►
From YouTube: PIM WG Interim Meeting, 2020-04-21
Description
PIM WG Interim Meeting, 2020-04-21
A
A
A
What
sir
exactly
the
offers
want
to
add
anything
to
it
or
if
they
believe
that
they
are
ready.
So
the
document
is
ready.
A
Not
sure
whether
it's
ready
for
our
last
call
yet,
but
humans
at
least
find
out
what
the
next
step
is
for
the
trust,
but
also
here
I
mean
it's
great.
If
people
can
look
at
the
just
and
give
comment
from
the
list
so
that
they
can
keep
improving
it,
and
then
we
have
the
IBM
of
the
expansion
draft,
which
is
a
new
draft,
and
it's
just
adopted
I'll
be
presenting
this
later
in
this
meeting.
Those
were
a
start
packing
draft
that
just
got
adopted
that
will
be
updated.
Next
meeting
I
can
just
say
quickly.
A
Then,
funnily,
here
they're
trying
to
move
out
improve
recently
and
m/l
delivers
into
two
internet
standard,
and
we
had
a
survey
trying
to
reach
out
to
anyone
that
has
implemented
these
protocols
or
deploy
them
and
try
to
get
some
inputs,
but
the
only
got
12
responses
so
far,
so
I
really
would
like
all
of
you
that
have
implemented
or
deployed
this
to.
Please
have
a
look
at
mr.
B
right
now,
if
we
open
until
the
end
of
this
month,
so
yeah
great.
A
If
you
can
fill
in
this
survey,
it
only
takes
a
few
minutes
and
it's
possible
also
share
it
with
others
that
you
know
have
deployed
multicast,
implemented
multicast.
We
would
like
vendors,
high-speeds
enterprises
and
anyone
using
multicast
to
respond
to
this
may
be
in.
Should
we
repeat
this
in
Hamburg
the
meeting
as
well.
B
A
Right
so
so,
basically,
we
will
be
like
N
and
an
internet
draft
that
has
like
the
proposed
Pearson
initial
proposal
for
the
new
version,
and
it
basically
addresses
a
router
that
we
have,
plus
it
might
remove
certain
things
in
a
document
that
we
decide
that
hasn't
been
implemented
widely
enough
or
or
also,
if
there's
interoperability,
issues
to
certain
things.
We
can
try
to
clarify
things
in
the
document.
So
the
important
thing
is
there
shouldn't
be
any
new
requirements,
any
one
that
is
compliant
with
existing
arts.
A
C
Because
it's
a
very
important
step
for
the
protocols
and
if
we
notified
other,
is
deals
especially
because
some
operators
might
not
participate
in
ITF
anthem
in
particular,
but
they
might
participate
in
others.
Deals,
for
example,
broadband
forum
and
multicast
as
part
of
IPTV
is
one
of
the
topics.
As
far
as
they
know,
in
discussing
broadband
forum,.
A
A
A
All
right,
I
think
that's
it
yeah.
So,
by
the
way,
as
long
as
we
have
pretty
much
one
person
at
a
time
that
wants
to
make
a
comment,
I
think
it's
fun
to
just
do
that,
like
we
just
have
done
so
far.
If
you're
multiple
people
like
to
feel
like
you
need
to
respond
early,
we
should
try
to
do
the
queue
management,
gonna
prosecute,
but
but
yeah
so
far
as
trying
as
long
as
they
don't.
Thank
you
so
comments
the
same
time.
E
E
E
E
So
yeah,
unfortunately,
he
passed
away
on
Sunday,
January
19th
and
the
funeral
was
held
the
weekend
following
that.
I
was
fortunate
enough
to
be
able
to
make
it
to
a
beautiful
funeral
and
celebration
of
his
life.
This
presentation
here
is
Steve
similar
one
of
his
close
colleagues,
giving
a
presentation,
ironically
or
poetically,
at
the
Barcelona
Cisco
live,
which
was
the
exact
week
after
that.
So
I'd
want
to
play
a
few
excerpts.
I'm
gonna
try
to
keep
it
within
10
minutes,
but
I
think
this
some
kind
of
captured
Bo
and
who
he
was.
E
He
did
participate
in
the
ietf
several
times
and
he
he
was
the
guy
with
a
huge
personality.
It
always
came
up
to
the
mic
and
always
made
us
understand
as
designers
I'm
how
to
make
people
understand
this
complex
technology
and
how
to
deploy
it,
and
that's
that
was
a
void
that
he
really
filled
Cooke
fielded
pretty
well
throughout
his
career,
but
anyways.
Let's
give
this
a
shot.
G
H
G
multicast
at
these
two
pictures
of
him
while
ago,
and
the
last
one
I
have
of
him
is
actually
from
Cisco
live
San
Diego
about
three.
Four
years
ago,
alongside
a
lady
called
Monique
Fidel,
who
used
to
do
Bonnie's
job
back
in
the
day,
I
thought
it
was
highly
relevant
that
she
would
be
up
here
as
well.
I
know
that
Bo
would
appreciate
that
fact.
G
G
G
G
G
Yet
did
you
eat
yet
that
question
is
squeak:
let's
go
eat,
okay,
so
if
you
actually
want
to
confuse
a
European
later
this
week,
I
highly
recommend
that
particular
line
in
humor
times
bow
actually
he's
the
only
person
that
I
first
met
that
made
humor
out
of
multicasts.
He
had
a
couple
of
really
really
good
jokes
about
multicast
and
I
just
loved
these
over
time.
Those
of
you
who
know
about
rendezvous
points,
these
aren't
multicast
dating
agencies
where
senders
and
receivers
come
together
and
have
shortest
path.
Trees.
G
It's
not
that
funny,
but
it
made
me
laugh
at
the
time
and
it
still
sticks
with
me
down
the
the
favorite
I
think
was
the
Eskimo
GRP
bit
prune.
Shannon
said
to
me:
you
better
talk
about
this,
or
was
it
Tim
I,
don't
remember
the
Eskimo
GRP
bit
prune
is
the
way
you
can
guarantee
yourself
a
quiet
evening
and
a
cocktail
party.
G
If
you
want
to
talk
to
the
person
next
to
you
and
you
think,
nah
just
engage
them
with
the
Eskimo
GRP
vet
prune
for
a
second
or
two
if
they
don't
walk
away
from
you
after
that,
try
the
fact
that
PIM
is
a
finite
state
automata
that
was
his
other
great
line
and
I
want
to
offer
this
up
to
all
of
you,
because
this
was
a
boat
Williamson
invention
and
he
didn't
put
a
patent
on
it.
So
you
can
all
take
it
away.
The
geek
ometer
who's
heard
about
the
geek
comments
are
before
I
know.
G
Some
of
you
have.
So
let
me
explain
the
geek
comments.
If
you
see
a
slide
with
the
geek
on
it's
her
and
it's
moving
up
towards
the
red,
you
know
that
there
is
a
significantly
geeky
amount
of
content
about
come
your
way.
G
What
bo
said
here
discovered
is
that
when
you
do
this
kind
of
thing
to
people
in
a
room
like
this
people
start
excreting,
geek,
endorphins
and
the
way
the
way
you
can
tell
the
way
you
can
tell
that
has
happened.
He's
like
this
guy's
turn
around
and
look
at
the
person.
Next
to
you,
okay,
those
are
the
evidences
of
what
that's
the
evidence
of
what
geek
in
dolphins
can
do
to
you
or
an
overdose
of
them.
I.
G
Wanted
to
share
just
a
few
pictures
of
Beau
and
I've
got
some
from
his
professional
life
and
some
of
him
as
a
family
man.
There's
a
guy
who's
been
really
helpful
in
pulling
this
together
and
I'm.
Not
gonna
skip
this
without
mentioning
his
name,
Dino
Ferren
Archie.
Many
of
you
all
know
of
him.
Maybe
if
you
all
know
of
him
as
the
inventor
of
Lists
before
he
invented
less,
he
invented
pen
and
Dino,
helped
me
pull
these
slides
together
and
give
me
content.
G
These
are
in
Dino's
house
about
15
20
years
ago,
and
I
think
the
top
left
was
the
first
and
then
Dino
started
coding
in
the
top
right,
and
he
looks
to
me,
like
Dino's,
just
thrown
out
one
of
his
many
infamous
bugs
there,
which
bo
has
just
spotted
in
the
middle
you
see
by
with
Chris
Long,
Vic
and
Leslie,
who
used
to
look
after
the
consulting
team
in
the
US
on
the
bottom.
There
there's
a
there's,
a
family
picture
or
there's
a
family
member.
G
The
very
young
guy
in
the
blue
shirt
is
bosun
Chris
who
works
for
Cisco
these
days
and
that
was
taken
at
the
Dallas
IETF
bo
lived
in
or
just
outside,
Dallas
in
Texas,
and
then
on.
The
left-hand
side
is
a
picture.
I
actually
took-
and
this
was
an
interesting
evening,
because
I
did
quite
a
lot
of
work
about
ten
years
ago
on
multicast
VPN,
that
side
I'm.
Looking
this
side,
it's
that
side.
G
G
Some
quotes
from
some
people-
I'm,
not
gonna,
try
and
read
through
all
of
these.
Some
there's
at
least
one
or
two
of
you
in
the
room
here,
Chaim
Butler,
who
is
actually
the
person
who
introduced
me
to
Bo,
gave
me
the
job
of
learning
his
course
so
that
I
could
teach
it
so
Thank
You
Jane
for
that
Max
is
here
and
Max
says
he
hopes
he
can
pick
up
some
of
those
skills
in
presenting.
Do
you
have
all
of
them,
but
take
more
this
there's.
G
Some
really
lovely
quotes
up
here
from
many
of
those
contemporaries.
The
one
I
probably
like
the
most
is
from
Andrew,
but
har,
see
who
says
he
will
be
sorely
missed
in
the
global
networking
community
to
use
a
geeky
parlance.
The
LED
on
the
switchboard
got
a
little
dimmer
for
all
of
us
yesterday
and
it's
true
and
it's
lovely
and
his
poetic,
and
over
here
dinos
with
Bart
board.
It
was
a
former
customer
of
mine
at
Goldman
Sachs.
He
was
in
New
York.
Last
week
there
were
people
talking
about
beau
in
New,
York.
G
Some
family
pictures
for
you
beau,
was
a
pilot.
He
was.
He
was
more
than
the
pilot
beau
built
his
own
f-16
simulator
in
his
garage
taught
GPS
training.
He
taught
Morse
code.
He
was
a
ballroom
dancer,
he
was
a
woodworker
and
he
did
multicast
better
than
just
about
anybody.
I
ever
met
this
guy
was
multi-talented
and
we've
really
lost
somebody's
battle
and
the
only
picture
I
was
never
lucky
enough
to
meet
his
wife
and
his
son,
but
the
only
picture
of
his
wife.
G
G
So
finally,
I
just
want
to
close
with
some
words
to
you
all
about
how
you
can
take
Beau's
spirit
and
memory
into
this
event.
This
week.
First
of
all,
if
you
didn't
know
already,
he
is
still
captured
for
eternity
on
the
fantastic
Cisco
life
365
archives.
There
are
recordings
of
his
sessions
from
the
last
3-4
years.
He
spoke
as
recently
as
2017
and
sitting
alive
in
Las
Vegas.
He
was
an
accomplished
speaker,
a
distinguished
speaker
is,
is
Denise
Fishburne
in
the
room,
I
think
so
fish.
G
If
those
of
you
who
know
Denise
Fishburne
is
going
to
dedicate
her
upcoming
session
this
week
and
her
future
multi
cast
sessions
and
Cisco
live
Capote
Williamson,
she
has
the
remaining
multicast
session
in
the
Cisco
live
agenda,
and
this
over
here
is
Bo's
book,
which
is
still
to
my
knowledge.
One
of
the
finest
books
written
about
multicast
I've
ever
read.
G
Would
Bo
ask
you
all
to
do
as
a
result
of
the
memory
that
I'm,
hopefully
passing
on
to
you
today,
I
think
the
three
things
boughten
anything
else
that
were
true
to
by
his
heart
and
that
made
him
such
a
great
speaker
was
make
people
laugh,
entertain
them,
make
them
think,
provoke
them
a
little
bit
above
all,
make
them
align.
You
all
know,
that's
what
you're
here
to
do
and
I'm
sure
you're
all
gonna
do
a
great
job.
G
B
B
B
Get
myself
the
ball
back.
Can
you
just
up
chef,
just
stop
sharing
I'm
trying
to
grab
the
ball?
What
it's
not
working,
I
think.
F
By
the
way,
a
real
you
are
not
recording
RV,
mr
lee
kuan.
I
did
a
record
kind
of
like
five
minutes
into
it,
so
we're
good
okay.
E
B
J
A
A
A
I'm
doing
this,
mainly
because
of
some
beer
extension
that
we
would
like
to
have,
but
it
cos
means
for
any
other
data.
