
by
pushing
an
acl
thing,
we'll
see
if
that
works
out,
we
don't
really
know,
but
that's
kind
of
the
direction
we're
aiming
to
head
at
least
one
of
the
trials
assuming
we
we
do
get
by
n
would
probably
include
a
prototype
of
the
of
the
nats
or
mnat
service
that
we,
we
think
will
do.
I
may
or
may
not
have
to
write
that
before
I
write
the
spec
for
it,
but
you
know
at
some
point
if
that
does
seem
like
it,
it
makes
it
more.
D
Approachable
for
people
will
probably
be
I'll,
probably
be
bringing
that
you
know
a
version
of
that
spec
back
to
this
group
and
then
other
next
steps
also
include
updating
drafts,
obviously
getting
our
chromium
experimental
api
checked
in
and
deferred
for
now,
because
it's
not
on
the
critical
path
to
getting
the
the
operator
trials
in
place
is
ambi,
but
that's
obviously
going
to
be
necessary
before
I
think
before
we
can
even
do
like
an
origin
trial
in
chromium.
D
Just
because
you
know
you
have
to
authenticate
traffic
from
the
network,
if
you're
going
to
run
a
real
browser
and
based
on
the
feedback
from
from
dino,
especially
the
it
might
be
worthwhile
to
to
actually
add
the
alta
work
that
we
were
talking
about
to
flesh
that
out
into
an
actual
wire
line
protocol,
even
if
it's
a
a
specific
profile
of
it
that
just
has
like
signed
packets.
D
You
know
individually
signed
packets
that
contain
the
hashes
that
are
in
ambi,
so
that
we
can
so
that
we
can
do
this
without
having
to
do
a.
D
You
know
a
live
feed
of
the
hashes
that
goes
over
https
and
has
to
do
its
own,
separate
unicast
fan
out,
so
that
that
might
be
something
that
we
want
to
do
just
to
to
sort
of
ease
the
the
deployability
on
that.
D
But
that's
all
gonna
be
probably
after
the
end
of
this
year,
because
I'm
gonna
be
pretty
busy
on
the
trials.
I
think
so
I
think
that's
all.
I've
got
any
other
questions.
B
Well,
just
one
bigger
question
have
have
you
gotten
feedback?
I
mean,
I
think,
the
the
three
drafts
nbc
back
and
dorms.
The
goal
is
to
you
know,
make
multicast
safe
for
the
browser
and
get
past
the
gateway
or
the
gatekeepers.
Have
you
gotten
feedback
thus
far
that
this
is
good
enough?
If
you
know
with
this,
yes,
they
will
allow
multicast
into
their
precious
browsers.
D
Nothing
quite
so
firm
the
the
authentication
story.
I
mean
that
was
all
the
browser.
Folks,
I've
talked
to
that's
the
first
question:
it's
like
how
do
you
authenticate
this
and
when
I
told
them
the
scheme
that
we're
using
in
mb
and
and
tried
to
just
honestly
describe
it
that
was
they
said
that
that
that
would
work
for
authentication
the
you
know.
D
There's
often
some
further
discussion
about
encryption
and
privacy
is
a
sort
of
unknown
still
because
there
is
some
privacy
what
they
call
it:
a
there's,
a
fingerprinting
footprint
to
to
the
the
joining
of
flows
from
a
browser
like
from
a
web
page.
So
this
is
a
concern
that
we're
not
really
so,
I
would
say,
there's
some
risk
on
that
front
as
well.
That's
that's
not
even
technical
risk,
but
sort
of
you
know
policy
risk.
D
D
You
know
if
what
it
might
end
up
being
is
more
like
you
know,
the
browser
will
have
to
ask
permission
in
order
to
join
in
order
to
enable
the
joining
of
multicast
flows
and
have
a
sort
of
you
know
like
the
way
cookies
are
where
you'd
have
a
sort
of
thing
that
pops
up
that
says
hey
this
browser
wants
to
join
some
flows.
This
could
expose
information,
learn
more
here
right.
D
If
that's
what
it
turns
into-
and
I
think
there's
still
value
in
in
doing
this-
and
you
know
similar
to
the
other
kind
of
permission
granted
stuff
it's.
It
would
be
possible
to
sort
of
configure
it
for.
D
For
the
you
know,
the
sites
you
go
to
often
that
kind
of
thing
and
for
people
to
have
different
sort
of
levels
of
protection
that
they're
interested
in
it
also
might
be
possible
that
that
it
would
only
be
necessary
when
you're
in
an
incognito
mode,
and
it
would
be
disabled
entirely
if
you're
on
tour
or
something
like
that
totally
makes
sense
right.
D
But
you
know,
I
think,
there's
a
lot
of
unknowns
on
that
front.
But
those
are
the
main
concerns
I've
heard
raised
with
it
so
far
outside
that
you
know
yeah,
the
authentication
scheme
has
been
reasonably
well
received.
They
think
that
as
long
as
it's
actually
anchored
properly,
it
should
be
okay.
B
So
should
we
be,
you
know,
is
there
a
formal,
you
know
request
for
review,
or
can
we
get?
