►
From YouTube: IETF112-MBONED-20211110-1200
Description
MBONED meeting session at IETF112
2021/11/10 1200
https://datatracker.ietf.org/meeting/112/proceedings/
B
All
right,
while
we
wait,
do
we
have
a
volunteer
for
to
be
a
note
taker
who
wants
to
be
the
note.
C
I
do
see
my
slides:
are
you?
Are
you
not
able
to
see
them
in
the
in
the
meeting?
Okay,
let
me
give
you
the
link,
I'm
looking
at.
B
The
I
do
see
them
in
the
meeting
manager,
jake,
okay,
but
where
I
don't
see
them
is
you
know
the
the
little
button
next
to
join
the
queue
slot?
It's
the
share,
slides.
B
That's
exciting
only
your
deck,
though
so
are
you
cool
to
present
yourself
just
share
your
screen.
A
Yeah
another
minute
I
mean
agenda's
pretty
small,
so
this
is
really
the
jake
show.
C
It's
gonna
be
way
shorter
than
my
slot
there.
I
think
so.
B
So
all
right
well
welcome
to
mbondy
for
those
on
the
west
coast.
Thank
you
for
waking
up.
So
early
here
is
the
notewell.
B
Here's
our
agenda
for
today
didn't
render
as
nicely
as
it
did
in
powerpoint.
Any
have
anyone
have
anything
to
add
or
bashing
to
be
done
to
this
agenda.
E
B
All
right
so
we'll
get
started
with
here's.
The
list
of
currently
active
working
group
docs,
the
wi-fi
multicast
problems
draft
after
a
long
long
journey
has
finally
made
it
to
its
fine
final
resting
place
as
our
published
as
rfc
9119.
B
Congratulations
thank
you
for
to
mike
for
getting
us
across
the
finish
line
with
all
those
final
discuss
items
covered
and
that
whole
team,
warren
and
others
who
other
authors
congrats
the
yang
models.
Draft
sandy.
Do
you
want
to
have
anything
to
add
here
sounds
like
you
said:
you
updated
it
based
on
the
yang
doctors
and
a
couple
other
reviews
you'd
like
more
one,
more
review
any
before
a
working
group
last
call
anything
else
sandy.
You
want
to
add
to
that.
B
B
How
you
mentioned
that
there's
a
plan,
it's
it's
related
to
the
iom,
draft
doc,
that's
soon
to
be
at
rfc,
and
he
planned
there's
a
plan
to
update
the
draft
after
that
and
resume
we'll
have
more
discussion
on
that
next
ietf.
B
And
there's
another
doc
redundant,
ingress,
failover
and
yisong
will
provide
an
update
on
that
one
today
and
we'll
probably,
unless
there's
major
objections
begin
a
working
group.
Adoption
call
after
that.
So
that's
everything
so
far.
That
was
all
the
drafts
we
have.
First
up.
B
Let
me
stop
here:
tommy,
you
are
up,
you
can
share
your
screen
or
you
can
share
pre-loaded
slides
the
share
pre-loaded
scroll,
slides
button,
which
is
next
to
the
join
q.
Button
works
reasonably
well.
B
All
right,
so
we
invited
tommy
yeah.
So
let
me
just
give
up
a
little
intro
here,
so
we
invited
tommy
from
apple
to
join
us.
Ios
announced
recently
a
new
entitlement
program
for
s,
applications
that
want
to
utilize
multicast,
and
there
were
some
folks
in
the
working
group
who
are
kind
of
curious
about
this
process.
B
So
we
invited
tommy
on
to
basically
describe
the
the
process
and
you
know
what
what
led
to
it
and
any
experiences
they've
been
having
at
apple
with
support
of
multicast.
So
with
that
tommy
I'll
turn
it
over
to
you.
F
All
right,
thank
you,
and
this
should
be
a
pretty
quick
update
here,
but
you
know
we
do
have
time
in
the
agenda,
so
I'd
be
happy
to
have
discussion,
questions
around
this
and
suggestions
of
what
you'd
like
to
see
going
forward
all
right.
So
what
are
we
talking
about?
F
So
specifically,
this
new
entitlement
when
an
entitlement
for
people
who
are
not
familiar
with
it
on
at
least
mac
os
and
ios
is
essentially
a
privilege
that
has
to
be
signed
into
the
binary
of
the
application
when
it's
before
it
can
be
released
on
the
app
store
and
some
entitlements,
you
can
just
kind
of
check
a
box.
This
is
one
that
you
have
to
request,
so
it's
a
little
bit
heavier
weight
anyway.
F
This
is
part
of
a
general
suite
of
changes
that
we
did
to
add
restrictions
around
what
you
could
do
on
the
local
network,
and
these
are
all
aimed
at
trying
to
improve
user
privacy
from
various
types
of
attacks,
and
so
this
was
introduced
in
ios
14.
So
that
means
we
announced
this
in
our
developers
conference,
which
occurred
in
june
of
2020.
F
So
a
year
a
year
and
a
half
ago
and
the
release
came
out
over
a
year
ago
at
this
point,
so
this
has
kind
of
been
the
reality
for
the
last
little
bit
and
there
are
several
different
changes
that
were
part
of
this
local
network
privacy
feature
overall,
first,
all
apps
that
are
communicating
with
hosts
on
their
local
network,
whether
that's
unicast
multicast
using
apis
that
will
trigger
local
network
interactions.
F
F
The
user
just
has
to
give
permission.
This
really
makes
a
lot
of
local
network
interactions
more
consistent
with
other
kind
of
nearby
device
interactions
that
required
a
similar
privacy
permission
such
as
connecting
to
a
bluetooth
device,
etc.
So
a
lot
of
this
is
about
consistency
and
user
awareness.
F
The
next
category
of
change
is
that
you
know
most
of
the
multicast
discovery
that
occurs
on
ios
devices
is
using
system
apis
for
multicast
dns,
using
dns
service
discovery
on
apple
platforms,
that's
branded
as
bonjour
and
any
app
that's
using.
That
must
declare
kind
of
at
compile
time
the
static
list
of
all
of
the
different
service
types
that
they
plan
to
use.
So
they
can't
use
any
service
type.
That's
outside
of
that
list.
They
can't
just
start
doing
wild
card
discovery.
F
And
then
the
third
category,
which
is
the
one
that
we're
talking
about
here,
is
specifically
for
multicast
and
broadcast.
That's
not
using
any
system
api
like
what
we
have
for
multicast
dns.
An
app
must
have
an
entitlement,
because
this
is
a
kind
of
even
more
privileged
and
powerful
operation.
F
F
And
so
in
general,
you
would
see
when
we
turn
this
on
first
and
just
to
say,
hey
what
lights
up?
What
starts
scanning
your
network
on
different
ports
and
on
different
addresses?
F
It
was
quite
a
lot
of
apps
doing
it
very
frequently,
and
you
know
some
of
these
can
be
legitimate
use
cases,
but
a
lot
of
them
also
could
be
efforts
to
detect
what
devices
were
around.
What
types
of
things
would
respond
on
the
local
network
to
create
a
signature
of
the
local
network
that
the
user
is
on.
Currently
we
had
for
years
before
prevented
apps
from
getting
directly
at
some
of
the
ssid
or
bssid
information
to
know.
F
F
Now,
in
general,
you
know,
as
I
mentioned,
for
bonjour,
we
kind
of
started,
preferring
apis,
where
you
could
list
a
specific
service
type.
So
this
guarantees
that
the
app
developer
is
aware
of
what
they
are
doing.
So
you
know
if
I
have
an
app
that's
trying
to
talk
to
a
printer
or
I'm
trying
to
discover
a
tv
that
I
can
display
something
on
indicating
at
the
app
compile
time.
F
Yes,
this
is
the
type
of
service
I'm
trying
to
do,
or
maybe
I
have
a
custom
service
for
my
own
special
devices
or
I'm
doing
some
peer-to-peer
game
if,
if
they
can
specify
here's
the
exact
scope
of
what
I
need
to
do,
it's
much
easier
to
guarantee
that
there's
not
other
code
running,
that's
doing
nefarious
things
or
trying
to
do
a
full
address
and
port
scan
of
all
devices
around.
F
And
then
the
other
aspect
here
is
that.
F
There
are
uses
of
multicast
that
we
see
as
kind
of
the
most
prevalent
as
we've
been
granting
these
entitlements.
By
far
the
most
common
is
oh
yeah.
I
just
want
to
use
ssdp,
which
is
you
know,
pretty
old
protocol.
Not
standard
has
a
lot
of
security
vulnerabilities,
and
this
is
also
an
opportunity
to
tell
people
hey
if
you're,
just
trying
to
do
this
discovery
going
and
using
a
system
api
that
is
more
audited
and
using
a
standard
protocol
like
multicast
dns
is
going
to
be
a
better
route
if
you're
building
a
new
solution.
F
So
going
forward,
you
know
apps
that
do
have
legitimate
reasons
to
need
custom
multicast
can
get
approved.
They
just
need
to
fill
out
a
form
to
say
what
are
they
trying
to
do
and
almost
everything
does
get
approved
generally.
What
wouldn't
get
approved
is
if
you
cannot
sufficiently
explain
like
what
address
and
ports
are
you
trying
to
use
like
what?
What
what
is
your?
Why
does
your
app
have
to
do
this
or,
if
they're,
trying
to.
F
The
common
sdks,
you
know
like
what
we
have
with
you
know,
there's
google's
casting
sdk.
F
F
So
just
some
other
resources.