Nothing
would
like
to
add
so
this
draft
was
just
adopted
and
yeah
people
were
generally
in
favor.
It
actually
had
some
suggestions
that
I
will
discuss
on
my
later
slide
here,
but
yeah
I
should
add
so
the
first
person
on
the
draft
did
actually
include
this
ear.
Extension
prepared
since
then
knew
this
to
be
a
specific
craft.
So
so
the
current
traffic
is
just
talking
about
the
generic
extension.
K
D
B
L
B
L
So
in
case
is
something
that
has
already
been
discussed,
because
it's
kind
of
a
thing
that
we,
while
we
were
working
on
multicast,
came
as
a
problem
to
us
and
we
wanted
to
discuss
whether
this
is
a
problem
that
people
have
seen
and
do
we
really
want
to
fix
it
or
we
keep
it
implementation
specific
next
slide.
Please
I
just
have
one
or
two
slides,
so
I
will
not
take
a
lot
of
time.
L
So
while
we
were
trying
to
look
into
the
MSM,
RIT
I
couldn't
see
much
on
MTU
specific
specifications,
so
we
thought
and
we
actually
have
something
path.
Mtu
specific
in
the
P
MSM
RFC
I
looked
back
into
the
ipv6
path,
MTU
RFC,
which
clearly
says
that
there
has
to
be
a
path.
Mtu
has
to
work
for
both
ipv4
sorry
for
unicast,
as
well
as
multicast
traffic
and
for
a
multicast.
It
should
based
upon
the
list
of
outgoing
interfaces.
L
It
seems
to
me
that
something
is
little
missing.
Is
it
intentionally
kept
implementation
specific
to
really
handle
that
or
we
can
really
improve
the
protocol
in
a
way
to
handle
situations
where
path
MTU,
where
the
MTU
exceeds
of
the
egress
interface
and
we
can
take?
We
can
take
some
action
on
top
of
that,
because
the
reason
in
itself
is
a
like.
A
data-driven
protocol,
like
you
program
the
multicast
routes
based
on
data
itself.
L
So
is
there
a
way
we
can
really
enhance,
based
on
the
size
of
the
data
that
we
need
to
forward
on
the
egress
interface?
Can
we
change
something
in
the
protocol
itself
so
that
the
MTU
part
is
considered
because
a
lot
of
issues
we
see
where
the
empties
are
not
properly
set,
especially
for
multi,
has
been
used
for
IPTV
servidio
screens
and
where
the
MTU
size
could
actually
go
high?
So.
L
What
we
are
superb,
what
we
are
trying
to
propose
is:
is
there
a
way
that
inside
the
PIM
protocol
we
can
say
that
before
computing,
the
outgoing
interface
list
and
programming
that
the
EM
route
entry
we
compute
the
MTU
and
if
the
MTU
is
less,
we
send
it
whatever?
That
leaves
zero
RFC
a201
field
similar
like
that.
L
There
are
empty
path,
m2
stops,
but
if
there
are,
if
there
are
actually
SG
joins
and
I
am
transitioning
from
say,
a
software
out
into
a
Hardware
routing,
State
and
I
compute,
the
outgoing
interfaces
and
I
figure
out
that
I
actually
cannot
really
program
based
on
the
packet
size
out,
sender,
error
and
stay
in
that
particular
state.
And
then
the
TR
keeps
on
that
doing
the
same
thing
that
it
does
for
unicast
path,
MTU
and
the
the
same
implementation.
That
typically
is
used
for
a
unicast
path.
L
Mtu
is
extended
for
multicast
and
we
find
the
exact
same
to
you
from
the
path
from
the
source
will
fill
the
host
and
we
get
the
optimal
MTU
so
that
we'd
never
have
packet
drops
or
fragmentations
on
the
because
segmentation
itself,
reassembly,
is
a
pain
and
cannot
then
typically
is
not
doesn't
work
at
line
rates
and
for
ipv6
intermediate
router
fragmentations
are
also
not
allowed.
So
it's
it's
good
to
have
that
path.
Empty
of
support
inside
the
protocol
itself.
L
L
L
It's
not
the
insight
that
what
I'm
trying
to
say
is
widely
you
are
in
that
connection,
establishment
phase
and
you
are
not
programming
the
entry
into
M
drought
table
I
just
mentioned
it
as
a
hardware,
orden,
softer
rock.
It's
not
something.
The
protocol
itself
talks,
so
I
will
probably
correct
it.
It's
something
there
in
the
slide
itself.
L
A
This
is
just
a
couple
of
quick
comments.
One
is
yeah
for
ipv6
a
corrected
RFC.
You
mentioned
you're,
actually
doing
path,
MTU
discovery,
also
from
home,
because
so
this
works.
This
actually
works
today,
together
with
Tim
I'm,
not
sure,
if
all
vendors
support
person
to
discover
some
of
the
custom,
but
that
this
is
it's
a
possible
always
done
for
ipv6
and
the
way
that
yeah
Thomas
yeah
putting
out
a
bit
the
waste
is
basically
all
just
relying
on
sending
and
ICMP.
A
You
know
package
a
big
message:
if
the
optimum
interface
is
chipping,
you
know
basically
gets
like
a
three
MTU
kind
of
a
so
very
based
on
you
know.
Maybe
some
your
receiver
joins,
then
she
might
go
down
and
so
on
so
one
one
issue:
I
can
see
it's
in
some
cases,
one
one
receiver
with
a
very
small
empty.
A
You
might
cause
problems
for
all
the
other
receivers,
so
there
might
be
cases
where
you
maybe
don't
want
to
honor
empty
another
thing,
I
just
sort
of
mentioned
quickly
is
you
might
want
to
check
out
your
hard
seat
six
days
out,
seven,
which
has
pop
count
extension
tipping.
So
that
has
this
mechanism
where
a
pin
joint
can
include
an
empty
you
and
by
doing
that,
it's
possible
for
the
source
of
the
treat.
You
know
the
DMT
you
even
before
there
is
data
being
sent.
A
L
N
No
very
quickly,
I
think
we
we've
we've
been
regurgitating
this
for
twenty
years.
My
conclusion
has
always
been:
unless
there
is
very
strong
evidence,
we
need
to
make
sure
that
all
the
multicast
applications
actually
run
with
a
minimum
and
to
you
1280
in
ipv6,
and
that's
basically
what
I
got
into
RC
35:42
the
extended
socket,
API
and
I.
Think
that's
what
pretty
much
you
know.
All
the
applications
are
doing
today,
I'm,
not
aware
of
any
use
of
larger
MTU
or
path,
MTU
discovery,
so
before
I
think
doing
more
work.
It
would
be
good
to.
L
N
B
N
N
So,
as
I
said,
you
know
every
2
years,
I'm
asking
the
working
group
and
we're
saying
yeah.
Of
course
we
do
want
in
ipv6
MLD
to
be
used
also
for
link
local
addresses,
because
we
kind
of
missed
out
of
that.
In
ipv4-
and
it
has
proven
to
be
really
bad
with
the
amount
of
ipv6
link-local
multicast,
especially
when
you
got
bridges
over
to
wireless
LAN
segments,
you
really
would
like
to
be
able
to
filter
all
the
stuff.
N
N
So
when
we
actually
wrote
that
into
ml
dv1,
there
is
this
sentence
here
that
says
that
it
it
they
are
used
for
scope
to
then,
unfortunately,
when
we
went
to
our
FD
3810,
there
is
really
no
text
at
all
where
to
use
MLD,
on
which
scott
addresses,
at
least
I
couldn't
find
any
and
the
way
idea
rise
about
how
that
happened
is
that
ml
d
v2
was
written
not
by
you
know,
amending
and
improving
27:10,
but
by
actually
copying
IGMP,
v3
3376,
and
then
you
know
changing
all
the
address.
Links
to
a
hundred
to
it.
N
N
Multicast,
of
course,
can,
in
my
opinion,
should
explicitly
state
that
they
mandate
the
use
of
an
OD,
like
you
know
in
my
AC
key
draft,
because
there's
nothing
in
38
10
that
states
that
this
cannot
be
done
or
should
not
be
done,
because
there
just
is
no
text
about
that
right.
So,
and
it's
really
from
my
interpretation
right
now
for
the
up
to
the
application
when
to
use
n
or
db2,
which
of
course
is
bad
right.
N
N
This
I've
open
an
errata,
and
it's
where
I'm
kind
of
you
know
at
the
end
of
my
process,
unreasoning
I'm,
not
sure
what
you
know
if
the
errata
is
good
enough,
if
we
as
the
pin
working
group,
need
to
kind
of
approve
it
or
so
so,
I
hope
that
Alberto
can
chime
in
or
somebody
knowledgeable
with
the
process
that
you
know
I
should
know
about
and
if
not
a
rather
than
what
and
that's
it.
Thank
you.
B
O
A
O
No
so
apparently
no
one
actually
knows
what
updates
officially
means.
Sorry,
there
is
a
draft
in
the
gem
dispatch
working
group,
so
in
the
energy
we
have
been
discussing
this
for
a
long
time
because
it
is
not
exactly
clear,
has
never
been
defined.
The
only
place
that
it
was
defined
was
in
the
RC
editors
style
guide,
which
just
says
you
should
put
up
the
tag
somewhere,
but
there's
no
official
meeting.
So
some
people
take
it
to
me
what
you
just
said.
O
Other
people
take
it
to
mean
that
maybe
a
multi
to
was
just
an
enhancement
to
a
monthly
view.
One
others
take
it
to
mean
that
it
was
maybe
a
replacement
and
others
just
say.
Oh
look
at
that
other
thing
as
well.
So
there's
this
draft
in
gin,
dispatch
that
tries
to
say
well,
we
should
get
rid
of
those
tags
and
put
something
that
is
more
specific.
That
is
clear
tag
and
I
forget
the
names
for
the
text.
Exactly.
O
O
N
N
O
So,
yes,
if
you
want
to
change
a
behavior
of
something,
yes,
you
need
to
go.
Do
an
update,
whether
in
this
case
it's
a
one-sentence
update
or
review
the
whole
thing
yeah
that's
up
to
the
through
the
working
group
to
decide
what
the
best
way
to
do.
That
is,
if
nothing
else
is
going
to
change
in
this
case,
I
think
what
you're
saying
is
that
for
link-local
ipv6
you
must
use
a
multi
right
well,.
O
N
O
N
N
B
C
B
A
J
We
are
all
under
this
lockdown
here
in
Canada
and
I
was
a
little
bit
forward
beside
putting
a
picture
up
with
the
snow.
On
my
lawn,
my
kid
pulling,
my
hair
out,
I
decided
to
go
with
the
later.
That's
my
personal
update,
but
from
the
draft
point
of
view,
so
there
is
a
replication
segment
draft
that
is
in
the
working
group
and
as
of
now,
as
you
guys
know,
there
are
some
challenges
in
the
spring
working
group.
There's
a
single
chair.
J
J
One
thing
that
I
wanted
to
bring
into
the
table
here
is
the
replication
segment.
Is
a
small
portion
of
the
point-to-multipoint
policy
draft,
meaning
that
the
point
multiple
in
policy
draft
is
really
the
architectural
draft
and
is
the
prof
that
explains
this
idea
inside
out.
Meanwhile,
the
replication
segment,
even
that
it
has
the
forwarding
information
of
the
multicast
state.
J
This
is
why
we
decided
to
put
that
draft
and
that
a
small
part
of
it
into
the
screen.
So
one
question
here:
I
guess
the
working
group
is
even
the
current
situation
with
the
spring.
Will
it
make
sense
for
PIM
working
group
to
pick
up
the
torch
and
and
to
lead
this
effort
forward?
So
then,
all
these
other
drafts,
including
the
PCE
and
IDR,
we
just
start.
You
know
we
could
start
having
conversations
on
them.
J
B
I'm
gonna
put
your
face
back
up
there,
because
this
is
my
favorite
slide
so
far,
but
yeah
so
that
alvaro,
I'm
gonna.
Ask
you
to
chime
in
here
again,
because
I
know
that
we've
shared
a
little
bit
different
opinions
on
this
I'm
kind
of
more
leaning
towards
this
yeah.
Let's
accept
it
and
see
what
happens,
but
it
does
have
reliance
upon
the
draft.
That's
in
spring,
this
replication
segment
draft.
So
what
we
agreed
between
the
chairs
of
spring
and
is
that
the
spring
chairs
just
said
yeah
go
ahead.
L
B
Wait
on
that
draft
to
be
adopted
in
spring
before
we
dropped
it.
Your
draft
in
PIN,
although
I,
know
your
names
on
both
of
those
drafts,
but
so
it
may
be
in
this
case
better
to
have
the
the
working
group
get
some
feedback,
because
right
now
it's
this
kind
of
been
between
chairs
and
Alvaro
Alvaro.
Do
you
have
any
comments
further
on
whether
or
not
PIM
should
be
adopting.