You
know
those
folks
to
review
these
drafts?
I
guess
you
know
my
my
concern
would
be
you
know
we
go
through
this
process.
We
come
up
with
these
drafts
and
docs
and
we
think
they're
good
and
then
you
know
the
whole
purpose
is
so
that
we
can
get
into
the
browser
party
and
then
the
person
at
the
front
door
of
the
browser
says
no,
that's
not
good
enough.
D
That's
a
good
question,
so
there
is
for
me,
certainly
so
there's
a
sort
of
in
w3c
there's
a
tag
process
like
maybe
maybe
we
should
reach
out
to
a
w3c
liaison
and
ask
for
advice.
That
would
be
a
reasonable,
a
reasonable
idea.
I
think
you
know
I've
been
sort
of
like
there's
this
whole
reasonably
well
documented
path
for
how
individuals
are
supposed
to
try
to
get
things
adopted
by
by
w3c.
But
you
know
I
I
don't
know
if
it's
any
different.
D
If
a
working
group
chair
asking
on
behalf
of
the
working
group
acting
on
behalf
of
a
working
group
at
itf,
makes
a
more
formal
request
for
it
and
provides
a
support
or
if
I
just
sort
of
include
a
note
to
that
effect,
when
I'm
making
the
request,
but.
B
You
know,
let
me
see
how
about
wine.
Do
you
have
a
thought
about
how
best
to
request
review
from
those
folks.
E
Because
I
don't
participate
in
w3c,
but
I
could
probably
break
around
and
find
someone,
but
it
kind
of
sounds
like
proposed.
While
they've
got
an
infinite,
mirror
thing
proposed
idea
of
yeah,
I
don't
know,
but
okay
yeah
I
mean
it
seems
like
people
who
really
participate
in
w3c
should
make
better.
Just
ask
nicely
like
this
is
how
an
individual
does
it?
Do
you
give
any
special
priority
to
the
ietf
or
a
working
group?
I
believe
we
do
have
a
liaison
with
the
w3c.
E
B
D
I
I
think
that's
been
a
risk
the
entire
time,
but
well,
I
think,
like
anything,
you
know
if
I
one
of
the
things
we're
asking
for
with
the
trials
is
permission
to
use
people's
names,
both
the
content
owners
and
the
and
the
network
operators.
D
So
I
think
once
we
get
some
of
those
nailed
down
and
we'll
start
to
be
able
to
say
you
know
here's
the
coalition,
we
have
that's
trying
to
make
this
happen
and
that
you
know
yeah
at
that
point,
maybe
I
would
also
include
you
know
mbondi
asking
in
a
formal
sense
for
support
like
if
I,
oh,
oh
I'll,
if
anybody
says
like
who's
interested
in
making
this
happen,
then
I'll.
D
Well,
I
guess
at
this
point
I
don't
have
any
consensus,
but
besides
adoption
of
the
draft,
so
I
shouldn't
you
know
state
that
go
home
in
the
room.
Sure.
B
Wait
I
thought
we
already
adopted
these
drafts.
What
do
we
have.
D
Attention
so
the
the
question
at
this
point
is
more.
Like
you
know,
I,
I
guess
what
I
I'm
not
sure
I
haven't
done
as
much
with
the
w3c
engagement
process,
as
maybe
maybe
what
we
should
do
is
in
november
have
a
have
a
action
plan
for
making
sure
it
gets
on
the
on
their
list
on
the
tag
list
of
of
things
to
consider
for
real,
you
know,
and
maybe
that
would
include
updates
to
the
drafts
to
make
sure
they're
in
in
good
enough
shape.
D
D
D
Yeah,
but
we
could,
I
can
bring
back
a
report
of
any
any
roadblocks
that
it
would
be
helpful
to
have
formal.
D
You
know
a
formal
consensus
on
you
know
in
november,
if
that
or
or
on
the
list.
If
that
you
know,
as
I
encounter
them.
D
I
feel
like
right
now:
the
barriers
are
more
technical,
there's
still
some
skepticism
as
to
well
this
will
this
actually
work
and
I'm
hoping
that
the
trials
will
address
part
of
that
you
know,
I.
I
don't
think
we're
going
to
get
away
from
the
risk
that
that
people
won't
likely,
but
that's
no
different
than
you
know.
Multicast
was
in
the
first
place.
B
B
B
Okay,
well,
I
would
say
first
of
all,
we
had
awesome
attendance
for
the
last
meeting
on
the
last
day.
I
don't.
I
think
this
is
in
20-something
years
of
going
to
ietfs.
This
is
the
first
time
I've
ever
been
in
a
working
group
meeting
at
the
end
of
on
the
last
on
on
friday
afternoon.
So
maybe
there
are
some
benefits
to
being
virtual
you're
being
you're
going
to
be
stuck
on
a
with
friday
on
the
friday
agenda.
I
guess
it's
good
to
be
a
virtual
friday
agenda.
B
B
Well,
our
european
friends,
have
you
guys
are
ahead
of
us
it's
happy
hour,
for
you
guys.