We
have
our
2020
presentation
on
local
network
privacy
in
general
and
then
probably
more
interesting.
Is
this
separate
article
we
put
about
out
about
how
to
use
multicast?
F
C
Yeah
hi
tommy,
thanks
for
thanks
for
this,
I
did
have
a
couple
of
questions.
So
the
mdns
you
you
seem
to
say
it
was
bonjour
or
dnssd,
but
does
this
distinguish
between
like
just
using
a
dot
local
name
that
also
generally
uses
mdns.
F
Right,
so
if
you
use
a
dot
local
name,
that
does
not
require
any
special
declaration
of
what
service
types
you're
using
the
special
declaration
is
just
for
browsing
since
that
was
more
like
okay,
you
know
people
are
using
this
to
scan
and
you
know
there
definitely
are
so.
You
have
to
also
apply
for
the
entitlement
if
you
want
to
be
able
to
do
wild
card
launcher-
and
there
are
you
know,
utilities
that
are
just
about
scanning
every
device
around
you
and
reporting.
F
This
is
what
your
local
network
looks
like,
but
we
recognize
that
those
are
usually
pretty
niche
products.
C
So
so,
there's
nothing
that
prevents
scanning
just
according
to
a
bunch
of
dot
local
names,
for
example.
Or
is
that
just
a
sort
of
separate.
F
Configuration
I
mean
yes,
you
could
try
to
brute
force
like
every
possible
string
in
the
world.
No
note
or
known
devices.
F
C
C
My
second
question:
maybe:
are
there
any
differences
between
ssm
and
asm
or
between
globally
or
locally
scoped,
addresses.
F
No,
no
we're
not
doing
any
distinction
there.
Currently,
okay.
C
And
then
my
last
question
was:
do
you?
Are
there
any
sort
of
known
changes
you
anticipate
to
this
going
forward
or
you
think
this
kind
of
covers
the
the
sorts
of
efforts
you've
seen
and
and
nothing
else
would
be
necessary.
F
C
Sure
I
can
answer
for
myself.
At
least
there
was
a
an
article
forwarded
where
someone
was
kind
of
bemoaning
the
extra
bureaucracy
for
for
just
you
know,
experimenting
with
multicast
in
an
app
development
sense.
Saying
kind
of
multicast
is
hard
enough.
C
I
I
think
that
had
some
sympathetic
ears
on
the
list
when
it
was
forwarded,
but
you
know
I
mean,
we've
been
working
on
the
on
the
security
and
and
noting
the
I
don't
know.
If
you
saw
the
draft
insect
dispatch
right
after
your
slot,
but
yeah
there
are
some
sort
of
potential
exposures.
C
C
F
Yeah,
I
I
know
that
you
know
there
are
a
handful
of
people,
who've
requested
to
have
kind
of
complained
for
just.
F
F
You
know
oftentimes
if
there
are
requests
for
just
like.
Oh
I'm
doing
experimentation.
Is
it's
often
just
to
discover
like
peer-to-peer
like
hey
where's,
the
other
iphone
around
me,
and
for
that
I
think
we
try
to
steer
people
to
saying:
hey,
just
use
the
system
apis
for
discovering
your
app's
service
nearby.
That
works
just
fine.
Why
are
you
trying
to
invent
a
brand
new
discovery
protocol?
That's
not
dns
service
discovery.
It's
not
ssdp,
but
you're
like
rolling
your
own
thing,
like
that's
you're,
probably
making
it
harder
for
yourself.
C
C
B
So
I
think
I
understand
the
entitlement
process
for
for
apps
as
you're
sourcing,
so
as
just
as
a
user,
if
you
want
to
join
a
multicast
group,
there's
no
entitlement
for
that
or
do
you
just
get
a
pop-up
that
says,
do
you
really
want
to
use
multicast
or
so
can?
Can
you
just
explain
what
that
looks
like
for
not
the
app
developer,
but
the
app
user?
Is
there
anything
on
their
part
that
they
need
to
do
or
that
just
works.
F
F
F
B
Gotcha-
and
did
you
mentioned
this
this
you
this
isn't
a
your
your
act?
Can
you
give
any
kind
of
experience
in
terms
of
you
mentioned?
Are
there
lots
of
apps
that
have
applied
for
this
entitlement?
Did.
G
You
say
you
guys
haven't
denied
anyone.
No,
I
mean
some
people.
F
There
there
have
been
some,
we
have
not
approved
the
vast
majority
get
approved
because
the
vast
majority
are
legitimate,
gotcha
yeah
I
mean
it's,
it's
a
bunch.
I
mean
I,
I
think
it's
thousands.
A
Great
ask
a
question
there
too,
and
so
the
I'm
not
looking
for
names,
but
those
that
you
said
that
were
not
approved
is
this
in
in
the
general
case,
because
you
can
point
them
in
the
direction
of
something
already
exist,
rather
than
reinventing
something
exactly.
F
D
D
A
Compliance
button
you
used.
C
Actually,
no
that's
that's
the
request
screen
share.
Okay,
told
me
at
the
beginning
there
that
that,
for
whatever
reason
my
particular
slides
were
not
in
the
pre-loaded
list.
For
some
reason
I
was
probably
late
right.
So
thanks,
I
have
just
a
short
update
this
time
in
the
notes,
I
added
a
link
to
my
last
presentation
at
111.
C
If
you
do
need
any
background
on
these
drafts
and
kind
of
what
they're
doing
I'll
refer
you
to
to
take
a
look
at
that
one.
Instead,
I
don't
want
to
run
through
the
overview
of
them
each
time,
so
I'm
just
going
to
skip
that.
C
Mostly.
What
I'm
going
to
cover
is
the
sort
of
progress
in
that's
relevant
to
this
work.
That's
been
happening
externally,
though,
I
will
touch
on
the
relatively
small
amount
of
forward
progress
on
these
docs
and
and
discuss
that
a
little
bit
and
then
I'll
go
over
kind
of
what
my
next
steps
are
in
this
direction.
C
So
the
first
thing
is
that
the
multicast
community
group
has
been
meeting
regularly.
We
did
some
some
stuff
at
tpac.
C
Actually
I
probably
should
have
posted
the
the
link
to
the
demo,
video
that
we
put
up,
but
that's
that's
there
in
the
list
of
demo
videos
that
they
had
for
the
for
the
tpack
meeting
under
the
web
and
networks
interest
group,
the
web
networks,
interest
group
in
particular
had
some
nice
things
to
say
about
about
the
multicast
community
group
and
the
multicast
use
case
that
they
presented
during
their
tpac
meeting.
C
I,
like
the
slides
there,
there's
a
good
there's,
some
sort
of
openness
to
this
in
the
web
community.
The
most
compelling
part
of
that
seemed
to
be
about
the,
or
at
least
the
one
that
got.
The
most
discussion
was
its
impact
on
some
of
the
sustainability
efforts
that
are
getting
started,
and
it's
it's
relevance
to
that.
C
To
that
point,
which
I
I
also
highlighted
in
some
of
the
stuff
that
I
talked
about
there
and
I
also
had
a
presentation
with
the
web
transport
working
group.
It
was
a
sort
of
minor
part
of
their
of
their
overall
meeting,
but
I've
opened
an
issue
requesting
that
they
add
multicast
as
a
as
a
use
case
that
they
would
consider
valuable,
which
would
be
relevant
feedback
to
bring
to
the
ietf
web
transport
working
group
and
and
other
relevant
parties
that
is
under
discussion.
C
I've
had
a
follow-up
meeting
with
the
chairs
and
and
they'll
be
discussing
it
in
the
working
group
to
see
what
they
think
about
that,
but
that
is
sort
of
necessary
steps
toward
moving
it
to
a
browser
use
case.
I
also
presented
yesterday
insect
dispatch
about
so
the
background
on
this.
As
I
think
I
went
over
before,
but
I'll
I'll
recap
it
here.
C
We
we
submitted
an
intent
to
experiment
to
chromium
their
feedback.
Was
it's
not
ready
in
particular,
that
it
needs
confidentiality
as
well
as
the
integrity
and
authenticity
that
that
was
part
of
the
design?
C
C
But
that's
that's
going
to
be
some
work,
but
but
really
the
the
sticking
point
was
more
about
the
design
before
even
getting
to
that
point,
and
so
confidentiality
with
multicast
comes
with
some
complexity,
we
tried
to
capture
the
complexity
in
this
in
in
the
draft
k
rose,
sex,
dispatch,
multicast
security
and
we
are
looking
for
a
home
for
it,
so
insect
dispatch
their
the
recommendation
for
dispatch
was
that
we
start
a
mailing
list,
it's
to
be
determined,
whether
that's
going
to
be
the
msec
mailing
list
for
the
previously
existing
but
closed
multicast
security,
working
group
or
whether
that'll
be
a
new
mailing
list,
and
that
I'm
supposed
to
follow
up
on
and
we'll
be
sending
an
email
today
to
kick
that
off.
C
Hopefully,
that
mailing
list
will
be
like
in
place
by
the
end
of
the
week.
I
I
I
hope,
but
we'll
see
and
we'll
see
if
we
can
get
any
security
experts
on
it
to
really
dig
in
on
this.
We
did
get
some
discussion
from
ecker
on
the
list,
he's
sort
of
against
the
general
idea
and
is
making
the
case
that
nobody's
particularly
interested,
and
so
there's
no
reason
to
look
at
it,
but
the
objections
he
raised
to
it
were
at
on
the
technical
side
were
not
particularly
compelling.