O
I
have
exactly
the
same
comments
that
I
told
you
before
when
we
talked
with
the
spring
chairs,
which
were
and
I
was
just
frantically
looking
for
my
email,
because
I
obviously
don't
remember
everything
say,
but
when
I
said
I
think
was
that
I
will
prefer
us
to
delay
the
adoption
until
spring
the
opts.
The
reason
for
that
is
that,
as
far
as
I
understand,
there
is
a
dependency
with
the
draft
here
in
PIM
on
the
draft
over
there
and
there's
nothing
that
precludes
the
working
group
from
working
on
the
PIM
draft,
even
if
not
adopted.
N
Miss
typing
trying
to
get
hue
but
I'm
missing
it
and
now
I
think
I
would
support
whatever
I
was
saying
and
maybe
return
in
the
same
way.
That
Alber
is
saying
that
we
need
to
inform
six
men
on
stuff.
So
basically,
if
this
goes
to
two
to
spring,
then
they
should.
Basically,
you
know,
update
us.
You
know.
Whenever
you
know
there
is
some
affection
to
what
in
working
group
should
know.
F
O
That
we
did
talk
to
do
right
so
yeah,
so
just
that
everyone
knows
since
I
think
the
last
site
nothi
whatever
ITF
106,
maybe
even
before
yeah
there's
nothing
had
come
up
and
and
so
Mike
and
state
had
already
been
in
touch
with
Bruno
and
Martin.
The
ad
for
spring
was
well
on.
You
know:
what
do
we
do?
You
read
thoughts
on
this
or
not,
or
you
know
where
it
wouldn't
so
yeah.
So
that's.
O
We
had
talked
about
now,
I
just
want
to
say
one
more
thing,
the
same
way
that
nothing
precludes
the
working
group
from
working
on
the
document
that
hasn't
been
adopted.
In
fact,
it
would
be
great
if
people
actually
reviewed
documents
at
all
stages
of
the
lifetime
that
the
life
of
the
document
option
you're
in
adoption
after
adoption
before
last
call.
You
know
all
that
stuff,
there's
nothing
that
it
forces
the
publish
anything
that
has
been
adopted.
H
O
B
I'm
gonna
meet
you
again
tour
list
because
you're
being
and
pretty
good
there,
so
yeah,
okay,
thank
you.
What
we
need
to
move
on,
Thank,
You,
human,
but
I,
think
we
just
need
to
as
Alvaro
suggested,
just
continue
to
work
on
your
draft.
We've
already
had
a
adoption
call
and
we,
as
a
working
group,
have
agreed
to
adopt
it
I
think
for
now
we
should
maybe
we
should
take
it
to
list
and
get
people's
opinions.
That
may
be
the
best
thing
to
do,
and
we'll
do
that
actually.
B
B
So
what
I'm
trying
to
think
of
what
we
should
do
here,
so
we've
got
one:
it's
a
it's
a
good
Jason!
It's
an
operator
presentation,
it
kind
of
fits
in
both
of
our
working
groups.
Would
you
prefer
that
we
have
it
be
shortened
and
then
so
you
can
go
on
to
yours
or
do
you
want
to
just
you
know
what
just?
Let's
all
assume
that
we
can
hang
on
this
call
for
another
hour
or
what's
your
what's
your
preference
here.
Q
L
B
R
Here
will
do
so
I
stick.
It
send
out
a
survey
about
regarding,
like
IGMP,
v3,
mode2
implementation,
so
I
responded
back
to
that
survey
and
then
I
had
mentioned
like
Verizon's,
worldwide
SSM
deployment,
so
they
wouldn't
make
sent
out
the
agenda
for
the
interim.
I
had
responded
from
adding
to
the
agenda,
so
it
was
good,
useful
use
case
for
operators,
I
guess
with
large
SSM
deployments.
So
with
that
move
on
next
slide,
so
Verizon
we
so
we've
had
a
an
ASM,
I
guess,
architecture
for
many
years
and
we've
always
thought
of
you
know.
R
Kind
of
looking
at
SSM
has
a
way
to
kind
of
simplify
our
architecture,
so
we
so
just
kind
of
given
of
our
architecture.
So
we
have
a
typically
we've
used
SS
at
we
use
multicast
for
a
enterprise-wide
webcast
and
the
enterprise
eunuch
stands
worldwide.
So
we
do
have
over
a
hundred
thousand
some
employees
that
watch
webcast.
So
with
the
with
the
way
art
the
architecture
works
is
we
would
have.
Let's
say
you
could
have.
R
We
could
have
a
remote
presenter
that
could
be
like
anywhere
in
the
world
and
in
and
they
would
actually
send
a
stream
and
it
would
it
be
a
stream,
a
unique,
a
stream
that
would
be
sent
from
anywhere
in
the
world
and
it
would.
It
would
come
down
to
a
broadcast
center
domestically
within
the
US
and
then
then
from
there.
We
would
send
it
to.
R
We
used
a
product
called
ramp,
which
we
we've
used
and
I've
another
slide
on
that,
but
really
that
the
the
main
key
of
that
product
for
multicast
is
that
we
wanted
a
product
that
actually
is
is
able
to
use
a
browser
for
webcasts
and
not
actually
have
a
separate
application
on
the
endpoint
like
a
Windows
or
Mac
or
whatnot
endpoint.
So
if
folks
are
familiar,
say,
there's
a
application,
API
Netscape
and
it's
called
net
NP
API,
so
Netscape
Netscape,
plug-in
application,
programming,
interface
and
most
of
the
newer
browsers
today,
like
Chrome.
R
Firefox
and
Safari,
and
some
of
the
really
more
common
ones.
They
don't
support
that,
so
the
real
challenging
part
was
trying
to
find
a
browser
that
we
didn't
want
to
have
to
load
an
application
on
endpoints,
so
ramp
has
a
product
and
I'll
go
into
that
like
a
little
more
detail,
kind
of
how
it
worked,
and
it
all
supports
us,
a
Sam
which
was
nice.
So
we
were
able
to
you,
know
procure
this
product
and
so
that's
kind
of
really
on
our
endpoints.
R
So
we
have
this
stream
that
comes
in
you
know
so
anywhere
in
the
world,
you'd
have
a
brought
you'd
have
a
presenter,
it
would
get
backhaul
down
to
domestic,
then
that
would
actually
get
yet
dub,
audio/video,
filtered
and
then
shipped
over
the
stream
it
would.
It
would
be
an
HLS,
so
HTTP
Live
Streaming
stream
that
would
actually
get
shipped
over
to
a
receipt
to
a
yob
store
server
and
then
that
server
would
source
the
multicast
out.
You
know
enterprise-wide,
so
we
know
worldwide.
R
So
that's
kind
of
basically
the
overall
architecture
with
this
with
the
deployment
go
ahead.
Next
next
slide,
so
we're
so
SSM.
The
nice
thing
was
with
with
the
product
with
with
the
ramp,
it
did
support
and
even
with
SSM
we
didn't
have
any
issues
with
Wi-Fi.
You
know
Wi-Fi
in
general
with
multicast
is
you
know
it's
kind
of
prone
to
issues,
and
it's
really
just
with
based
on
like
connection
speeds
and
I.
Think
there
are.
R
Some
RFC
is
related
to
that
that
when
you
try
to
connect
over
multicast,
it
drops
to
like
the
lowest
speed,
and
then
you
always
have
drop
packets
and
connectivity
issues
and
we
use
the
product
Aruba
as
the
vendor
that
we
used
for
our
Wi-Fi
deployment
and
they've
got
a
got,
a
feature
that
when
the
SSM
stream
comes
in
it
would
it
would
unicast
that
with
like
an
optimization
down
to
all
the
endpoints,
so
it
would
be
an
incoming
SSM
and
then
it
would
replicate
that
out
to
all
the
associated
Wi-Fi
users,
so
that
was
kind
of
a
nice
success
that
we
had.
R
R
R
You
can
have
like
an
h,
a
type
of
concept
like
active
standby,
where
you
have
to
with
the
SSM
subscribed
channel
when
you
subscribe,
you
can
have
the
same
multiple
sources
for
the
same
group,
so
we
had
an
h
a
group,
so
you
could
have
inactive
with
multiple
backups
or
a
round-robin
style,
where
you're
actually
sharing
a
group,
so
that
was
that
was
really
nice
to
take
advantage
of
that
with
SSM.
So
with
with
the
ramp
product.
What
it
would
do
is
we
would
actually
have
a
source
stream.
It
would
be
ingest
it
would.
R
Multicast
plus
receiver
agent,
it
would
just
take
that
it
would
take
the
packet,
the
payload,
onion
capsule
ate
the
rat,
remove
the
wrapper
and
then
it
would
play
it
out
on
the
local
loopback
and
then
you
could
bring
it
up
on
any
browser.
So
it
was
really
nice
to
actually
not
have
to
have
extra
software.
You
just
have
this
agent
running
and
now
you
can
actually
play
play
it
out
on
any
any
other
common
common
browsers.
So
that
was
really
nice
thanks
next
slide.
R
So
there
were,
there
were
some
challenges
with
Verizon
deploying
multicast
video
with
SSM
and
I.
Think
probably
one
of
the
biggest
ones
was
related
to
SSA
IGMP
v3
support
I
mean
it
seems,
like
you
know,
we've
gone
so
many
years
and
IGMP
v3
has
been
around
for
so
long,
but
they're
still
vendors
out,
you
know,
don't
support
it
or
have
issues
or
you
know
whatever
whatnot,
but
you
know
having
a
workaround
to
get
IGMP
v3
to
work.
You
know
where
we're
getting
us
a
stem
to
work
where
IGMP
is
not
supported
on
the
hosts.
R
That
was
kind
of
a
big
challenge.
So,
overall
it's
a
I
had
to
some
applications
listed
on
the
source
side.
We
had
a
Windows
Server.
That
was
ok.
We
had
allows
the
media
server
did
did
some
of
the
encoding,
and
then
we
had
encoders
and
décor
appliances,
high
vision,
and
that
was
no
problem.
They
all
supported
and
then
of
course,
ramped.
R
You
know
the
product
for
the
endpoint
and
the
source
that
all
supported
it
and
then
the
receiver
OS
as
it
supported
it
like
Mac,
Microsoft,
Windows,
10
and
7
I,
think
Windows
has
been
supporting
I,
think
I
believe
since
Vista,
so
that
was
nice,
Apple
Mac
that
that's
all
supported
in
the
latest
in
the
versions
over
the
last
few
years,
no
issues
and
then
most
of
the
Linux
versions
we
didn't.
We
didn't
have
issues
as
well,
where
we
did
run
into
issues
and
I
actually
said
in
the
slide,
I
kind
of
had
a
Mis
type.
R
It
should
so
where
it
says
receiver
OSS
that
do
it
should
have
been
do
not
so
we
had
issues
with
them.
You
know,
so
they
were
set-top
boxes
where
we
that
we
had
it's
around
the
world
where
we
have
apply.
Basically,
a
set-top
box
of
all
appliances
connected
to
a
plasma
screen,
so
it
get
an
SSN
stream
that
would
come
in
or
a
multicast
stream,
and
then
it
would
hit
the
the
the
last
hop
router
and
the
immuno
would
set
that
box
would
be
sitting
next.
R
You
know
connected
to
the
plasma
and
it
would
actually
do
a
static
join
so
that
join
there
was
issues
with
that
join
actually
supporting,
even
though
the
vendor
sided
supported.
V3
we
just
could
not
get.
We
just
couldn't
get
it
to
work
after
trying.
You
know
so
many
different
syntaxes
to
try
to
specify
like
a
source.
You
know
for
the
joint
so
and
then
we
had
Citrix
VDI
desktops
that
didn't
supported
Chromebooks
lansquenetes
other
miscellaneous,
but
it
was
enough
that
you
know
it
definitely
made
it
difficult.
R
R
R
R
R
So
over
overall
I
think
with
Verizon
with
the
use
cases.
You
know
you
know
as
far
as
a
SM
and
SM
considerations
and
comparisons,
I
would
say,
with
a
SM
I
think
what
we
found.
I
mean
definitely
complexity.
We
really
you
know
tremendous
complexity
with
the
MSTP
meshing
and
peering
and
whatnot
for
inter-domain
routing
and
then
the
RP.
You
know
you
know
having
to
have
you
know
all
the
ACLs
and
whatnot
sa
filters.
R
What
so
that
complexity
kind
of
completely
eliminated
with
SSM,
so
that
was
really
nice
and
then
I
think
the
and
then
I
think
we're
also
with
ASM
the
controls
bar
built-in.
It's
all
network
based
control
vision,
network
based
application
now
based
source
discovery.
Where's
SSM
has
application
based
Louis
discovery.
So
all
that
control
mechanism
built
in
ASM
is
all
done
on
the
network.
So
you
have,
you
know
a
million
ACL,
so
you
have
a
lot
more
complexity
due
to
those
ACLs.