C
I
was
not
able
to
give
him
compelling
evidence
that
anybody
is
interested
either,
so
there
might
be
more
such
objections
if
we
kind
of
get
to
that
stage,
but
I
thought
the
discussion
overall
was
pretty
good.
I
think
he
had
some
some
good
points
of
feedback
that
we've
started
incorporating
into
the
draft
to
cover
it.
C
But
this
is
this:
this
security
draft
is
not
just
about
the
web
safety
model,
the
web
security
model,
as
it
relates
to
multicast
it's
you
know
we
we
intend
to
do
both
web
and
non-web
traffic
at
akamai.
Assuming
this,
you
know
actually
reaches
a
deployment
stage,
and
you
know
at
the
moment
it's
looking
likely
that
it
will
start
with
the
non-web
traffic
where
we
have.
C
You
know
we'd
be
relying
on
on
receivers
that
have
some
deployment
footprint
and
this
would
include
potentially
some
of
the
gaming
platforms
for
the
game
downloads
and
also
some
of
the
sort
of
smart
tv
on
a
stick
platforms
that
that
some
of
our
potential
partners
have
some
deployment
footprint
with
and
and
that
doesn't
have
to
be
web
traffic.
Although
being
a
part
of
a
web
api
and
doing
the
same
thing
in
all,
the
cases
might
actually
be
kind
of
handy,
but
not
strictly
necessary.
C
So
we
might
be
able
to
continue
moving
forward
with
some
of
that,
and
I
think
the
target
still
has
to
include
the
browser,
at
least
from
our
point
of
view,
for
this
to
be
worthwhile
in
the
end,
because
there's
such
a
you
know
some
of
the
peaks
that
we,
the
peak
traffic
days,
that
we
hope
to
address
with
this
work
come
from
web
video
traffic
and
we
we
do
need.
We
do
think
we
need
to
address
it,
but
the
the
sort
of
next
steps
on
the
mailing
list
front.
C
With
a
security
question,
would
I
mean
the
the
target
I'm
I'm
aiming
towards
ultimately
is
well.
Ultimately
is
rfc
is
with
consensus
behind
them,
but
to
get
there
it
would
take
a
buff
to
open
a
working
group
again
either
the
existing
mtech
or
a
new.
C
You
know
maybe
broadcast
media
multicast
security
or
some
such
I
don't
know,
but
because
it's
a
slightly
different
security
model
than
the
previous
msec
work
did
so
it
would
either
have
to
come
with
a
recharger
or
it
would
have
to
be
a
new
working
group.
I'm
not
sure
what
the
what
the
right
answer
will
be
there
and
that'll
be
up
for
discussion.
C
I'm
not
sure
it'll
even
happen.
It's
going
to
depend
on
security,
people
agreeing
that
this
is
worth
looking
at
and
that
they'd
be
willing
to
to
contribute,
but
we'll
try
to
get
that
discussion
going
between
now
and
march,
or
whenever,
however
long
it
takes
and
and
hopefully
get
somewhere,
but
that's
kind
of
the
what
it's
going
to
take
to
get
into
browsers.
C
It
looks
like
because
the
the
sort
of
not
encrypted
at
the
at
the
level
that
the
browser
would
be
maintaining
the
encryption
is
sort
of
a
non-starter.
C
According
to
the
the
web
security
review
that
we
saw
the
the
sort
of
the
apis
we're
looking
at
are
listed
in
the
multicast
community
group
charter.
In
the
w3c
thing,
I
think
I
had
a
link
to
the
github
there,
which
has
a
link
to
the
page.
Anybody
can
join
that
group
by
the
way
it's
not
restricted
to
to
people
whose
companies
are
paying
w3c
membership
fees
and
whatnot.
C
I
I
should
say
anybody
whose
company
permits
them
there
is
a
contributor
license
agreement,
but
you
can
as
an
individual
if
your
company
allows
it
make
a
make
a
free,
w3c
account
and
join
the
group.
So
please
do
if
you're,
willing
and
able.
If
your
company
is
a
member
of
w3c,
then
you
might
have
to
get
approval
from
somebody
who's
who's
sort
of
heading
up
your
participation
with
w3c,
but
for
most
participants
that
hasn't
been
too
big
an
issue.
C
I'd
love
to
have
anyone
who
was
interested
enough
to
come
to
this
meeting
might
be
interested
enough
to
follow
this
too.
I'm
particularly
interested
in
people
who
would
be
interested
in
doing
the
testing
and
implementation
work
as
that
as
that
progresses,
but
that'll
be
a
slightly
faster
cadence
for
updates
than
the
then
the
mbondi
updates.
C
Let's
see
the
some
of
the
early
feedback
that
we
saw
from
the
chat
insect
dispatches
is
suggesting
that
we're
probably
going
to
need
to
have
the
the
actual
protocol
proposal.
So
for
that
we're
I'm
intending
to
start
those
experiments.
You
know
towards
the
end
of
this
week
and
and
to
get
these
rolling
in
the
in
the
coming
months.
C
C
There
are
likely
to
be
some
differences
with
that,
because
I
want
to
make
sure
to
support
datagrams
to
make
it
such
that
existing
multicast
apps
can
be
deployed
with
just
the
sort
of
web
transport
shim
around
them
at
the
at
the
sending
side
and
then
received
by
the
browser.
That's
what
I'd
I'd
like
to
aim
for.
C
So
the
doc
status
for
the
mbondiwork
I've
been
engaged
on
some
some
internal
stuff.
In
my
in
my
day
job
and
on
some
of
this,
some
of
this
you
know
w3c
and
and
working
out
the
security
stuff.
So
I've
been
kind
of
low
activity
on
the
drafts
that
are
active
here,
but
I
did
do
some
updates
in
dorms.
C
I
I,
when
I
checked
it
against
the
reviews.
I'd
had
seen,
particularly
particularly
rashad's
as
young
doctor.
There
was
an
unaddressed
comment
that
I'd
left
out,
so
I
added
some
text
to
address
that.
So
I
think,
as
of
now,
it's
got
all
the
all
the
updates
that
people
have
recommended.
C
If
anything
has
been
left
out
that
somebody
mentioned,
then
please
do.
Let
me
know
if
you
see
it,
it
probably
would
make
sense
to
get
at
least
another
review
if
possible.
But
you
know
I
would
love
to
move
this
to
last
call
at
some
point
and
have
that
you
know
get
the
the
forward
motion
that
would
that
would
make
the
other
two
that
build
on
it
more
considered
more
stable.
C
Likewise,
the
mnat.
I
got
a
few
comments,
but
nothing
really
telling
me
to
change
things.
I
think,
and
so
I
would
love
to
get
reviews
on
that,
one
in
particular
as
well
ambi,
I
think
I'm
gonna
be
I'm
gonna
be
messing
with
as
I
work
on
the
security
stuff.
C
So
so,
if
you're
looking
to
hold
off
on
something
you
can
hold
off
on
that
one
once
storms
is
done,
then,
then
see
back
would
be
certainly
a
useful
one
to
do
as
well,
but
my
priority
order
on
reviews,
if
you're
willing
to
do
them,
would
be
dorms
and
m-net
and
then
see
back
and
then
ambi
at
this.
At
this
stage,.
C
Let's
see
the
in
in
m
I'll
highlight
dnssd
in
particular
the
the
the
search
domain
portion
of
that
as
the
as
the
part
of
the
architecture.
That
worries
me
the
most
because
it
proposes
to
have
a
a
mnat
server,
that's
discoverable
by
clients
inside
their
home
networks
and
via
dns
sd,
and
this
requires
probably
passing
the
search
domain
as
a
as
a
dhcp
option
into
the
home
network,
which
I
I
am
not
seeing.
C
Currently
I
mean
I
haven't
done
a
wide
review,
but
currently
it
anecdotally
does
not
seem
super
common
that
home
routers
are
are
implementing
that
functionality
today
it
might
be
because
the
service
providers
aren't
aren't
advertising
in
their
dhcp
offerings,
but
this
is
a
space
that
that
I
don't
think
has
been
necessarily
that
that
well
explored
and
and
would
bear
some
some
investigation
and
and
possibly
alternatives.
The
other.
The
main
alternative
I've
been
thinking
about
is
is
making
a
dns
rr
type
that
would
describe
an
sg.
C
C
It's
more
drafts
than
I
was
expecting
when
I
started
anyway
in
terms
of
next
steps.
C
I'd
like
to
consider
dorm's
last
call
so
I'd
like
to
open
that
up
for
discussion
now.
The
other
thing
I'm
going
to
ask
for
is
to
put
dormancy
back
and
ambi
into
a
cluster,
the
cluster
being
so
that
they
can
refer
to
one
another
and
would
would
become
rfc's
together
or
in
the
order
and
order
determined
by
by
their
lack
of
loops,
and
then
there
are
some
new
docs
that
are
going
to
be
relevant
to
this.
C
So
if
you
know
anybody
who
is
willing
to
talk
about
this
stuff
then
make
sure
to
get
them
on
the
list.
When
I
send
out
the
invitation
to
join
the
msec
or
whatever
it
will
be.
So
with
that,
do
please
give
me
any
questions
or
comments
that
you
cut.
H
C
I
wouldn't
expect
so
mld
and
pin
and
beer.
I
would
expect
to
be
unaffected.
C
In
fact,
I
would.
I
would
hope
that
we
could
even
consider
so
the
trans,
the
the
security
considerations,
the
sort
of
initial
direction
that
I
have
not
yet
written
down
and
is
highly
subject
to
change.
Is
that
going
to
be
trying
to
make
a
a
quick
version
of
this?
That
uses
a
symmetric
key
for
encryption
and
maintains
the
same
kind
of
out
of
band?
C
The
unicast
based
secure
transfer
of
hashes
to
do
the
to
do
the
authentication
of
the
data
channel
of
the
multicast
data
channel
and
and
the
main
difference
with
what
it
was
before,
which
was
just
authenticating
generic
udp
packets,
which
may
or
may
not
be
encrypted,
is
that
the
udp
packets,
for
this
part
would
be
sort
of
specified
as
being
quick
encoded
with
a
with
a
with
a
symmetric
encryption
key
that
is
passed
over
an
associated
quick
channel.
C
These
the
intent
is
to
make
it
so
that
so
that
existing
udp
payloads
from
outside
from
like
existing
multicast
applications,
would
be
possible
to
translate
into
these
udp
packets
as
as
quick
datagrams
and
be
transported
with
multicast
that
way,
so
that
the
applications
themselves
even
legacy
applications
would
not
have
to
change
at
all.
C
We'd
have
to
do
is
run
a
receiver
sort
of
inside
the
sender
that
would
pump
out
the
quick
packets
and
the
authentication
hashes,
and
the
authentication
hashes
would
have
to
be
distributed
to
the
to
the
receivers
in
a
much
smaller
channel
than
than
the
data
channel
in
the
same
way
that
ambi
works.
But.
C
In
terms
of
the
the
transport
under
udp
under
the
the
quick
layer
of
udp,
there
would
be
no
change
and
so
anything
that
that
works
today,
mpls
beer
pim.
All
these
things
would
be
unaffected.
E
B
One
of
the
pieces
of
feedback
was
or
a
gating
factor
was-
and
I
forget
who
you
mentioned,
which
group
that
know
that
there
is
a
lack
of
demand
for
this
is
that
is
that
pertinent
from
a
technical
standpoint,
and
that
seems
like
a
subjective
discussion
and
and
also
one
that
would
fly
in
the
face
of
the
existence
theorem
that
you
indeed
have
enough
interest
to
be
doing
all
of
this.
So
can
you
so
so?
C
Yeah,
if
you
can't
get
enough
interest,
then
they
won't
open
a
working
group
for
you
in
general,
but
with
that
said,
the
specific
case
here
was
with
regard
to
the
multicast
security
for
browsers
in
particular,
the
main
question
was
like:
well,
what
browsers
are
going
to
take
this?
They
came
from
ecker.
He
asked
that
both
on
the
list
and
in
the
in
the
cube,
insect
dispatch
and
made
the
point
that
I
mean
yeah.
C
I've
been
talking
to
some
chromium
people,
but
they're
not
eager
for
this
and
seem
pretty
lukewarm
on
it.
You
know
I
I
would.
I
would
like
to
get
some
counter
examples
to
speak
up
and
and
and
say
that
I
mean,
I
think
you
know
my
response
to
his
to
what
he
said
was
that
well,
they
think
it's
probably
not
going
to
work.
C
They
are
skeptical
of
this,
and-
and
this
is
perhaps
a
justified
position
based
on
multicast
history-
they
want
to
see
you
know
more
people
engaged
on
this
before
they're
willing
to
to
dive
in.
I
assume
and
that's
some
feedback.
I've
gotten
from
others
in
the
in
the
multicast
community
group.
So
I
think
increasing
engagement
would
really
show
something
worthwhile
here.
C
I
I
should
also
maybe
mention
you
know,
I'm
starting
to
get
a
little
bit
more
traction
in
this,
like
virgin
media
previously
had
given
us
permission
to
to
say
their
name
is
having
been
involved
in
our
our
trials
and
evaluated
the
tech
and
supportive
of
moving
it
forward
since
then,
or
since
last
time,
I've
gotten
so
comcast
said
that
they
would.
They
think
this
makes
sense
for
the
future.
C
C
Likewise,
I've
I've
gotten
so
dizone
sent
an
email
to
the
list,
but
wasn't
it
wasn't
forwarded
to
the
sexist
patch
list?
But
I
do
feel
comfortable
saying
I
I
think
I
will
probably
be
forwarding
it,
but
they.
They
also
said
that
they're,
supportive
of
multicast
standardization
amazon
also
said,
did
give
permission
to
say
that
we'd
talk
to
them
and
that
that
they
would
like
to
see
multicast
capabilities
happen
in
news
that
may
or
may
not
be
related
design
and
amazon.
C
Each
individually
have
signed
major
major
sports
delivery
deals
for
the
upcoming
years.
In
the
like,
you
know:
10
billion
and
2.7
billion
dollars
in
euros
respectively.
That
they've
publicly
announced-
and
you
know,
and
suddenly
they're
interested
in
multicast
for
some
reason,
or
at
least
tentatively
interested.
So
I
mean
you
know,
will
it
happen?
C
Who
knows
right?
But
but
but
these
are
names
that
that
are
not
nothing.
I
did
not
get
a
chance
to
say
that
in
sex
dispatch
my
time
was
pretty
short,
but
but
we'll
probably
be
discussing
it
a
little
more.
C
B
You
mentioned
one
of
the
other.
Concerns
was
integrity.
Does
that
mean?
Is
that
where
you
need
this,
you
need
to
spin
up
a
new
working
group,
or
is
that
just
another
draft
that.
C
C
C
You
know
novel
crypto
or
authentication
approaches.
These
are
all
well
proven
in
other
contexts,
and
so
as
long
as
we
and
it's
not
something
that
that
has
like
tricky
trade-offs
on
crypto
agility,
we
just
have
a
very
you
know,
extensible
sort
of
framework
for
for
how
to
identify
which
algorithms
we're
talking
about
via
the
yang
model,
and
so
the
you
know
it's.
C
It's
there's
not
many
concerns
on
the
security
side
with
the
with
what
was
proposed,
but
I
think,
having
worked
on
it
since
then,
there
might
be
some
subtleties
in
like
the
way.
The
the
course
model
works
that
I'd
like
to
get
some
more
security
expert
eyeballs
on
really
before
before
I'm
comfortable
with
it.
C
So
you
know-
and
you
know
that
that
was
also
part
of
the
advice
that
we're
gonna
need
to
have
a
security
review
of
it,
but
I
do
think
it.
It
would
make
better
sense
if
there
was
an
existing
msec
working
group
to
live
inside
the
msec
work
working
group
once
it's
there,
so
so
I
I
might
propose
moving
it
if
possible
at
that
time,
but
you
know
ju,
that's
just
a
heads
up
for
the
future.
There's
nothing
actionable,
I'm
suggesting
at
the
moment.
B
C
Yeah,
it's
an
ambi
and
there
might
be
something
very
similar
that
would
go
into
the
quick
protocol
itself.
That
would
be
a
a
little
more
integrated
instead
of
purely
out
of
band,
but
I
think
I'm
gonna
do
it
by
just
referring
to
ambi.
I
just
I
need
to
to
experiment
with
that
part
and
actually
write
it
and
and
write
it
up.
B
Okay,
frederick
and
saba
and
frederick
you
guys
are
up
so
you
can
choose
to
you,
can
choose
to
share
your
screen
or
you
can
click
this
button
of
share
preloaded
slides.
J
I'm
sorry
so
thanks
for
for
the
invitation,
my
name
is
frederick
louis
and
I'm
here
with
mate,
who
are
working
on
behalf
of
association,
which
is
the
pan-european
company
managing
the
pan-european
network.
So
we
are
here
today
to
talk
about
an
exciting
project.
We
have
started
in
2019,
which
is
related
to
implement
a
routine
stack
based
on
various
data.
J
J
So
this
is
the
the
agenda
very
quickly.
We
will
explain
what
is
the
rar
free
router
project,
the
lessons
learned
regarding
brfc
implementation,
the
the
first
br
experiment
that
has
been
launched
in
2017
and
that
has
been
completed
in
2021.
J
The
next
point
is
the
amt
relay
implementation
based
on
rfc
7450,
and
since
today
we
have
a.
We
have
a
very,
very
exciting
br,
combined
to
int
implementation,
experiment
that
you
you
can
that
you
can
test
by
yourself
and
listen
the
the
the
hunger
and
harvard
music
and
a
bit
of
conclusion.
Next.
J
So
basically,
what
is
the
the
r
project?
It's
essentially
combining
three
components,
which
is
the
control,
plane,
multiple
data
planes
and
create
a
specific
interface
between
each
data
plane
and
the
control
plane.
So
so
the
goal
is
to
have
one
familiar
platform
that
is
able
to
propose
multiple
solutions,
and
each
solution
will
address
specific
research
and
education
use
case
and
the
the
the
fun
part
is
that
multicast
is
one
of
this
use
case.
Next.
J
So,
for
now
we
we,
we
have
multiple
data
planes
based
on
p4.
We
have
the
vm
v2
target,
which
is
the
virtual
p4
switch
where
you
can
train
yourself
and
look
at
the
r
free
router
code.
We
have
an
implementation
for
tofino,
which
is
basically
essentially
the
same.
The
same
code
base
except
the
fact
that
you,
you
will
see
more
pragma
restraining
features
because
the
the
ace
the
tofino
asic
is,
of
course,
not
unlimited
in
terms
of
resource
compared
to
a
virtual
switch.
J
We
are
looking
at
fpgas
and
which
will
come
soon
next
year.
We
have
a
backup,
tcp,
damper
library
and
also
a
dpdk.