Where
you
know
it's
all,
there's
always
a
trade-off.
R
You
know
with
a
double-edged
sword
with
that,
with
essence
am
having
application
based
source
of
discovery.
It's
all
built
in
the
application.
But
now
you
don't,
of
course
there
are
some
near
cons
and
not
having
the
network
based
sources
cover
e,
but
the
nice
thing
is
all
your
controls.
Are
there,
so
you
don't
it
eliminates
ACLs
and
whatnot
that
you
have
to?
You
know,
manage
and
administer
on
the
network
with
us
at
Sam,
so
overall
that
really
it
really
simplified
simplified
the
architecture.
R
You
know
just
going
completely
SSM
you
know
and
and
go
ahead
next
slide.
So
this
is
kind
of
like
an
idea
of
like
SAS
M.
You
know
with
the
MSTP
meshing.
Just
kind
of
show
like
this
is
like
a
typical
region
where
we
have
s
it
had
ASM,
and
all
of
that
elimination
was
really
really
nice
to
completely
eliminate
all
the
MSTP
peering,
so
that
would
that
was
really
a
big,
huge,
huge
plus
and
again
with
us
with
them.
Go
ahead
next
slide.
R
So
this
kind
of
just
so
we
we
have
webcasts
that
we
have,
you
know
almost
almost
daily
and
they're
like
to
every
alder,
our
Verizon
employees.
So
it's
kind
of
nice
they've
kind
of
a
big
success.
You
know
with
SSM,
so
one
hundred
thirty
nine
thousand
four
hundred
employees.
You
know
worldwide.
You
know
whether
you're
on
a
wired,
LAN
or
Wi-Fi,
you
know
you're
getting
that
multicast
b2b
SSM.
So
that
was
a
big
big
win
for
us
and
I
guess,
overall
I
guess
the
rest
of
family
gets
a
nice
big
worldwide
deployment.
R
R
So
these
were
just
go
down
to
the
last
last
two
slides
as
you
go
up
to
go
up
to
more
yeah
right
here,
yeah
right
here,
so
this
was
something
that
actually
came
up
when
we
were
deploying
you
know
with
with
ASM.
You
have
that
domain
concept
in
you
had
ERP
concept,
so
you
know
with
SSM
I
guess
the
actual
RP
context
completely
goes
away.
So
now
you
have
just
one
big
SSM
domain,
so
you
know
there
are
we
in
our
accused
case?
R
We
didn't
have
it,
but
there
may
be
other
operators
that
may
have
used
cases
where
you
have
different
administrative
domains
or
for
whatever
reason
you
want
to
like
break
up
your
your
big
domain
or
your
tree
but
segmented.
So
the
only
idea
is
that
I
could
think
of,
like
you
know,
probably
outside
of
like
really
at
the
at
the
SSS.
You
know
at
the
multicast
layer,
but
maybe
you
know
to
provide
earlier.
You
know,
maybe,
if
you're
doing
in
Ras,
because
of
course
way.
R
If
you
have
a
you,
know
large
large
network,
you
know
you
may
be
going
through
providers,
multiple
service
providers
or
multiple.
You
know,
enterprise
providers
so
doing
an
nras
to
segmented
or
I.
Guess,
if
you're
doing
SR,
if
you're
running
s
are
either
SR
MPLS
or
s
or
v6,
using
SRT
to
segment
it
outside
of
that
I
couldn't
think
of
any
other
real
way.
It's
not
a
use
case
that
we
had
today,
but
it's
something
that
was
thought
of
and
it's
something
for
M
go
in
and
and
maybe
the
PIM
workgroup
to
think
about.
R
You
know
if
that
use
case
ever
came
up
and
somebody
was
migrating
from
ASM
and
you
got
all
these
domains
with
the
RP
context.
You
know
now
everything
is
just
one
tree,
but
you
still
want
to
have
some
kind
of
like
administrative
segmentation
that
doesn't
really
exist,
I
think
winter
with
SSM.
Now
it's
really
one
tree
that
build
all
the
way
through
and
those
are
the
two
ideas
that
I
had
thought
of
go
on
to
the
next
next
slide,
and
this
is
a
final
slide.
R
So
is
another
thought:
I
had
and
just
kind
of
just
deal
it
working
with
essence,
a
man
after
our
deployment
just
kind
of
food
for
thought.
So
with
SSM
I,
guess
you
don't
have
network-based,
horse
discovery
and
ASM
dollars
and
just
kind
of
the
pros
and
cons.
You
know,
as
a
femme
definitely
has
a
lot
to
gain.
You
know
with
the
simplicity
which
is
really
nice,
so
an
idea
I
had
was
you
know,
maybe
using
like
Saffy
to
multi
cast
on
oral
and
lri.
R
You
know
to
actually
did
say
if
you
had
sources
and
they
were
geographically
dispersed,
but
the
challenge
with
application
based
sources
discovery
is:
if
you've
got
you
know,
if
you,
you
have
hundreds
of
application,
it's
a
view.
Many
applications
not
even
hundreds
our
menu
about
many
applications,
but
then
they're
all
and
you
have
all
these
disparate
sources
there
from
you
know,
they're
spread
everywhere.
Is
there
any?
Is
there
an
easy
way?
You
know?
And
if
let's
say
you
know,
the
application
basis
discover
really
being
not
as
scalable.
R
If
you
have
a
large
number
of
score
sources,
how
could
you
actually,
you
know,
make
SSM
provide
for
network
based
sources
cover
you.
So
this
is
one
idea
I
had
the
only
caveat
is
I.
Think
I.
Think
I
did
something
like
basic.
You
know
like
proof
of
concept
and
then
I
as
an
idea
and
I
think
it
would
work
I.
Think
in
general,
this
happy
to
I
mentioned
it's
used
for
in
congruent
technologies.
R
You
know
where
you're
in
congruent,
unicast
and
multicast
with
ASM,
where
you
know
where
MSD
p4
PRP
of
check
but
using
a
process
that
would
just
be
to
basically,
you
know,
advertise
all
sources
who
all
the
receivers
get
all
did
last
operator
gets
all
the
sources
and
then
you
could
actually
map
the
sources
to
group.
So
just
an
idea
and
then
I
just
lasted
I,
guess
questions
or
comments.
That's
that's
all
I
had
Mike
awesome
thanks,
I
think
Tyler
said
a
question.
N
Thank
you.
So
you
listed
the
compatibilities
on
the
receiver
side,
which
obviously
is
very
interesting
to
us
and
I'm
very
surprised,
and
you
know
somewhat
doubting.
For
example,
chrome
OS
shouldn't
support
IGMP
v3.
Is
it
possible
that
this
is
just
you
know,
missing
support
for
SSM
in
the
receiver
application,
or
could
you
elaborate
on
what
you
try
to
determine
that
the
OS
ready
doesn't
do
IGMP
v3,
so
you.
R
Know
what
I
think
that
would
be
Chromebooks
I
think
it
was
really.
You
know
what
I've
it's
been
some
time
since
I've
tested
but
I
think
Chrome
the
Chromebooks.
They
did
not
support
multicast,
not
just
say
yes,
every
day,
SM
or
SSM.
It
was
just
unicast,
but
it's
been
it's
been
a
little
bit
of
time.
So
I
think
it
was
really
so
it
was
not
really
kind
of
from
Nia,
so
I
think
I
think
once
it
supports
a
assembler
would
support
us.
I
mean
I
should
have
probably
mentioned
that,
but
it
didn't
support.
R
R
R
The
application
they
were
really
just
web
pages,
their
homegrown
applications.
It
was
just
this
you'd
had
a
web
portal
and
the
portal
would
have
whatever
group.
So
the
information
formation
was
off.
The
portal
like
the
channel
channel
einem,
would
be
on
so
it'd
be
homegrown,
so
we'd
have
our
own
application.
That
would
be
like
a
head
end
page
that
you
would
hit,
and
that
would
have
that.
Would
that
would
have
the
group,
whatever
groups,
I,
guess
the
channel
lineup
saying.
R
R
The
host
would
actually
do
that
join,
but
it
would
actually
be
it
would
be
on
the
it
would
be
on
this
when,
when
the
stream
was
when,
when
the
redirect
happened
to
the
whatever
stream,
so
let's
say
whatever
stream
you
were
gonna
head,
it
would
actually
be
a
redirect
and
then
that
redirect
would
be
to
that
ramp
source
server,
and
then
that
would
actually
send
that
that
that
would
that's
that.
That
would
give
you
the
NBS
comedy
channel
so
that
subscribe,
and
then
you
would
do
that
scripts
right.
That's
worse!
Right!.
R
Yes,
yes,
yes!
Yes!
Yes!
So
on
the
receiver,
there's
an
agent
and
it's
what
the
is,
what
the
agent
is
doing
is
it's
actually
sending
a
joint
out
and
then
once
it
once
the
yeah,
because
I
think
I
talked
about
it
earlier
on
a
slide.
But
when
we,
when
the
when
the
stream-
let's
say
our
dart,
be
the
native
content,
is,
let's
say
an
HLS
where
which
we
use
and
it's
to
be
live
streaming.
R
It
would
be
capped
and
packaged
in
a
you,
DB
payload,
so
it'd
be
sent
UDP
or
multicast
through
the
distribution
tree
down
to
the
receiver.
The
receiver
would
I'm
unpackaged
acidity
and
then
it
would
then
it
would
actually
play
visitor
before
that.
I
did
miss
a
step.
So
there
was
so
the
there's,
an
agent
Reed
grunting
on
the
receiver.
He
would
send
out
to
join
so
here
that
stem
that
joint
out.
The
next
of
that
channel
join.
R
R
You
so
I
think
I
was
a
very
good
success
without
time
and
I.
Think
overall
and
it's
been
pretty
solid,
I
think
for
MTTR
like
mean
time
to
recovery
into
complexity
and
not
having
to
do
with
like
shortest
path
switchover.
You
know
that
I
think
that's
always
been
kind
of
a
nightmare.
You
have
a
shader
tree,
but
now
you
don't
have
your
s
comedy
and
you're,
not
building
your
tree
that
that's
that's
always
been
a
nightmare
to
troubleshoot.
You
know
trying
to
figure
out
why
that's
shortest
paths,
whichever
is
not
happening,
so
it's
nice.
R
Just
you
know
the
tree
just
build.
You
know
as
long
as
your
unit
is
not
having
an
RPF
check,
though
you're
like
ten
design,
this
anywhere
along
the
path.
You
know
it's
and
there's
no
easy
else
to
deal
with.
Everything
is
just
kind
of
kind
of
flows
and
all
your
control
mechanisms
are
built
into
the
application.
You
know
visit
the
applications
providing
that
channel
to
join
so
that
was
really
nice.
I
would
say
for
sure
that
that
really
made
that's
what
I'm.
You
know
huge
win
for
Verizon.
Yet
a
question.
S
R
T
R
T
I
I
Q
I
R
T
L
T
R
Multicast
boundaries
and
that
there
are
some
cases
where
we
had
to
do
that:
multicast
boundaries,
so
I
think
I
mean.
As
you
said,
you
still
have
that
s,
comma
G,
like
even
with
the
ASM.
You
know
when
you
switch
over
to
that
shared,
create
the
shortest
path
tree
it.
It
looks
just
like
that's
the
time
you
got
that
shortest
path
tree
so
I
said.
D
T
R
F
L
R
L
R
T
So,
instead
of
doing
Safi
to
and
and
MBG
peakins
can
do
this,
and
in
fact
it's
doing
it.
An
ID
in
MVP
ends
be
GPM
VPN
with
type
a
life
outs.
And
if
you
look
so
the
best
working
group
just
adopted
a
basically
feed
PM
cast
and
multicast
bgp
or
bgp
multicast,
or
what
it's
called,
but
essentially
that
you're
carrying
all
the
multicast
state.
You
know
the
source,
the
group
and
I'm,
pretty
sure,
also
carries
the
si.
Information
is
what.
R
I
think
it
does
I
think
with
the
MVP
n
I
think
you
have
like
the
types
I
think
for
like
yes,
I
guess,
that's
true.
So
you
know,
I
was
just
actually
trying
to
draw
an
analogy
like
that,
but
I
think
with
them.
With
with
that
right
with
that
with
MPLS
Oh
SR
I,
guess
you
have
the
MVP
n
procedures
like
that
are
6513
6514,
so
you'd
you
would
actually
do
like
I
think
it
was
like
type
4
type,
5
and
type
4
I
believe
is
like
dennis
is
the
source?
Is
the
yeah.
R
R
T
D
Yeah
you
for
doing
this
I'm
interested
in
following
up
offline,
because
in
my
presentation
that
we
clearly
should
but
I'm
gonna
keep
doing
anyway,
so
M
Bundy
first
time
indeed,
slot
have
a
call
for
participation
at
the
end
and
based
on
what
even
subscribed
here
it
sounds
like
you
might
in
interesting
candidate.