We
have
some
discussion
with
in
order
to
pro
to
provide
the
npa
letterplane
next.
J
So
how
do
we
test
the
the
the
code?
Essentially
we
we
have
deployed
for
a
set
of
four
nodes
in
in
in
europe,
amsterdam,
frankfurt,
budapest
and
poison,
and
so
far
we
we
attracted
some
attention
to
various
organizations
from
around
the
world,
and
now
we
we
are
pleased
to
welcome
more
nodes
in
north
america
and
south
america.
We'll
have
some
new
nerds
in
in
indonesia,
in
korea,
for
example,
but
and
all
of
this
corresponds
to
a
beer,
an
entire
beer
domain
that
that
is
spread
onto
the
the
p4
lab.
J
Now
I
think
it's
about
the
floors
is
yours,.
I
I
How
can
it
transmit
my
my
frequently
broadcast
music
sessions,
and
this
is
this
is
what
we
mentioned-
that
the
the
building
started
in
in
2017
or
or
about
about
the
time
then,
when
this
refc
became
public.
I
For
for
this
to
happen,
we
did
obviously
the
interior
gateway
protocols
to
to
have
the
the
beer
load
index
and
the
b
mpls
labor
base
flooded
within
the
domains.
So
this
is
via
b
right,
size,
extension
and
b,
ospf
extension
as
a
started
to
appear
in
inframatter.
I
This
is
again
the
same
2017
when
it
happened.
I
This
is
the
point
where
I
would
call
you
if,
if
you,
if
you
implement
a
networking
operating
system
and-
and
you
would
like
to
and
and
you
have
beer
and
would
like
to
do,
some
interoperability
testings
then
feel
free
to
contact
us.
We
are.
We
are
very
happy
to
do
some
hacking
with
you.
The
thing
is
that
beerus,
spf
extension
have
already
interrupt
tested
with
isis
extensions.
Fortunately,
the
viral
shark,
decoded
it
very
well
but
not
yet
tested
in
interrupt
times
regarding
the
beer
bgp
draft.
I
It
is
also
supported
since
the
beginning,
when
it
when
it
started
to
appear
it
is.
It
is
basically
when
you,
when
you
send
the
prefix,
then
you
cannot
touch
a
beer
attribute
to
it,
then
that
prefix
will
behave
as
as
it's
s
as
it
saw
in
the
in
the
igp
domains
with
with
the
b
index.
I
That
is,
you
can
and
you
can
send
multicast
pockets
to
that,
that
prefix
or
or
parts
of
the
prefix
and
and
that
that
will
happen
in
beer
and
constellation
in
this
case
also,
we
support
the
the
requisite
version
of
this.
That
is,
you
get
a
prefix
without
would
be
attribute,
but
but
the
underling
root
have
a
beer
index.
Then
the
pockets
also
we
sent
in
in
the
rank
of
selection
and
and
the
last
thing
that
we
we
support
with
beer
things.
I
That
is
a
appear
at
this
p
in
a
in
a
b
rank
of
session.
It
is
needed
for
me
just
because
this
is
how
I
I
am
with
cast
my
my
music
in
the
so-called
global
table,
and
last
but
not
least,
this
year,
we
we,
as
we
started
discussing
with
with
andy
and
and
and
the
guys
from
juniper.
Then
he
added
the
empty
functionality.
I
That
is,
we
have
two
different
parts
and
we
have
the
amtv
and
also
the
anti-gateway,
which
also
tested
for
interrupt
with
the
video
rank.
You
know
the
erc
and
and
also
with
the
public
amt
rays,
that
are
provided
by
provided
by
various
universities
in
the
states
and,
as
far
as
I
know,
they
are
running.
I
25
or
so
nodes
and
and
then
all
are
participating
in
the
beer
domain,
which
is
now
since
last
year,
connected
to
the
b4
test
that
frederick
mentioned
before,
and
the
main
difference
between
the
two
is
that
my
local
whole
network
is
mostly
running
on
tpd
credit
appearance
via
the
p4
test.
But
the
adjunct
is
running
on
hardware-based
switches
and
since
there
are
still
it
is
cross-connected
with
cross-connected
and
we
are
doing
inter
entry
as
beer
set
up
between
them
since
that
time.
I
I
One
of
the
regular
broadcasts
of
my
office
music
is
is
recorded
on
this
video
on
this
demo
link,
which,
which
is
basically
a
unique,
hosted
video
at
the
moment.
But
it
is
showing
you
that
when
I,
when
I
press
the
play
button,
then
then
what
happens
is
in
the
network.
So
I
show
you
that
the
end
nodes
start
the
igmp.
I
This
is
a
triggered
by
the
vrc
startup.
Then
it
it
get
sent
through
the
beer
core
within
a
beam.
Anc
observation.
This
is
the
pier
beam.
This
is
the
pmb
interrupt
and
and
all
these
multicast
pockets
within
the
core
are
transmitted
in
in
a
beer.
At
the
moment,.
I
And
and
since
we
have
the
dpdk
data
plane,
I
think
it
is.
It
is
two
years
old
now
then,
since
then,
all
this
happens
within
the
dpdk
replication
here
and
this
whole
network
is
connected
to
to
public
accessible
amtv.
I
I
Guys
do
you
see
the
terminal?
Yes,
yes,
so
two
things
that
I
would
like
to
show
you
here
is
one
part
that
is
the
emt
gateway
at
the
moment.
I
I
have
started
it
with
the
roka
router
that
that
I
spawn
up
and
when
the
meeting
started,
it's
basically
at
the
moment,
this
very,
very
stupid
configuration
it
points
a
default
multicast
through
to
to
to
a
random
address,
which
is
on
the
tunnel
interface
and
tries
to
join
that
multicast
group,
which
is
sent
out,
obviously
through
the
through
this
next
hope,
which
is
on
the
tunnel
interface.
I
I
That
is
interesting
when
I,
when
I
showed
the
multicast
routine
v
to
vrf
for
the
given
group,
then
the
pockets
and
and
the
bytes
are
increasing
slowly,
and
this
is
this-
is
a
constant
stream
here,
not
too
much
on
this
node
to
show
you,
because
this
is
just
an
empty
gateway,
a
local
gateway
that
can
decapsulate
an
amt
stream
and
basically
it
can.
It
can
handle
the
multicast
pockets
to
the
local
lawn
receivers.
I
I
And
this
is
how
an
amt,
an
amt
relay
that
is
configured
in
free
router,
some
basic
security
things
that
we
added
just
today
to
other,
only
eight
clients
from
a
subnet,
a
total
of
256
listeners
at
all
and
some
access,
some
creative
meetings
on
the
new
terminals
and
and
loads
dropped
to
drop
the
attempts.
I
Finally,
an
interface
to
bind
to
and
and
the
vr
have
to
bind
to
and
an
interface
that
you
can,
that
that
will
be
cloned
for
all
the
clients.
I
This
interface
is
the
points
toward
the
integrands
that
are
the
starting
emt
sessions
at
the
same
tv
dummy
app
address
here
and
access
group
to
only
accept
igmp
pockets,
viva
stream,
multicast
igmp
here
and
the
same
for
v6
the
same
for
v6.
I
And
after
after
all
these
things
at
the
moment,
this
client
is
connected
here.
So
this
is
why
I
have
an
amt
access
interface
cloned
from
the
template
here
so
and
also
the
multicast
group
operate
here
with
the
with
one
interface
is
on
on
the
outgoing
interface
list
and
also
the
pockets
and
the
bytes
are
increasing
on
the
group.
It
have.
It
have
an
ingress
interface
here,
which
is
which
is
a
beer
in
a
bird
interface.
I
Sorry,
but
I
can
show
you
the
and
the
npr's
forwarding
counters,
so
this
is
this
is
how
it
looks
like,
and
you
can
spot
that
it
is
extras
in
bites
and
it
increases.
I
If
we,
if
we
take
a
closer
look
on
this
river,
then
it
is
it's
basically
an
ospf
v4pr
label.
It
did
not
sent
out
at
all.
I
And
it
is
a
beer
local
b
string,
which
is
our
beer
index
expressed
in
in
the
in
the
beer
b
string
here,
and
we
have
some
peels
on
this
on
this
beer
label.
Basically,
they
are
the
same.
They
are
the.
They
are
the
different
nodes
in
the
in
the
igp
domain
and
they
are
expressed
with
these
b
strings.
As
you
can
see,
we
we
have
two
routing
interfaces
here
and
also
the
pockets
and
the
bytes
are
increasing.
J
Knowing
that
we
started
in
2019,
we
are
now
in
the
in
the
in
the
situation
where
we
we
need
to
to
calibrate
all
profiles.
So
if
we
want
to
have
a
pure
p
router,
we
have
to
calibrate
it
in
order
to
fit
most
of
the
to
find
a
baseline.
J
That
would
fit
off
the
most
of
the
use
case
in
the
in
in
our
research
institution,
customer
or
organization,
and
of
course,
br
is
a
multicast
profile
and
beer
profile
is
a
must-have,
because
we
have
some
users
who
have
a
multicast,
enable
application,
and-
and
this
is
why
we
think
that
it's
it's
really
great,
because
this
this
p4
lab
is
giving
us
a
way
to
to
be
ahead
in
terms
of
innovation
and
new
implementation
and
just
to
let
you
know
that
you
are
free
to
test
this
amt
relay.
That
is
public.
J
You
have
club
amateur,
showed
you
the
the
link,
and
we
can,
of
course
share
the
link
with
the
domain
list.
The
good
news
is
that
we
are
we
are
we
are.
We
can
use
this
fifa
app
to
to
validate
two
things.
J
The
first
solution
of
the
program
was
to
to
provide
a
way
to
link
users
to
multicast
feed,
which
is
the
role
of
the
interior,
but
we
think
we
think
that
we
can
solve
the
other
problem,
which
is
tying
new
source,
adding
new
source
to
this
multicast
platform,
so
that
we
can
solve
the
other
part
of
the
program
which
we,
which
is
also
which
will
provide
an
end-to-end
path
to
the
source
toward
the
the
the
receivers
and
that's.
That
would
be
great
because
we
really
think
that
this,
I
wouldn't
say
a
cost.
J
Effective
solution
for
multicast,
but
just
to
know
that
this
runs
in
in
an
overlay
in
an
overlay
on
top
of
a
real
network
is
a
very
great
asset
to
to
provide
maybe
a
new
way
to
provide
multicast
services
that
is
totally
independent
and
safe
in
terms
of
operation
operational
point
of
view-
and
this
is
also
I
think
back
in
the
time
when
I
was
working
in
the
engineering
team.
J
It
was
a
kind
of
scary
to
activate
multicast,
because
at
some
point
you
you,
you
don't
know
if
you
can
geoparadise
some
resources
from
your
backbone
and
not
in
this
case.
This
is
totally
not
the
case.
We
are
free
to
to
exploit
to
explore
new
to
new
new
solution
and
feel
free
to
to
to
reach
us
and
I'm
not
sure
sub.
I
should
have
the
link
to
the
to
the
current
world
map.
J
J
We
are
starting
to
have
new
nerds
in
the
in
the
us
on
the
west
coast
and
north
america
and
in
latin
america,
and
we
will
leverage
a
new
submarines,
cable
in
an
entrepreneurial
capacity,
but
just
to
let
you
know
that
all
of
this
is
a
br
domain
that
that
can
be
used
for
testing
and
also
one
very
good
point
is
that
if
you
are
implementing
a
client
we
have
this
br
domain
is
ipv6
enabled.
So
this
means
that
you
can
implement
your
ipv6
int
client
and
test
it
against
this
before
lab.
C
C
Yeah,
that
is
very
cool,
guys,
nice
work.
So
these
there's
a
couple
of
mailing
lists
you
mentioned,
I
think,
are
these
for.
C
Is
there
a
what
a
charter
or
purpose
for
these
mailing
lists
like
I'm,
not
I'm
not
clear.
I'm
interested
in
following
this
work,
I'm
not
clear,
which
is
exactly
the
right
one
that
I
should
sign
up
for.
I
think
I'm
I
think
I'm
on
at
least
one
already,
but
I'm
not
sure
if,
if
that's
for,
but
I'm
not.
J
C
J
Sorry,
I
wasn't
clear
you
are
already
on
the
on
the
multicast
that
is
between
us,
but
I
I
was
mentioning
the
the
mbondio
mailing
list,
so
people
can
freely
test
it,
get
used
to
it
and
approach
it
discuss
with
us
how
we
can
improve
things
because
for
sure
there
are
things
to
improve
security,
encryption
or
new
number
of
parallel
client,
and
the
good
news
is
that
we,
this
mt,
really
has
a
10
gig
connection
toward
the
follow,
and
we
should.
We
hope
that
soon
we'll
move
to
100
gig.
J
So
we
can
have
a
very
tremendous
toy
to
play
with
and
and
the
good
news
is,
that,
the
more
client
we
test,
the
more
interoperable
it
will
be,
and
even
for
you,
it's
a
it's
a
win-win
situation.
So
if
you
are
a
pure
vlc
developer,
I
would
like
to
create
an
ipv6.
Vlc
client
feel
free
to
reach
us
it's
for
you.
Just
let
us
know.
Sometimes
there
will
be
breakage
and
so
on,
but
it's
it's
here
for
to
for
everyone
to
to
progress.
J
On
the
contrary,
we
can
add
new.
We
can
think
about
how
to
to
bind
new
source
into
this
platform,
even
if
it
is
an
experimental
platform,
but
in
the
end
the
idea
would
be
to
solve
this
problem
where
we
we,
we
bind
the
source
and
the
receivers,
and
after
that,
we
will
give
you
all
the
recipes
to
create
your
own
platform
in
your
own
company,
whatever
with
your
favorite
hardware
and
that's
pretty
much
about
it,.
C
That's
that's
outstanding.
Do
you
have
a
sort
of
timeline
in
mind
for
for
when
you'll
be
ingesting
external
traffic,
and
and
is
that
is
that
gonna
be
using
the
the
sort
of
amt
and
and
dryad
solution,
or
is
that
going
to
be
or
did
you
have
something
else
in
mind.
I
And
for
the
empty,
the
second
empty
in
the
p4
that
was
just
started,
but
as
as
I
showcased
in
the
demo,
we
have
the
empty
gateway
functionality
that
can
pull
traffic
from
from
an
existing
gmt
relay
and
as
I
saw
and
on
the
m2m,
I
cannot
recall.
So
there
is
a
public
list
where
there
are
some
some
interesting
feeds.
I
think
a
a
nasa
video
with
8k
and
and
some
other
universities
from
the
states
which
are
accessible
through
ent.
I
I
I
played
a
bit
back
in
time
with
them
and
noticed
that
internet2
has
shut
down
its
services.
So
basically
we
have
three.
Today's
behind
the
the
domain
that
that
is
the
amtv
and
and
some
sources
are
reachable
through
one
gateway
and
I'm
sorry,
one
really
and
the
other
source
is
behind
otherwise.
I
But
I
so
this
urgent
entv
is
just
started
today,
but
I
I
have
plans
to
to
populate
some
some
static
tumors
to
these,
these
public
other
emt
rays
and
and
they
have
a
constant,
constant
feed
feed
from
them
and
and
make
them
available
in
into
the
giant
multicast
network.
Because
the
more
more
streams
then
we
are,
we
are
more
happy
so
so
in
summary,
yes,
we
have.
We
have
plans,
but
I
think
it
will
happen
in
in
the
upcoming
days.
C
Okay,
great,
I,
I
may
open
a
throat
on
the
mailing
list
to
talk
a
little
bit
about
the
the
dns
based
discovery
mechanism
and
whether
that
would
be
feasible
for
you
as
well.
Thank
you.
J
That
that
that
perfect,
one
thing
that
we
would
also
like
to
to
to
focus
on
is,
of
course,
there
is
a
technology
aspect
where
you,
of
course,
all
of
us
are
technical
people
and
we
can
find
and
understand
things.
But
the
real
idea
that
would
like
to
achieve
in
the
end
is
to
achieve
something
really
brain
dead,
simple.
J
I
would
say
that
receivers
and
source
should
not
know
all
of
these
complexities
or
should
work
instantly,
and-
and
this
is
the
difficulty
technically
complex,
but
make
things
simple
for
users
so
that
they
just
don't
think
about
using
anything
else.
I
will
just
go
this
way
because
it's
simple
and
it
works-
and
this
is
the
the
idea
of
this
whole
entropy-
that
is,
that
we
are
trying
to
to
to
buy.
C
That's
exactly
what
I
want
to
talk
about
yeah,
so
we'll
we'll
take
that
up
offline
but
I'll
just
say:
take
a
look
at
rfc
8777
and
we'll
talk.
We'll
talk
some
more!
Thank
you.
E
Yeah,
I
got
my
hand
up
yeah,
it's
it
comes.
It
does
come
back
to
content,
you're,
absolutely
right
at
the
moment,
there's
very
little
multicast
content
in
the
research
and
education
networks.
There's
a
project
called
eu
metsat,
which
is
meteorological
related
data.
That's
multicast!
We
looked
on
our
uk
network
janet
and
it's
pretty
much
only
you
met
sat
traffic,
that's
multicast!
E
That's
pretty
much!
It
there's
nothing
else
really
and
as.
E
Too,
have
taken
a
didn't
want
to
take
the
decision,
but
at
the
moment
they
don't
have
a
multicast
service
because
of
the
way
they're
transitioning
to
new
a
new
platform.
They'll
reinstate
it
later.
So
it's
it's
currently
multicast
is
a
bit
of
a
second
class
citizen
for
most
things.
So
anything
we
can
do
as
you
say
to
make
it
easier
would
be
great.
So
I
think
what
we're
looking
to
probably
do
with
rare
in
this
multicast
work
is
to
find
ways
of
allowing
people
just
to
stream.
E
Even
from
you
know,
bandwidth
challenged
areas,
so
they
can
stream
out
and
reach
multiple
people
by
automatically
multicasting
from
that
single
stream
that
they're
sending
out
and
making
it
transparent
to
them
as
far
as
possible.
So
yeah,
but
yeah
multicast
is,
is
not
in
a
particularly
happy
state
in
rna
networks,
which
is
unfortunate.
B
Any
other
any
other
questions.
B
Saba,
I
have
a
question
for
you.
What
was
your
experience
in
implementing
amt?
How
much
did
you
know
about
it
before?
Did
you
just
pick
up
the
spec?
Was
it
really
easy?
Was
it
clear.
K
Did
it
take
months
or
weeks
or
hours?
So
basically,
I.
I
I
I
first
heard
I
think
from
you
about
the
ant
this
d-spring,
then
I
I
did
a
lot
of
other
things
and
the
first,
the
first
reading
and
and
the
first
impression
was
not
wasn't
too
positive
here,
because
I,
whereas
I
am,
I
am
thinking
about
end
of
sedations
and
tuning
like
like
pokemon,
so
catching
them
all
is,
is
always
good,
but
but
this
this
kind
to
tune
a
multicast.