So
I
would
I
look
for
contact
information,
your
slides,
I
didn't
see
any.
Can
you
either
reach
out
to
me?
I'm
Jay
Holland
did
Akamai
or
paste
your
contact
at
Akamai,
dot-com
or
Facebook
contact
in
the
chat.
Perhaps
a
second.
D
T
T
Yeah
so
I
guess
we'll
use
the
same
ground
rules
that
we're
using
with
PIM,
which
is
ad
plus
and
minus
your
name
in
the
chat
window
and
we'll
put
you
in
the
queue
and
I
would
say.
Please
sign
the
blue
sheets
if
you
haven't
already
and
if
somebody
could
paste
the
blue
sheet
thing
last
I
looked,
there
was
about
30,
27
or
28
sign
names
in
the
blue
sheet
and
there
was
like
39
participants.
So
there
may
be
a
few
people
missing
all
right.
So
bringing
up
the
slide
here.
T
T
H
T
Well,
we're
still
under
the
PIM
Noel,
but
it's
our.
So
it's
the
same
note.
Well,
if
you
so
just
keep
in
mind
all
the
things
and
note
this
well,
here's
our
agenda,
a
quick
status
of
working
group
items
has
three
drafts
under
the
collective
multicast
of
the
browser
umbrella
and
I'll,
give
common
an
update
on
the
multicast
to
the
grandma
deployment
architecture
that
I
presented
I
think
it
was
in
5/4
104.
T
Okay,
alright,
so
status
of
the
currently
active
working
group
items,
Jakes
Dryad
draft
is
in
off
48
keep
an
eye
out
that
will
that
is
eminently
about
to
become
RFC,
we've,
eight,
seven,
seven,
seven,
something
like
that,
so
an
early
congrats
to
Jake
on
that
a
deprecated
ASM
draft
which
Jian
might
be
checking
out
it's
because
it
kind
of
covers
a
lot
of
some
of
the
topics
that
you
talked
about.
It's
currently
at
the
RFC
editor
queue
so
that
should
yeah
published
very
shortly
the
multicast
problems
draft
so
that
one's
still
awaiting
some
revisions.
B
Yeah
so
we've
been
kind
of
going
back
and
forth
between
authors
and
the
iesg
to
take
care
of
some
issues
and
then
recently
as
of
last
week,
there
was
something
that
you
may
have
seen
that
Venus
loosing
brought
up
on
the
list
with
regards
to
issues
with
MLD
and
an
Android
power
saving
and
sweet
nut.
Wouldn't
you
know
it
shared
some
solutions,
he
thought
maybe
doable
and
so
I
invited
leanness
to
join
us
today.
B
B
One
one
big
rebel,
a
transport
type
issues
that
the
transport
area
gave
us
to
do
and
it's
a
little
bit
overwhelming,
but
we
just
got
to
get
it
done.
Cuz
we
put
so
much
effort
into
it
and
then
new
issues
like
what
I
just
mentioned
from
we
just
keep
popping
up
and
we're
like
okay,
fine
one
could
it
be
eventually
we
just
gotta
just
done
and
publish
it,
but
it
seems
to
be
a
worthwhile
document.
T
B
You
know
what,
if,
if
any
and
I
need,
who
are
the
authors
in
the
draft
need
to
let
it
go
we'll?
Let
it
go.
If
it's
not
worthwhile
to
the
working
group,
we
thought
it
would
be.
If
there's
not
really
interest
in
sending
it
off
the
AIA
to
the
iesg,
then
I
mean
it's:
it's
a
draft
it's
recorded
and
we'll
use
it
as
a
you
can
use
it
at
your
reference,
but
I
again,
I
will
just
take
the
working
groups.
B
T
So
what
here's,
what
I
would
say,
I've
I,
read
I,
was
one
of
the
commenters
I,
think
value
I!
Think
it's
you
know
within
charter
and
the
folks
who
have
responded
and
reviewed
the
draft
seemed
to
say
similar
things
that
there
is
value
here.
There
just
hasn't
been
enough
so
kind
of
a
message
to
the
working
group.
B
T
B
T
An
easy
read,
take
a
look
and
comment
if
you
think
that
this
is
of
value
and
we
should
advance
it
if
you
don't
have,
if
you
don't
think
it's
the
value
then
say
so
as
well,
but
we'd
love
to
have
more
comments
and
thoughts
here
and
provide
feedback
on
this
one
Greg
great.
Did
you
have
any
other
thoughts
on
on
this?
One,
blue,
giraffe.
T
Here's
the
question
Mike
because
a
cat
a
lot
of
substantive
comments.
Would
you
want
to
rev
the
draft
based
on
Jake's
comments
or
because
it
doesn't
make
sense
for
you
to
rev
it
and
then
find
nobody
else
know
he's
interesting
or
do
we
want
to
do
one
last?
You
know
chance
at
it
and
see
who
all
is
interested
yeah.
T
T
D
Wanted
to
reiterate
I
think
I
said
in
my
in
my
review
that
I
do
think
there
is
value
in
that
draft
and
I
think
it's
pretty
good,
just
yeah,
so
I
I
don't
mean
to
derail
it.
If
my
comments
are
there,
I
just
wanted
to
mention
that
and
and
encourage
the
authors.
Okay,
well,
yeah
I
see
I
see
where
you're
coming
from,
but
I
kind
of
wish.
We
could
get
more
comments
too
yeah.
Thank
you.
P
P
I
D
I
D
D
Q
P
Q
D
D
No
I'm
I'm
trying
not
to
realize
your
fears,
so
I
guess
the
one
thing
I
want
to
say
up
front
before
I
lose
people
I
have
a
call
to
participate
at
the
end
I'm
looking
particularly
for
carrier
partners
to
ingest
traffic
according
to
the
architecture
outlined
here.
We
want
to
validate
the
architecture
and
make
sure
that
we
get
something.
D
That's
really
going
to
work
and
we
are
in
talks
with
a
number
of
people,
and
so
anybody
who
is
interested
please
just
jump
to
the
last
slide,
send
me
an
email
and
we
can
now
cup
offline,
but
I
will
go
through
a
quick
update
at
the
slides,
a
quick
update
of
the
of
how
things
stand
or
the
benefit
of
anybody
who
can
stick
around
yeah.
So
the
next
slide,
please.
D
So
as
you
after
adopted
I'm,
not
going
to
do
a
detailed
presentation
about
the
about
the
techniques.
There's
a
link
there.
If
you
need
to
review
it
or
you'd
like
to
watch
it
the,
but
just
as
a
quick
reminder,
there's
three
main
things:
dorms
AM,
B
and
C
back
and
what
they
do
is
the
metadata
is
in
dorms.
That's
the
the
substrate
that
provides
the
metadata
ambi
is
the
one
that
provides
the
data.
Integrity
and
Seebeck
is
the
one
that
provides
hooks
for
bandwidth,
management
and
I
have
just
lost
a
view
of
your
slides.
D
Yeah
that
adequate
yeah
go
on
next
slide.
Please
so
I
got
a
great
set
of
office
reviews
on
all
three
dress
from
Dino.
There
were
a
lot
of
comments
on
Andy,
with
several
rounds
of
back
and
forth.
There
were
that
some
substantial
comments
on
feedback
and
some
brief
comments
on
dorms
I'll
go
over
a
little
bit
of
what
my
DVDs
are
on
these
in
light
of
having
gotten
these
these
comments
so
far,
obviously,
it's
still
interested
in
more
feedback
from
other
people
and
big
thank
you
did
you
know
and
yeah
next
slide.
D
That
is
my
the
word
count
by
the
way
so
15,000
included.
This
is
basically
just
a
text
dump
of
the
email
threads,
but
you
know
1,500
lines
and
15,000
words
with
you
and
me
it
was.
It
was
good
times
I
encourage
getting
a
dinner
review
anytime.
You
can
all
right
next
slide,
please
yeah!
So
for
dorms
the
comments
were
were
pretty
straightforward:
I
have
several
DVDs
that
are
still
in
there
and
I
didn't
list
was
as
I
still
need
to
add.
J
D
D
Had
the
slides
back
or
okay,
there,
we
go
something's
coming
great,
thank
you
yeah!
So
there's
just
a
few
minor
clarity
enhancements
and
a
clarification
about
the
the
metadata.
That's
provided
my
incoming
assumption
with
dorms.
As
you
already
know,
the
SG
out-of-band
there's
not
a
a
network-based
discovery
of
the
SG,
so
this
is
coming
from
apps
from
app
logic
of
some
sort.
D
The
primary
value
comes
from
spreading,
providing
that
that
metadata
to
other
management
domains
that
are
not
otherwise
privy
to
the
app
level
data,
so
I
thought
those
were
useful
suggestions
and
I'll
be
watching
that
out,
I've
not
updated
and
he
had
the
docs.
Yet
this
was
just
like
six
weeks
ago,
which
feels
like
a
longer
time
by
now,
yeah,
that's
like.
D
D
That
tries
to
not
collide
with
the
circuit
breaker
terminology
so
heavily
and
to
and
to
explain
more
in
context
of
a
sort
of
Seebeck
centric
way
of
looking
at
it
with
just
a
section
that
sort
of
ties,
those
that
terminology
together
and
and
points
out
exactly
how
it
maps
to
the
circuit
breaker
concepts.
So
that's
going
to
be
a
probably
a
fairly
significant
text.
D
Overhaul
I
think
I'll
be
aiming
to
get
that
done,
as
as
some
of
the
other
workers
for
produced
little
actually
worthwhile,
so
Andy
and
also
an
operational
consideration,
section
I
think
is
another
really
good
suggestion
to
add
and
to
be
recover.
The
concept
of
what
to
do
when
there's
multiple
sivak
circuit
breakers
in
the
same
network
and
possibilities
for
some
optimizations
that
a
good
perform
that
are
out
there
for
the
document,
but
to
point
the
way
to
some
some
useful
deployment
models.
D
D
D
D
Even
if
there
is
a
you
know,
a
requirement
in
within
that
context
that
integrity
be
provided
as
long
as
the
integrity
is
provided
in
some
other
way.
That's
also
fine.
One
thing
that
might
work
is,
for
example,
a
generic
UDP
solution
for
Tesla.
Now
some
of
the
people
don't
like
ambi,
also
don't
like
Tesla,
and
you
know
so
I'm-
not
sure
that
that
will
necessarily
do
the
job
I
like
any
better,
so
I'm
gonna
start
with
with
that
one.
But
at
some
point
I
might
end
up
writing
the
Tesla
idea
as
well
and
certainly
I'll.
D
Any
developments
along
these
lines.
Look
for
future
updates
to
go
there,
but
this
is
this,
is
you
know
very
much
sausage-making
kind
of
figuring
out
where
we're
going,
but
for
now
the
plan
is
still
to
go
forward
with
it
with
an
AM
be
implemented
and
to
get
and
to
get
that
moving
pending
any
better
ideas.
So
what
I'm
hoping
to
do?
But
this
also
is
the
dock
that
I
want
to
get
furthest
along
I.
D
Think
it's
the
most
goal
for
getting
things
into
the
browser,
and
so
what
I
would
like
to
do
is
get
my
my
top
priority
in
the
dock.
Work
is
to
get
this
into
shape
for
requesting
a
security
review
and
make
sure
that
it's
in
sort
of
a
reasonable
State
for
a
first
round
heavy
technical
review.
Yeah
next
slide.
T
E
Well,
I
think
there
is
a
soup,
a
simple
solution.
If
you
just
say
we
use
a
symmetric
keys
and
what
I
mean
was,
if
you
just
put
a
signature
in
the
packet
and
have
the
source
sign
it
with
its
private
key
and
it's
like
he
has
put
in
dorms,
then
all
receivers
could
verify
each
packet
with
the
public
key.
However,
we
believe
that
there's
no
good
asymmetric
algorithms
that
are
fast
enough.
If
we
can
fast
enough,
would
you
think
that's
a
better
solution?
Question?
Yes,.
D
That
that's
a
fair
point.
So
if
we
had
an
adequate
cipher
suite-
and
we
were
reasonably
confident
that
that
we
could
do
this
at
the
right
level
of
performance,
then
yes,
I
would
have
done.
That
was
my
first
idea.
In
fact,
and
the
you
know
my
section
about
that
implement.
So
that
is
going
to
be
one
of
the
designs
for
sure
that'll
go
and
then
cover
in
the
rejected
designs
that
don't
work
and
I'll
be
I'll.
D
Explain
how
the
reason
it
doesn't
work
is
because
of
benchmarks-
and
you
know
point
to
something
saying
about
that
and
the
the
like.
The
only
reason
that
doesn't
work
is
because
of
the
performance
like
if
you
try
to
run
video
that
way,
you
can't
do
it
and
that's
you
know
not
on
not
on
today's
Hardware,
maybe
one
day
we
can
do
it.
Jake.
E
Do
you
think
any
of
the
security
area
guys?
Are
they
working
on
any
new
state-of-the-art
stuff
to
make
that
work
better?
Do
you
know
because
I
know
I
know
elliptic
curve
and
Edwards
curve
is
much
faster
than
RSA,
but
obviously
not
fast
enough
right
right.
D
D
You
know
another
document
that
can
they
can
do
the
simpler
job
and,
and
that,
like
you,
could
very
well
define
a
new
asymmetric
signature
approach
where
you
inject
the
signature
as
a
shin
into
a
UDP
packet,
and
you
have
sort
of
a
generically
authentic,
awp
payload
that
you
can
deliver
in
exactly
the
same
way.
Doing
exactly
what
you
just
said.
I
do
agree
that
works.
D
If
you
go
slow
enough,
my
problem
is:
it
doesn't
fit
my
use
case
because
I
want
to
do
video
and
you
know,
among
other
things,
and
and
nobody
can
sign
it
fast
enough
and
nobody
can
verify
it
fast
enough,
and
so
the
the
only
problem
with
that
approach
is
that
it
doesn't
work
because
of
performance.
So
I
do
think
that
although
I'm
not
going
to
use
it
having
that
protocol
in
our
back
pocket
for
when
a
when
a
goes
algorithm
when
a
fast
enough
algorithm
comes
out
might
be
worthwhile
for
the
future.
D
E
Quick
question:
if
we
can
authenticate
if
we
can
use
an
authentication
mechanism,
that's
different
than
the
integrity
check.
Would
that
work
as
well?
Because,
if
like
in
my
comments
to
you
I
said
if
we
could
sign
a
few
set
of
bytes
like
16,
bytes
or
even
eight
bytes,
that
would
be
relatively
faster
to
authenticate.
But
you
want
by
working
now,
you're
you're
doing
both
with
one
solution
right.
So.
D
E
Could
we,
if
we
had
a
different
mechanism
to
do
integrity,
checking
than
authentication
you
could
make
authentication
go
fast
now
to
make
integrity
go
fast?
Unfortunately,
you
have
to
go.
You
have
to
process.
Every
byte
of
the
packet
has
to
be
part
of
either
a
checksum
or,
or
you
know
what
I
mean
or
yeah.
D
D
There
are
zero
days
in
video
players
and
we
cannot
allow
it,
in
my
opinion,
browsers,
it's
not
safe
to
allow
browsers
to
accept
on
a
thank
you
data
it
because
there
are
easy
injection
paths
in
the
network,
and
it's
and
I
mean
that
is
my
opinion
and
also
the
the
opinion
of
every
person.
I've
talked
to
who
checks
code
into
the
browser.
So
I
I
really
think
that
it
has
to
be
a
straight
and
secure
solution
that
provides
integrity.
D
D
So
I've
implemented
a
very
simple
dorm
server
by
taking
jack
conte
and
putting
right
yang
models
into
it.
I
have
it
sort
of
it's
it's
a
work
in
progress,
it's
in
the
kind
of
infrastructure
that
that
becomes
a
thing
that
gets
deployed,
not
just
a
like,
barely
running
on
my
on
my
laptop,
but
but
something
with
an
eye
toward
getting
it
there.
It's
you
know
not
something
I
can
demo
to
you
today,
but
it
is
something
that
is
on
its
way
to
being
something
that
you'll
be
able
to
access
and
next
slide.
D
The
browser
work
we
had
started
a
dev
team
that
started
working
on
it.
So
it's
not
just
me
hacking
on
the
weekend
anymore.
It
is
you
know,
people
who
actually
do
it
for
real,
still
and
and
and
they're
doing
a
great
job.
They
I
mean
it's
amazing.
They've
got
like
tests
and
everything
it's
awesome
and
the
our
internal
goals
before
we
before
we
engage
externally
are
coming
to
a
head
imminently.
D
Our
internal
of
proof
of
concept,
sort
of
criteria
is
to
make
sure
it
actually
fits
our
use
case
and
we
are
going
to
so.
We
have
an
LMS
product.
That's
that's
how
we
deliver
stuff.
It
sounds
actually
quite
similar
to
the
RAM
product
described
before
and
also
quite
similar
to
the
new
DVD
spec
that
came
out
from
the
DVB
group,
but
the
this
is.
D
This
is
a
deployed
product
and
our
intent
is
to
just
leave
an
a
roughly
unmodified
server
version
of
that
encoding
data
according
to
the
way
that
we
encode
data
and
to
use
the
webassembly
SDK
that
we're
sticking
into
the
browser
or
sorry
to
use
the
API
that
we're
sticking
into
the
browser
in
a
web
assembly.
Sdk
that
can
be
delivered
through
a
page
with
the
new
proposed
API
that
we're
working
on
and
to
and
to
actually
make
that
play.
D
Video
using
the
browser's
MSE
ap
is
for
for
playing
video
and
then
the
next
steps
after
that
are
the
lean
in
on
getting
our
API
up
streamed
into
chromium
to
the
best
of
our
ability.
So,
obviously
this
depends
on
many
external
factors
and-
and
we
are
doing
our
best
to
do
a
a
credible
and
worthwhile
job,
it
will
face
something
you
know
there
with
any
new
kind
of
API
like
this.
There's
always
questions
about
like
how
important
is
this
participation
and
comments
about
the
w3c
API,
especially
after
we
update
it,
because
we
have
some
updates.
D
We're
going
to
be
doing
will
be
will
I
feel
like
will
be
helpful
to
getting
the
sort
of
credible
for
progress
on
that.
That
has
some
obvious
external
support,
so
I'll,
be
maybe
I,
don't
know
on
a
monthly
cadence
or
so
begging
for
people
to
chime
in
in
various
forums.
If
they,
if
they're,
able
to
and
I,
would
very
much
appreciate
it
support
there,
it
might
make
the
difference
between
things.
Moving
forward
or
not
and
yeah,
that's
it.
D
D
If
you
wanted
to
scale
so
the
and
then
obviously
the
doc
updates
for
the
w3c
API,
as
well
as
the
three
IETF
draft
and
and
I
am
hopeful
that
at
some
point
the
multicast
ingest
platform,
the
the
published
and
also
talked
about
here
before
that
we
would
be
adding
and
the
authentication,
as
a
sort
of
you
know,
ingest
firewall
checker
did
that
if,
if
it
turns
out
reasonable
but
that's
a
I'm,
not
making
any
promises,
I'm
just
saying
it's
on
my
Ritter
and
maybe
we'll
do
it
next
slide.
Please.
D
So
we
are
currently
looking
for
participants
to
work
on
these
kinds
of
trials.
We
have
either
several
people
we're
talking
to
yeah
next,
please
in
in
some
we
have
several
content
owners
that
are
that
have
already
said
yes
and
that
are
willing
to
they're
willing
to
work
with
us.
We
are,
we
have
several
carriers
that
have
not
said
no,
but
are
also
not
like
a
clear.
D
So
it's
not
just
our
traffic
but
also
potentially
other
people's
traffic.
So
we
are
looking
for
contacts
with
with
carriers
introductions
to
people
that
you
think
would
be
interested
in
this
from
carriers,
and-
and
this
is
a
very
timely
part
time
to
get
started
with
this,
because
if
we,
if
we
you
know,
have
reached
out
to
30
people-
and
we
get
30
knows
you
know
within
the
next
month,
then
we
there
is
a
chance
that
we
would
put
the
shot
put
this
on
a
shelf
and
come
back
to
it
some
other
time.
D
N
Thank
you
very
much
great
progress,
so
I'm,
not
sure
you
know
what
you
really
like
to
do
about
MB
right.
It
certainly,
you
know
cryptographically
a
very
interesting
thing
to
do,
but
if
you
feel
that
this
is
also
a
possible
roadblock
for
more
adoption,
then
I
think
it
would
be
good
to
better
analyze
under
which
circumstances
MB
is
really
necessary
right.
Maybe
that
could
be
done
by
appropriate
exponential
explanatory
text
in
ambi
itself.
N
The
way
I
see
it
is
that
you
know
M
is
really
only
necessary
if
you
have
something
like
free
content
that
you
are
able
to
stream
to
non
DRM
encumbered
receivers
right
as
soon
as
you
have
DRM
encumbered
receivers.
I
think
the
the
whole
picture
already
changes,
because
these
receivers
have
to
be
trusted
to
comply
with
the
DRM
system
and
all
the
existing
broadcast
systems
are
based
on
the
fact
that
you
can't,
basically,
you
know,
extract
the
keying
from
Saturday.
N
Our
M
receiver
not
only
do
to
decrypt
on
other
places,
but
also
not
to
be
able
to
re-inject
fake
content
using
the
same
key
and
I
think
for
most
of
the
commercial
traffic.
We
unfortunately
have
to
fear
that
zebras
work
erm
encumbered,
but
that
should
relax
the
requirement
on
something
like
MMB
and
instead
moves
it
into
the
DRM.
D
That
is
an
interesting
approach
and
yeah
I'd
be
willing
to
explore
that
maybe
I'm
concerned
that
that
would
not
satisfy
the
kind
of
threat
model
that
that
is
so,
if
you,
if
you
consider
RFC
3550
to
I,
think
it
is
in
section
3,
where
it
talks
about
the
internet
threat
model
and
the
notion
that
you
know
we
have
obviously
observed
on
path
attackers
for
video
traffic.
In
the
past
things
like
ad
hijacking,
where
you
know
somebody
who
is
in
position
to
actually
just
drop
in
actually
different
traffic.
D
N
Is
no
there
is.
There
is
nothing
that
has
seen
more
money
and
worse
being
used
than
broadcasted
DRM
traffic
right.
So
the
whole
premise
of
being
able
to
lock
down
receivers
to
the
extent
that
symmetric
encryption
keys
cannot
be
extracted
from
them
is
I,
think
something
that
has
been
proven
the
most
on
the
planet.
So
that's
actually
the
answer
to
exactly
that
threat
model.
D
N
N
And
the
DRM
would
already
be
the
piece
yeah
I
think
you're,
just
you
know
removing
the
requirement
of
MB
when
you
say
I
have
a
DRM
system.
That
is
exactly
one
of
these
existing
DRM
systems.
Right,
we
can
still
argue
about
whatever
other
channel.
Is
there
to
authenticate
a
DRM
receiver
and
give
it
the
decryption
tree.
But
it's
locked
down
the
keys
are.
Q
We
are
really
pushed
hard
to
decline,
but
email
lists
and
what
I
ask
if
you
have
text
specifically
offer
the
text
to
discuss
on
the
on
the
list.
Yeah.
Thank
you,
like
almost
you
know,
at
40
minutes
over
and
we
slow
them
one
more
item
to
get
you
and
now
we're
running
other
people's
agendas.
Thanks
for
your
input,
please
take
it
the
list.
Anyone
else
have
questions.
You've
got
more
stuff
to
do
here.
Jake,
you
got
more
slides,
that's
it.
Okay,
pour
I
want
to
make
sure
we
get
some
traction
on
this.
Q
Jake
I
want
to
see
people
pick
this
up
and
follow
through
and
anyway
we
can
help
as
a
working
group
as
a
chair
is
the
IETF.
Please
let
us
know
you
will
do
something
on
you
know,
specifically
in
the
working
group
and
track
it
that
way.
We
can
just
want
us
to
reach
out
to
the
people,
operator,
communities
and
stuff
I'm
sure
we
can
help
you
make
connections
if
you
don't
already
have
them,
but
it's
not
a
bad
place
to
start
here.
Sure.
D
Q
T
Q
T
T
We've
been
we've
been
making
steady
progress
so
first
off
what
is
multicast
to
the
grandma,
and
that
is,
you
know,
multicast,
it's
so
multicast
over
the
Internet
that
is
ubiquitously
available
over
for
everyone,
not
just
for
the
folks
who
are
participants
on
this
call,
and
it's
also,
you
know
not
just
a
hope
for
a
distant
future,
but
something
that
actually
is
working
today.
So
I'm
going
to
talk
about
what
we
have
in
place
today
and
hopefully
see
if
other
folks
want
to
get
involved.
T
T
And
if
you
just
look
at
the
bit
rates,
you
know
the
numbers
start
to
skyrocket,
pretty
quickly
from
HD
to
4k,
to
8k
to
VR
and
I
stole
this
from
Jake
at
least
the
last
time
I
heard
every
year,
Akamai
apparently
breaks
their
traffic
record
and
the
last
time
I
saw
it
was
72
terabytes.
That
was
the
most
traffic
over
the
Internet
and
if
you
just
divide
72
terabits
by
40
Meg's,
you
know
how
many,
how
many
4k
streams
would
that
be
if
and
then
I'm
the
1.8
simultaneous
4k
viewers?