I
Multicast
didn't
didn't
get
too
much
laugh
here,
but
then
we
had
some
chats
with
freak
and
and
mentioned
that,
okay,
that
we
should
catch
this
book
cable
also,
so
afterwards
I
started
reading
the
recipe
initially.
I
I
also
didn't
found
the
the
needed
parts
for
me,
so
I
I
started
doing
the
whole
one,
which
is
very
rare
here.
I
I
usually
have
the
idea
and-
and
I
only
see
for
the
pocket
headers
in
the
rfc,
but
then
we
pointed
out
that
okay,
they
are
also
there,
but
I
have
to
scroll
down
a
lot
at
this
point
at
that
point,
so
about
two
or
three
days
passed
with
with
reading
the
refse
and
and
cast
the
the
the
idea
behind
it
afterwards
it
it
was
some
hours
because
adding
a
new
general
language
is
not
too
hard
here
then
I
at
that
point
I
I
was
able
to
to
see
the
multicast
streams
from
from
internet,
to
I
mean
from
the
universities
of
in
states
and
then
holding
the
empty.
I
B
J
Just
to
to
to
come
to
to
conclude,
I
I
just
would
like
to
to
to
hammer
the
fact
that
this
this
lab
is
is
open
to
anyone
who
want
to
improve
their
imt
client
v4
v6.
So
we
are
now
in
the
situation
where
we
we,
we
need
very
good
clients,
so
we
can
demonstrate
things
so,
whether
into
using
a
browser
or
vlc
or
whatever,
whatever.
J
E
B
Great
any
other
questions
for
saba
and
frederick.
B
Cool
well,
thank
you
very
much
guys.
This
is
really
really
interesting
stuff
and
the
combination
of
beer
and
amt
is,
is,
you
know,
potentially
really
powerful,
so
very,
very
neat,
very
useful
stuff,
and
for
the
next
section
I'm
actually
going
to
do
a
demo
that
hopefully
addresses
well,
it's
it's.
It's
an
attempt
in
the
direction
of
making
of
improving
source
content.
B
B
Can
everyone
see
powerpoint,
I
need
to
swap
displays
how
about.
I
B
Right
cool
all
right,
so
this
is
a
project
that
we
had
some
some
interns
this
summer
work
on.
So
what
is
off
net
sourcing
so
today,
there's
native
sourcing-
and
it's
been
around
for
you
know
more
than
two
decades
on
the
mbo
and
the
multicast
enabled
part
of
the
internet,
and
we
can
call
that
you
know
onnet.
B
But
you
know
the
mbone
is
a
very
small
fraction
of
the
internet.
What,
and,
and,
and
you
know
what
amt
does
is
it
allows
off
net
receiving?
So
if
you're,
not
on
the
multicast
enabled
network,
you
can
access
multicast
content
through
an
amt
tunnel
you
can
receive.
But
what,
if
you
wanted
to
do
the
same?
What
if
you
wanted
to
source
and
and
the
fact
is,
most
sources
could
potentially
be
on
unicast
only
networks.
B
So
we
have
a
you
know,
an
interesting
portal
for
discovering
and
launching
off
net
receiving,
and
that
is
the
multicast
menu,
often
at
sourcing.
This
is
where
users
on
unicast
only
numbers
can
stream
some
type
of
stream,
say
video
to
a
translator,
that's
sitting
on
the
m
bone
and
that
converts
the
stream
from
unicast
to
multicast.
B
Then
it
can
be
accessed.
The
idea
is
to
access
it
natively
for
on-net
receivers,
as
well
as
amt
for
off
net
receivers.
This
is
kind
of
how
it
would
work,
how
it
would
look.
We
have
the
the
big-I
internet,
that's
you
know
mostly
unicast.
Only
we
have
the
m-bone,
the
multicast
enabled
part
of
it,
mostly
rna
networks,
and
you
have
a
multicast
source
and
a
multicast
receiver.
That
works
the
way
it's
always
worked.
Now.
What?
If
you
have
an
off
net
receiver,
somebody
on
a
unicast
only
network
who
wants
to
receive?
B
Well,
we
have
these
amt
relays,
send
it
natively
to
the
relay
and
then
the
relay
will
send
it
over
amt
to
receivers
on
unicast
only
networks,
but
what,
if
you
have
an
off
net
source
and
now
we're
introducing
this
unicast
multicast
translator,
which
will
take
a
unicast
stream
in
turn
it
into
multicast
and
spit
it
out
to
native
or
off
net
receivers?
B
So
this
translator
was
developed,
it's
essentially
a
python
script.
That
translates
just
very
simply
translates
a
unicast
stream
to
multicast.
It
was
written
by
one
of
our
interns
named
giannis
it
and
it
all
does
just
rewrites
the
source
and
destination
address.
It
picks
a
random
ssm,
address
out
of
232
and
it
also
posts
and
creates
a
new
entry
on
the
multicast
menu
giannis
developed
this
in
conjunction
with
larry
and
maki
kind
of
working
together
on
a
project,
and
it's
this.
B
This
translator
is
deployed
at
george
washington
university
thanks
to
andrew
gallo.
In
terms
of
you
know,
what
are
the
goals
here?
We're
just
kind
of
working
our
way
through
functionality.
What
would
be
awesome
is
if
we
could
stream
from
an
iphone
or
android
camera,
to
stream
video
to
the
unica
to
the
translator
right
now,
we're
just
using
vlc
from
from
a
desktop,
but
getting
that
working
would
be
really
interesting.
B
The
translator,
which
just
translates
the
source
and
destination-
that's
already
done,
you
know
next,
would
be
you
know
if
it
could
index
the
stream
say
for
search
right
now,
there's
not
a
lot
of
streams,
there's
currently
zero
and
I'm
going
to
create
one.
But
you
know
if
this
takes
off,
there
could
be
zillions
and
you
know
creating
a
directory
service.
B
So
that
you
could
subscribe
to
different
types
of
streams,
say
you
know,
nature
cam,
slash,
zoo,
cam,
slash,
panda,
cam,
and
then
people
could
join
that
particular
channel
and
be
alerted
every
time.
There's
a
new
panda
cam
that
pops
up
so
that's
kind
of
the
the
long
term
vision.
Also,
potentially,
you
know
recording
for
for
look
back
purposes
since
obviously,
multicast
is
a
simultaneous.
You
know,
live
by
nature
technology
and
then
receiver
apps
right
now
we're
using
vlc
but
would
be
nice
would
be
nice
if
we
could
use.
B
You
know
something
a
little
easier
something
an
app
on
for
ios
and
android
as
well
as
something
that
you
know.
If
we
could
get
this
in
the
browser.
This
would
be
really
really
wonderful,
but
for
now
we're
using
vlc-
and
you
know-
we'd
love
some
other
ideas
just
to
make
this
easier.
I
think
you
know
frederick
said
it
well,
if,
if
we
could
could
make
these
things
easier
so
that
any
user
on
the
internet
could
do
it
that'd
be
great
here's
some
references
to
see
the
code.
B
This
is
an
open
source
project
and,
if
anybody's
interested
and
get
involved
just
shoot
me
an
email,
we'd
love
to
you
know
we
have
a
slack
group
of
co-conspirators
or
I
think
jake
has
called
it
a
a
cabal.
If
you
want
to
join
arca
ball,
please
just
shoot
me
an
email,
we'll
add
you
to
the
slack
group
and
with
that,
let
me
let
me
kind
of
show
give
a.
E
Yeah
that
looks
good
a
little
bit
small
if
you
could
do
a
plus
plus
maybe,
but
we
can
see
it
yeah.
E
B
Okay,
all
right,
so
here's
the
one
command
line-
it's
just
runs
vlc
launches
a
a
video,
an
mpeg4
and
it's
sending
it
to
this
magical
ip
address
here,
which
is
the
address
of
the
translator
and
which
is
looking
on
this
certain
port.
So
this
is
just
a
standard,
vlc
vlc
command
for
sourcing.
Let
me
say
this
so
folks
in
the
audience
this
is,
there
is
audience
participation.
Please
check
the
please
check
the
chat
window.
B
I
just
sent
a
link
to
the
multicast
menu
for
anybody
who
does
have
vlc4
be
great
to
give
it
a
shot.
So
this
is
currently
the
multicast
menu
and
you
can
launch
click
on
any
of
these
things
and
it
will
launch
one
of
the
video
streams
on
the
mbone.
So
I'm
I'm
about
to
hit
return
here
on
this,
which
will
create
that
next,
one
that
new
stream,
when,
if
I
reload
the
multicast
menu
you'll,
see
we
just
created
a
new
stream.
B
B
We
can
rename
this
to
cute
puppy,
cam
all
right
and
we
will
upvote
it
a
few
times,
because
this
is
the
internet
and
it's
all
about
shelf
self-promotion,
and
is
anybody
out
there
able
to
join
this
stream.
B
So
here's
the
amt
relay,
you
can
actually
see
the
relay.
We
can
see
some
multicast,
so
the
so
the
group
address
is
232
66
90
70..
That
is
this,
so
you
can
see
the
incoming
interface
and
here's
the
downstream
interface.
This
is.
There
are
two.
This
means
there's
two
tunnels,
so
we
can
do
a
show,
amt
tunnel
detail
and
you
can
see
these
tunnels.
This
one
is
one
amt
gateway.
B
This
happens
to
be
a
second
laptop.
That's
next
to
me,
and
here
is
another
and
here's
a
third.
So
it
looks
like
we
have
three
receivers:
jake,
I'm
guessing.
One
of
you
are
one
of
these
two
addresses
anyway.
Are
you
seeing
any
video.
C
B
J
B
Let
me
see
here
and,
if
you're
using
vlc
for
a
recently
downloaded
version
of
vlc4
it
has
by
default.
This
is
the
relay
that
is,
that
is
the
it's
just.
The
default
relay.
C
But
I
did,
I
did
get
some
video
after
restarting,
but
it's
a
a
bit
choppy
and
it's
it's
hanging
again.
It
looks
like
it's
dropping
some
frames
or
yeah.
C
But
well
for
what
it's
worth,
what
resolution
are
you
sending?
What
bit
rate.
B
I
think
it's
it's
not
that
big,
it's
only
like
eight
megs,
but
I
am
oh
actually
here.
Let's
sit
here
show
multicast
route
detail.
This
is
yeah
about
eight
megs.
B
Gotcha
anyway,
as
as
anybody
interested
in
working
on
this,
as
you
can
see,
this
is
you
know,
this
is
the
idea,
the
multicast
menu
by
the
way
lauren,
who
developed
it,
is
working
on
enhancing
it
to
she
is
close
to
releasing
a
2.0
for
the
multicast
menu
and
all
right
so
frederick
you
you're,
seeing
something
is
it?
Is
it
cute.
J
J
B
B
We
do
have
one
last
presentation,
so
the
song,
if
you
want
to
get
things
teed
up
if
you'd
like
to
share
your
preloaded
slides,
but
with
that
as
he
does,
that
anybody
have
any
questions
about
this
project,
and
you
know
the
idea
and
the
vision
for
this
is
anybody
can
start
streaming
and
sourcing,
something
that
anybody
can
start
receiving
and
it
should
lower
the
bar
for
generating
multi-gas
sources?
B
And
this
is
an
open
source,
simple
python
script.
So
if
anybody
wants
to
launch
a
translator
themselves,
please
let
us
know
we'd
love
to
we'd
love,
to
add
you
to
the
global
translator
pool
which
currently
consists
of
one
but
we'd
like
to
do
more.
It's
proof
of
concept.
But
you
know
you
can
see
the
vision
here,
any
other
questions
about
off
net
sourcing
and
the
multicast
translator,
jake.
C
C
B
Interesting,
so
the
injection,
so
the
translator
just
gets
the
source
onto
a
multicast
network
with
dryad.
You
could
use
that
to
hide
from
I'll
have
to
think
about
that.
I'm.
C
Saying
I'm
saying
I
would
perhaps
run
your
open
source
translator
externally,
just
in
in
my
usual
aws
multicast!
Oh,
yes,
yes,
yes,.
C
And
that
could
be
pulled
in
wherever
it
could
go
directly
from
from
those
relays
to
you
know
your
network
or
other
networks
as
well,
but
also
could
could
go
natively
to
to
various
i2
receivers
so
that
that
does
look
interesting
and
a
useful
source
of
traffic.
The
streams
I'm
running
are
I
mean.
We've.
We've
run
some
some
customer
streams,
but
they
usually
want
those
a
little
more
controlled.
C
D
B
Yeah,
I
definitely
think
that
could
work
and
that
could
be
a
good
test
of
of
dryad
and
for
what
it's
worth
the
next
multicast
menu
will
have
the
ability
to
specify
a
different
relay
so
right
now
it
just
uses
the
default
relay
in
vlc
but
which,
like,
for
example,
if
you
could
plug
in
a
different
relay,
then
it
could
launch
that
from
a
different
relay.
So
that
would
work
in
conjunction
with
what
you
know
saba
and
you
know
his
his
his
test
stream.
B
So
so
yeah
that
could
all
these
pieces
can
all
work
nicely
together.
But
with
that
it's
I
have
to
cut
myself
off
and
hand
things
over
to
you
song,
frederick,
yes,
very
quickly,.
J
Just
as
a
side
note
and
very
quickly,
we
will
try
to
to
play
with
this
at
least
and
see
how
we
can
use
it
in
in
the
p4
lab.
B
M
We
hear
you
okay,
thank
you
today.
I
will
present
the
multicultural
redundant
ingress,
router
failover,
and
this
is
the
second
presentation
in
itf.
So,
firstly,
I
will
give
the
quick
very
quickly
review
of
this
chart.
M
For
this
brief
introduction,
another
two
ir
and
the
in-grass
routers
are
used
to
avoid
a
single
node
failure
for
the
multicast.
The
two
irs
are
the
umh,
the
upstream
multicast
hub
candidates
for
the
ers,
the
er
means
the
egress
router,
so
for
the
pim
or
beer
or
rcpt,
p2mp
or
mmdp.
M
Whatever
the
specific
multicast
technologies,
different
functions
can
be
used
for
for
the
deployment,
so
there
are
three
stand
by
modes
can
be
deployed
for
the
ir
switchover
in
case
of
the
irs
failure
that
the
code
standby
and
the
wall
standby
and
the
heart
standby
and
the
details
you
can
find
in
the
draft
and
the
cost
and
the
influence
of
the
three
modes
are
different
and
now
no
one
of
these
three
modes
are
the
best
choice,
as
depends
on
the
actual
network
deployment.
M
M
So
the
update
is
that
in
some
deployments
of
the
warm
standby
mode
and
the
irs,
the
irs
made
in
charge
of
different
multicast
flows.
For
example,
if
we
have
a
10
multicast
video
flow,
so
we
can
divide
for
the
five
five
for
each
of
the
irs
order
is
made
charge
of
the
different
ers
for
one
specific
flow,
for
example,
robust
for
one
flow
ir1
in
charge
of
er2
and
er1
under
the
ir2
charge
of
er3.
M
The
ios
can
can
be
backup
devices
for
the
for
each
other
and
in
case
one
ir
fails
and
the
other
ir
may
notify
do
not
the
arc
may
notify
the
other
ir
to
forward
the
flow
to
the
ers,
for
example,
for
one
specific
flow,
i1
is
in
charge
of
er1
er2
and
the
r2
is
in
charge
of
er3
and
the
if,
while
l1
fails,
r1
fails
and
it
will
notify
the
r2
and
r2
will
forward
the
multicast
flow
to
the
er
two.
M
Yes
er
one
year,
two
and
yeah
three.
So
that's
the
update
of
this
version.
M
That's
of
the
draft
and
any
comments
are
kind
of
webcam
for
this
draft
and
we
quarters
think
it's
ready
to
requesting
the
working
group
adoption.
That's
all
thank
you.
C
Jake
you're
up
yeah
thanks.
So
in
the
draft
I
I
didn't
see
how
it
detects
failure
and
and
how
it
how
it's
doing
the
signaling
from
from
one
to
the
other
that
it
needs
to
that.
It
needs
to
be
in
charge
of
the
of
the
ingest.
M
So
for
this
chapter
we
we
we
just
give
the
a
general
way
for
the
multicast
root
root.
Router
failure
not
not
involved
the
specific
technology
in
that.
C
Okay,
so
I
guess
I
I.
I
think
that
this
is
a
really
valuable
space
to
develop,
but
I'd
like
to
I'd
like
to
see.
C
I
I
guess,
what
some
more
specific
I'd
like
to
see
some
some
recommended
mechanisms,
at
least
like
if
some
exist,
then
I'd
like
to
see
references
to
them,
or
you
know
some
some
guidance
on
how
to
do
the
detection
that
that
a
failover
is
right,
because
it
seems
like
that's
a
I.
I
thought
that
was
necessary
in
the
for
this
draft
to
be
to
be
headed
the
right
direction
to
me,
but
yeah,
but
thanks.
I
think
this
is.
D
C
A
H
B
So
for
those
who
noticed
there
was
a
we
had
two
polls
who
was
read
the
draft
and
who
would
like
to
see
it
adopted
as
a
working
group
draft.
We
had
seven
people
who
have
read
it,
so
it's
a
good
sign.
B
So
a
third
of
the
number
of
people
on
and
looks
like
if
you
could
just
put
your
hand
up
show
of
hands
who's
who
thinks
that
this
this
is
be
a
a
draft
that
would
who
would
like
to
adopt
this
from
the
work
as
a
working
group
document.
B
So
far,
we've
got
six.
Anybody
else,
no
objections.
So
I
think
we'll
take
this
to
the
list
for
an
official
working
group.
Adoption
call.
This
was
by
the
way
presented
back
in
09
and
figured
we'd
represent
it
again
to
get
it
fresh
on
people's
minds
first,
before
the
adoption
call
so
looks
like
there's
support
to
to
adopt
so
we'll
take
this
officially
to
the
list
for
an
official
adoption
call.
B
Please
speak
up
on
the
list,
both
if
you
want
to
see
it
adopted
or
not
adopted,
and
also
you
know,
any
any
suggestions
that
you
see
for
the
document
are,
of
course,
always
welcome
at
all
times,
and
it
looks
like
I'm
about
to
end
that
poll
there
we
have
seven
who
say
yay
and
one
person
who
did
not
raise
a
hand.
B
And
any
other
questions
or
comments
for
for
this
draft.
B
Okay
with
that,
thanks
for
thanks
for
coming
and
we'll
see
folks,
virtually
or
physically
in
ietf,
13
113
thanks,
everybody
thank.