D
T
And
and-
and
you
know
so,
what
we've
been
doing
thus
far
as
I
would
call
brute
force
unicast
bfu,
and
can
we
keep
up
so
that
was
one
point:
eight
million
viewers,
the
average
audience
size
of
an
the
television
audience
for
an
average
NFL
game,
not
a
soup,
not
the
Superbowl,
not
a
playoff
game,
not
Monday,
Night
Football,
but
an
average
NFL
game
is
15
million
viewers,
15
million
at
40
Meg's
at
4k.
That
would
be
600
terabytes.
T
T
Also
in
the
last
few
years,
I'd
say
the
last
two
years
or
so
the
court
cutting
evolution,
it
kind
of
started
with
things
like
YouTube
and
Hulu
and
and
Netflix,
and
it
was
all
about
on
demand,
but
recently
all
the
energy
it
seems
to
be
in
live
and
you're
seeing.
This
is
like
the
last
stage,
I.
Think
of
of
the
cord-cutting
evolutions:
how
do
we
get
live?
You
know
when
you
talk
to
neighbors,
they
constantly
say.
Well,
what
do
you
do
about
live
television?
T
You
know
cuz
they're
sports,
you
know,
and
actually
before
a
month
ago
there
used
to
be
sports
on
and
it
was
and
people
used
to
watch.
But
if
you
see
you
know,
there's
Hulu,
TV,
YouTube,
TV,
sling,
TV,
so
live
linear
television
is
not
dead.
Yet,
in
fact,
you
know
live
streaming
is
actually
trending,
and
it's
because
we,
you
know,
we
live
in
a.
We
live
in
a
fake,
we're
all
the
fake
news
and
deep
fakes
and
and
there's
it's
we're
becoming
a
much.
T
Our
return
on
the
investment
I'll
go
through
this
quickly
and
it's
you
know
actually
live
streaming,
is
the
fastest
growing
type
of
video
online
right
now,
of
course,
recent
events
with
the
you
know
the
current
pandemic
is,
is
illustrating
more
the
need
for
solutions
that
can
enable
people
to
gather
virtually
together
like
working
group
meetings
or
distance
learning,
worship,
services,
movie,
watch
parties,
happy
hours,
virtual,
happy
hours,
working
group
meetings,
interim
meetings,
things
like
that,
so
we're
seeing
now
that
everything
isn't
just
on
demand.
You
know
live
is
still
alive.
T
So,
what's
new
you
know,
Multi
multi
cast
has
been
around
since
the
90s.
It
hasn't
gained
a
whole
lot
of
traction
and
I'm.
Talking
about
internet
multicast,
regular
multicast,
I,
say
I
would
say,
has
been
fairly
successful,
but
internet
multicast
is
not
ubiquitously
deployed.
It
hasn't
gained
much
traction
well.
Why
should
we
think
it
should
be
different
now,
and
there
are
a
few
things
that
are
different.
First
of
all,
ASM
deprecation
and
that's
you
know
the
draft
that
we've
talked
about.
That's
about
the
becomin
RFC
and
a
BCP.
T
It's
already
been
applied
as
the
best
common
practice
in
Internet
to
which
is
the
largest
component
of
the
mbone,
and
once
you
get
rid
of
ASM
and
you
move
to
SSM
only
we
eliminate
about
90
percent
of
the
complexity
of
multicast,
all
the
stuff
that
people
complain
about
about
multicasts
too
complicated.
Don't
want
to
do
it,
the
juice
ain't
worth
the
squeeze.
Those
arguments
go
out
the
window
when
you
just
deploy
SSL.
Only
when
you
do
SSM
only
you've
got
no
RPS,
no
MSD
P.
T
Q
Got
his
hand
up,
you
got
a
shout,
cuz
I
can't
see.
So
what
do
you
got?
Well,
just
to
the
point
you
know
having
been
one
of
the
early
cheerleaders
for
the
beer
movement.
Some
of
these
you
know
resistance
came
from
get
go,
I,
hate,
multi,
guess,
I,
hate,
multi,
guess
eight
multi,
cashier
people,
you
know
beat
that
drum
and
it
was
clear
early
on
that
they
don't
hate
multicast,
they
hate
PIM.
We
make
that
distinction
in
our
discussions.
Network
replication
has
got
incredible
value.
Q
T
Further
and
say
you
think
you
hate
PIM,
but
what
you
really
hate
is
PIM
ASM,
you,
you
hate,
sparse
mode.
You
hate
rendezvous
points,
you
hate,
em
SDP,
you
hate
all
that
stuff
him
SSM
is,
actually
you
know
the
the
cooler
younger
cousin
of
and
that's
the
one,
that's
the
guy.
You
really
like,
because
he
looks
an
awful
lot
like
LDP.
So.
Q
And
what
you
got
me
ping
me
off
off
off
line
here:
an
email
because
I
did
a
pre
session
like
this
working
with
the
BBC
last
year,
and
they've
got
tons
of
great
new
information
about
how
they're
losing
bandwidth
in
terrestrial
RF
space,
and
they
will
not
have
enough
RF
bandwidth
further
eventing
soon
enough
and
there's
not
enough
CDN
space
in
the
entire
country
of
the
UK
to
handle
even
one
live
event.
So
yeah
they're,
wedged.
T
So
another
development
in
you
know
recent
development
is
overlay
networking.
So,
with
things
like
AMT
and
Lisp,
we
can
address
some
of
the
fundamental
all-or-nothing
problems
with
multicast
when
I
started.
I.
Remember
that
that
tunnels
were
seen
as
a
suboptimal
hack
to
be
avoided
and
we
need
to
go
native
because
tun
will
suck.
All
you
have
to
do
is
just
rename
it
to
overlay
networking
and
it's
cool
now.
T
So
that's
that's
kind
of
a
you
know
a
nice
recent
trend
that
we
can
ride
on
top
of
and
say,
oh
well,
you
know
we're
back
to
the
future
here
with
overlay,
networking,
remember
beers
and
overlay
network
as
well.
Exactly
so
all
the
pieces
are
in
place
for
a
working
solution
today
and
that's
the
key
thing
we
have
something
today
is
and
and
I
think
Dino
said
it
really
well
last
month,
and
you
know
the
thread
that
got
us
here
in
this
meeting.
T
Q
Say
it's
worse
than
that,
because
I've
worked
with
the
team,
where
we
put
together
a
product
and
there's
OHS
on
the
phone
here
that
were
involved,
we
showed
that
it
worked.
We
were
intent
as
all
of
your
you
know,
in
this
room,
probably
the
same
game.
We
look
at
the
network
restrictions
that
come
with
unicast
versus
multicast.
What
we
found
is
the
immediate
savings
was
in
the
data
center.
All
it
took
was
for
concurrent
receivers,
the
same
piece
of
content
and
we
had
a
90
percent
savings
and
data
center
costs.
Q
T
To
the
next
point,
which
is,
this
is
essentially
an
effort
that
we're
calling
the
multicast
the
grandma
architecture,
and
this
is
kind
of
a
grassroots
bottom-up.
Instead
of
this
as
being
one
one
large
corporation,
you
know
catering
to
the
whims
of
the
various
product
groups
and
hoping
that
the
right
VPS
like
this
or
don't
like
the
other
thing.
T
This
is
just
a
grassroots
movement
of
folks
who
are
just
deploying
stuff
and
getting
stuff
out
there,
with
the
hope
that
you
know
we
can
demonstrate
it
works,
and
you
know
get
some
kind
of
interesting
content
out
there
that
maybe
goes
viral
and
and
and
pushes
the
architecture.
So
this
is
more
of
a
bottom-up
approach
than
a
top-down
approach,
and
so
how
would
it
look?
So
this
is
kind
of
the
whole
architecture,
so
we
have
the
Internet.
Most
of
it
is
unicast,
only
some
small
portion
of
it.
We
call
it.
T
The
mbone
is
the
multicast
enabled
portion
of
it
and
for
all
intensive
purposes.
That
is,
you
know,
internet
to
that
is,
you
know,
probably
90%
of
the
Ambo
and
is
Internet
and
the
mbone
comprises.
Maybe
you
know
somewhere
between
you
know,
maybe
1
to
3%
of
the
Internet,
so
you
have
a
multicast
source
and
a
multicast
receipt
native
source
native
receiver
on
the
emam
that
works
the
way
it's
always
worked
right.
We
have
PIM,
sparse
mode,
still
works
today
running
over
i2.
T
Now,
imagine
you
have
an
off
that
receiver
and
that
is
a
receiver
who
is
on
a
unicast
only
network?
Well
we
have
these
AMT
relays
that
have
been
deployed.
We've
talked
a
lot
about
AMT,
I'm
sure
you
know
this
audience.
I
don't
need
to
go
through
what
AMT
is,
but
the
idea
is.
It
goes
native
to
the
relay
and
then
from
the
relay.
It
gets
tunneled
over
unicast
over
UDP
to
via
AMT
to
end-users
and
I
just
saw
a
comment
but
hold
on
a
sec,
ter,
Jake
off
net
sort.
T
So
that's
off
receiving
often
that's
sourcing.
How
do
we
get
somebody
with
a
camera
phone
who
just
whips
out
their
phone
and
starts
hitting
record
and
they
want
to
stream
it
natively
onto
this
infrastructure
if
they're
stuck
on
a
unicast,
only
Network,
that's
where
list
comes
in
and
we
actually?
This
was
demonstrated
with
Dino
in
Singapore,
where
he
was
able
to
get
I
believe
it
was
on
his
iPhone.
He
was
able
to
from
the
from
the
unicast
only
Wi-Fi
network
in
Singapore
inject
pings,
pinging
traffic.
That
was
seen
on
multicast
routers
and
generate
state.
T
So
imagine
we're
what
we're
missing
the
piece
we're
missing
is
camera
video,
that's
what
we
want
rather
than
king
traffic,
so
getting
camera
video
we've
got
all
the
other
pieces
in
place,
so
you
can
camera
video
streaming.
Multicast
lists
would
get
it
to
a
a
TR
at
the
border
of
the
mbone,
which
then
can
get
it
natively
to
native
receivers,
as
well
as
often
at
to
via
amt2
off
net
receivers.
Q
T
Cool,
so
so
what
is
our?
What
is
our
plan
for
world
domination?
Five
steps?
First,
we
got
to
get
relays.
Second,
gateway
implementations.
Third,
a
portal:
how
do
you
you
know
solve
the
search
problem?
How
do
you
find
the
content?
Next
is
getting
be
off
net
sourcing
to
work
and
then
Step
five,
oh
shoot
what
is
step
5?
Yes,.
T
So,
let's
dig
through
them:
where
are
we
you
in
each
of
these
steps?
So
the
first,
a
MT
relay
was
deployed
at
Thomas
Jefferson,
High
School,
subsequently,
George
Washington
University
has
added
two
more
relays.
I
spoke
at
I
to
tech
summit.
Tech
exchange
in
December
and
in
several
I
to
institutions
were
interested
in
adding
relays
my
day.
T
Job
has
prevented
me
from
nagging
them,
but
I
will
be
reaching
out
to
them
and-
and
we
need
more
so,
if
you're
interested
in
if
you're
connected
to
I
too,
if
you
have
say
and
juniper
MX
router,
it's
four
lines
of
config
is
all
it
takes.
No
extra
hardware,
no
licenses,
no,
no
nothing,
and
if
there
are
other
vendors
who
also
make
AMT
relays,
please
speak
up
and
join
the
effort
and
we'd
love
to
you
know,
use
your
relays
as
well.
T
The
real
a
discovery
for
those
who
are
curious.
What
do
you
do
when
you
have
more
than
one
relay?
It
was
really
easy
when
there
was
only
one
relay
what
we're
doing
right
now
for
now
the
solution
is
dns-based.
I
think
this
is
the
Jeff
Houston,
the
Jeff
Houston
advice
F.
It
just
used
DNS
right,
just
put
it
in
DNS,
and
so
we
have
a
full
equality.
T
You
know
a
MT
relay
got
em
I
cast
multicast
Internet
cast
net
that
maps
and
we
have
each
of
these
three
relays
has
an
a
record,
so
we've
built
resilience
into
the
application
layer.
So,
in
the
case
of
VLC
that
we've
implemented,
VLC
gets
an
ordered
list,
receiver,
we'll
get
an
ordered
list
or
the
Gateway
we'll
get
an
ordered
list
of
all
the
relays,
and
it
just
picks
one
and
if
it
gets
no
packets
it'll
pick
the
next
one
and
it
just
goes
down
the
list.
So
we
have
relay
resilience.
T
T
Someday
when
there's
a
lot
of
relays
out
there
and
there's
a
lot
of
content
out
there
and
will
I
have
some
ideas
as
to
how
to
solve
this
problem,
but
if
others
do
have
some
other
ideas,
please
reach
out
to
me
and
email
me
and,
and
but
here's
a
hint,
it's
probably
you
know,
I
could
see
using
DNS
to
solve
this
problem,
Gateway
implementation.
So
the
goal
is
yes,
RFC,
eight,
seven,
seven
seven!
T
T
To
go
through
the
whole
thing,
unless
you
have
something
that
needs
some
clarity
within
a
specific
point
but
yeah.
Let
me
continue
on
because
I'm
almost
done
all
right,
so
our
goal
was
to
build
it's.
What
I
would
call
the
last
foot
problem,
and
this
is
what
Jake
is
really
solving.
The
last
foot
problem
is
to
get
it
from
your
screen
to
your
eyeballs
or
how
do
you
get
it
yeah
you
know,
and
that
is
we
wanted
to
build
a
gateway
implementation.
That's
easy
enough
for
any
user
to
use
the
short
term.
T
You
know
we
settled
on
am
on
VLC,
because
VLC
is
multicast
friendly,
it
is
available,
it's
multi-platform,
it's
open
source
and
Nataly
Landsberg
and
Wayne
Brazel
implemented
an
AMT
gateway
for
VLC
and
that
patch
and
they
were
able
to
get
an
upstream
successfully.
It
is
in
VLC
4.0
today,
so
you
can
go
to
the
website
and
download
it
and
you
will
get
VLC
AMT
capable
VLC.
Now
this
is
just
the
short
term
solution,
because
this
is
the
best
we
could
do.
What
we
really
want
and
I
would
encourage
everybody.
T
Please
go
check
this
out
and
try
it
out
in
the
long
term.
What
we
really
need
is
multicast
to
the
browser
and
so
Jake's
doing
the
Lord's
work.
So
we
you
know
this
is
what
we
really
need
to
solve
this
last
foot
problem
all
right,
so
the
content
portal,
this
the
most
recent
envelopment,
so
Lauren
Dell,
which
who
is
a
another
student
from
Thomas
Jefferson,
High
School.
She
is
building
off
of
William
Jiang
when
he
was
when
he
first
deployed
the
first
AMT
relay
he
combed
through
all
the
multicast
content
on
the
on
I.
T
He
basically
went
to
the
looking-glass
routers
and
I
too
looked
at
the
multicast
state
in
the
actual
routers
and
came
up
with
you
know:
here's
a
list
of
content
out
there
and
she's,
basically
building
off
of
that,
starting
with
that
and
now
populating
onto
a
nice
web
pretty
web
page
that
you
can
hopefully
launch
VLC
from
she's
made
some
progress.
This
is
I
would
say,
I
believe
Lauren
might
be
still
on,
but
she's
she's
made
a
lot
of
progress.
T
She
is
certainly
looking
for
help
she's
having
some
issues
trying
to
launch
an
application
out
of
a
browser
which
is
remarkably
difficult
for
browser
security
reasons.
So
anybody
who
knows
and
can
help
with
this
I'll
shout
this
out.
You
know
some
apps
can
do
this.
Zoom
can
web
Microsoft
teams
can
WebEx
canyou
can
launch
an
app
from
a
link,
we're
struggling
to
get
that
working.
N
T
N
Ring
in
parallel
wirelessly,
so
I
think
you
know
it
sounds
very
much
aligned
with
what
Jake
is
doing
and
so
I
think
would
be
good
to.
Maybe
you
know
on
all
the
sides
make
it
clearer
if
there
are
any
differences,
because
I'm
hoping
there
is
one,
you
know
story
going
forward
and
everybody
is,
you
know
I'm
working
on
his
pieces,
I
think
Jake
primarily
on
the
browser
side.
N
I
was
wondering
specifically
and
sorry
if
I
should
be
aware
of
that
by
everything.
If
the
approach
that,
for
example,
Verizon
presented
earlier
on,
which
I
think
wasn't
pure
AMT,
but
was
this
trick
off
doing
multicast
and
then
the
transport
layer
create
a
unique
a
stream
locally
through
an
agent
on,
so
that
you
could
use
an
unmodified
browser.
That
only
knows
unicast
right,
I.
Remember
that
right.
L
N
R
That
I
think
there's
like
mass
again
I
think
there
is
a
way
that
you
could
actually
get
it
to
work,
but
it
actually
doesn't
require
some
development,
but
with
web
RTC
that
you
could
actually
take
the
stream
and
add
somehow
you
can
actually
get
it
to
play
like
in
a
web
browser,
but
we
we
didn't.
Actually
we,
instead
of
by
you,
know
trying
to
develop
it
in-house.
We
ended
up
just
buying
a
product.
I
did
it,
but.
N
I
think
there's
a
way
that
you
can
yeah
I
think
there
are
a
bunch
of
different
ways
on
how
to
do
this.
Like
you
know,
unmodified
browser
with
this
agent
and
may
be
saying
yeah
here
is
the
you
know
open
source,
the
best
recommended
agent,
something
like
that
and
I
was
just
bringing
up
the
point.
It's
interesting
for
the
whole
strategy,
obviously
native
within
the
total
world,
but
yeah
would
rather
know
how
long
away
that
is.
If
there
is
your.
R
R
Are
you
familiar?
You
know
that
NPAPI
look
like
that
was
kind
of
like
legacy
like
that
nuts
key
/
yeah,
so
that
was
legacy
and
I.
Think
it
support
was
supported
by
older
browsers,
but
then
they've
removed
that
support
for
all
browsers.
So
then
it
would
did
make
it
difficult.
So
in
a
new
browser
now
you
can't
play
your.
N
And
and
with
a
plug-in
issue
right
now,
it's
either
get
new
native
code
into
browsers
or
an
agent,
and
that's
why?
Basically,
this
agent
part
was
something
where
I
was
wondering.
Maybe
you
folks
feel
we
should
go
because
you
haven't
fully
running
with
some
proprietary,
eight
and
I'm,
not
sure
what
Jake's
thoughts
are
there
so.
D
N
R
You
know
did
does
sound
them
almost
identical
like
taurus,
what
you're
saying
with
this
Ram
product
so
like
when
it
actually
sends
it
it
sends
it
encapsulated,
just
puts
it
in
a
UDP
payload
ships
it
out
across
the
trendy
and
then
we're
sleeper
gets
it
young
packages
it
and
then
he
just
plays
it
out
so
that
stream,
that
native
stream
is
like
an
HLS
stream.
So
now
he's
able
to
play
out
that
way,
and
he
just
plays
it
out
on
that
move
back
sounds
like
the
software.
T
T
N
R
Looked
at
it
a
little
bit,
you
know
when
we
look
at
it
in
the
beginning,
but
I
think
there's
some
messaging.
There
is
some
kind
of
translation
that
you
have
to
do
to
get
it
to
work
and
I
think
you
know
sorry,
but
it
for
just
looking
at
it.
When
we
did
look
at
a
high
level,
it
seemed
like
it
was
possible,
but
I
think
you
had
to
do
some
translation
because
the
multicast
you
know
way
before
that.
R
For
that
stream
to
be
read
by
the
browser,
you
actually
have
to
take
that
multicast
and
translated
unicast.
So
there
was
a
translation
and
I
think
that
mess
web
RTC
I
think
the
way
it
work
it's
kind
of
doing
like
messaging
between
the
client
and
the
server.
So
it
would
actually
do
like
some.
It
would
have
to
do
part
of
that
translation
and
just
for
multicast
to
unicast,
for
the
BLA
browser
to
play
out
unicast.
T
T
So
there's
different
approaches.
You
know
we,
there
was
the
idea
initially
of
a
server
we
looked
at
and
then
Lisp
seemed
to
be
a
way
to
do
it
and
we
were
able
to
demonstrate
it.
What
we,
what
we
really
need
is
a
find
a
way
to
get
there
the
smartphone
camera
to
be
able
to
stream
the
video
as
multicast
instead
of
unicast.
Some
progress
is
made
on
on
this.
We
had
a
developer,
who
was
working
on
this
and
he
had
made
progress
but
still
working
on
it.
Q
In
terms
of
requirements
here,
I
think
the
requirement
that
the
source
is
multicast
is
misguided.
Multicast
is
the
header,
that's
the
transport,
that's
how
it
goes
out.
All
we
care
about
is
the
content
and
getting
into
that
multicast
infrastructure,
and
it
doesn't
need
to
come
from
the
phone
into
the
multicast
infrastructure
because
in
most
cases,
you're
going
to
need
to
transcode
you're
going
to
need
to
send
it
in
multiple
stream
rates
for
various
receivers,
and
you
want
to
archive
that
stuff,
so
it
all
has
to
go
to
a
cloud
storage
anyway.
T
That's
certainly
one
approach,
you
know
we
got
it.
I
think
we
we
have
it.
You
know
with
Lisp
gives
us
the
ability
to
do
that
today,
because
you
know
it's:
it's
not
impossible
to
have
a
camera,
video
that
spit
it
out.
Multicast
packets,
it's
just
getting
them
somewhere
and
that's
where
Lisp
comes
in.
But
let
me
and
yes.
E
You
know
so,
yes,
you're
I
mean
I
agree
with
schepis
was
saying
and
I
also
agree
what
you're
saying
it
turns
out
that
when
Lisp
runs
on
the
source,
the
application
still
thinks
it's
sending
out
multicast
packets,
but
they
are
in
effect
being
unicast
at
somewhere
else,
and
they
may
be
unicast
into
that.
You
know
reformat
or
whatever.
That
chef
was
talking
about
yeah
that
transcoder
or
whatever
so
I
mean
it
kind
of
doesn't
matter.
E
Don't
I
think
if
we
just
find
if
we
create
the
multicast
service
model
for
the
product
and
receivers
can
join
and
leave,
it
doesn't
matter
how
the
data
is
sourced
right,
we
could
even
just
use
a
browser
or
something
and
the
packets
that
leave
the
application,
could
be
unicast
it
and
they
could
be
unicast
into
a
spot.
That
actually,
then
translates
the
destination
of
multicast.
If
we
are
really,
you
know
had
bent
on
packets,
have
to
the
destination,
address
must
be
in
a
two
twenty
four
slash
for
our
range
or
whatever
you
know
so.
E
T
E
Other
thing
is,
is
if
you
wanted
to
really
be
on
a
cell
phone,
and
while
people
are
walking
down
the
street,
you
have
to
deal
with
NAT,
traversal
and
usually
unicast
is
a
better
way
of
doing
that,
and
that's
why
Lisp
worked
well
in
this
test
that
we
did
because
the
they
have
traversal
functionality
for
multicast
packets
they
were
encapsulated
in
unicast
were
able
to
pierce
the
the
Pierce,
the
mats
on
outbound.
We
also
have
the
problem
on
inbound
as
well
right,
so
I
think
it's
important
ugly.
T
Even
if
the,
if
the
browser,
if
the
if
the
the
source
is
spitting
out
multicast
packets,
that
doesn't
help
you,
if
you're
not
on,
if
you're
on
a
unicast
only
network,
you
need
that
overlay
that
somehow
gets
it
there,
whether
it's
Lisp,
you
know
tunneling
it
to
the
multicast
network
or
that
it's
a
unicast
stream
to
the
multicast
network.
You
know
it's
as
as
a
plain
old
unicast
thing:
I
think
you
know
I'm
agnostic.
We
just
need
something
that
works,
agree,
agree,
and
you
know
if
there's
folks,
who
are
interested
in
helping
with
that
effort.
T
E
Need
more
work
there.
So
what's
sorry
Lily
real
quick,
what
we
need
if
people
could
find
a
resource
is
we
need
somebody
to
write
an
app
on
iOS
and
Android
that
takes
camera
output
and
sends
it
on
the
network
in
any
form.
Okay,
and
if
somebody
wants
to
write
that
application,
then
we
can
urge
them
to
send
both
unicast
UDP
or
multicast
UDP.
T
Q
E
The
camera
on
a
phone
I
couldn't
get
VLC
to
source
any
data
unless
it
was
a
video
file
that
was
cached
locally
and
usually
it
pulls
it
from
a
server.
It
was
kind
of
it
was
kind
of
backwards,
so
send
me
appointed
to
the
what
you
found
Lenny,
because
maybe
we
can
turn
turn
that
into
multicast
at
like
a
head
end
somewhere,
like
chef,
was
saying
sure,
yeah
and
in.
T
Fact,
I
I
may
have
found
a
tool
that
actually
does
that
and
it
may
be
open
source,
so
yep
awesome
all
right
cool.
So
here's
just
references
if
any
of
the
things
you
know.
If
you
want
to
take
a
look,
so
you
start
with
the
VLC
repo
here,
because
this
it
gives
you
a
pointer
to
where
you
can
download
the
the
correct
images
of
VLC,
and
it
also
gives
you
usage
documentation
how
to
actually
show
that
it
works.
T
But
this
is
something
that
everybody
can
download
right
now
and
start
wiping
content
on
the
ember
and
then
there's
you
know
other
other
past
projects
and
and
I
encourage.
You
also
take
a
look
at
Lauren's
multicast
menu.
We
do
have
a
slack
group
we'd
if
you're
interested
shoot
me
an
email
and
we'll
we'll
add
you
to
the
slack
group
and
with
that,
that's
everything.
I
had
questions
for
anybody
else
who.