►
From YouTube: IETF93-BIER-20150722-1300
Description
BIER meeting session at IETF93
2015/07/22 1300
A
B
That
that
still
doesn't
take
it
over
good.
Now
my
clock
says
straight
up,
so
I
think
we
should
roll
guys
up.
For
that
any
objections.
Anything
no
we're
waiting
for
in
particular,
I
can't
see
everyone's
face
directly
see
if
the
folks
were
looking
for.
There
was
a
late
shift,
the
agenda,
but
we'll
get
to
that
in
just
a
bit.
B
No
well,
you've
all
read,
understand,
agree
with
us
nods,
just
post
lunch.
It
is
a
snap
time.
That's
how
I
figured
nap
slots
always
fun
so
time
to
throw
rocks
the
agenda.
I
moved
peers
to
the
top,
because
he's
got
a
conflict
with
homenet
and
actually
I
saw
anything.
How
do
we
do
that?
Have
we
not
put
that
on
our
our
conflict
list,
but
we'll
add
it
next
time,
I
think
it's,
because
we
had
to
get
you
Jenna
together
before
you
even
submitted,
so
you
came
after
the
agenda
was
already
established.
B
A
B
C
B
B
C
C
Yeah
so
I'm
Jeff
Eastin
and
I'm
here
to
present
you
a
brand
new
draft
called
dear
ingress
with
you
gasp
overlay
using
Lucas
recent
discovery
protocols.
Next
slide,
please
so
coming
from
the
beer
architecture,
we
have
this
very
fancy.
Slicing
of
the
gear
stack
on
the
bottom.
We
have
the
routing
under
later
to
know
that
does
the
gear
very
clever
forwarding,
and
on
top
of
that
we
have
the
earlier
that
is
actually
supposed
to
insert
and
remove
the
beer
leads
inside
the
packets,
and
on
top
of
that
we
have
this
multicast
flow
overlay.
C
That
is
suppose
to
tell
the
BL
layer
what
is
the
year,
what
are
the
beer
that
should
actually
be
included
in
packets
that
enter
a
beer
domain
such
that
it
can
reach
the
right
egress
route
routers
when
moving
through
the
the
Bo
domain?
So
regarding
that
we
have
a
few
proposals
and
lots
of
things
are
going
on
for
the
underlay.
We
already
have
mpls
a
top
two
documents
for
the
beer.
C
We
have
nice
eyes,
extensions,
ospf
extensions,
but
try
now
right
now
for
the
multicast
overlay
it
still
little
bit,
I
mean
still
open
to
some
work
and
right
now
we
can't
really
implement
a
full
stack
of
beer
working
with
only
our
draft.
So
that's
where
this
proposal
comes
next
next
slide.
Please
and
this
proposal
doesn't
intend
to
be
the
only
solution
ever
as
a
flow
flow
overly
from
here.
Let's
not
try
to
oil,
the
ocean,
the
scope
of
this
architecture.
C
In
my
mind,
if
this
proposal
is
for
small
to
average
average
sized
network-
maybe
bigger
I,
don't
know
I
mean
I,
can't
guess
the
scalability
of
the
proposal
and
it's
a
distributed
proposal,
it's
a
distributed
design.
Maybe
some
people
will
want
more
centralized
approaches
such
that
they
can
have
more
control
on
what
is
going
on.
So
it's
perfectly
fair
and
I.
Think
it's
in
scope.
Do
you
working
with
multiple
solutions,
yeah.
B
C
I'm,
a
homeless
guy
I
come
from
home,
meds
I.
Imagine
these
very
simple
scheme
could
be
applied
to
home
that
it's
not
on
today's
agenda.
It's
not
I
mean
it's
only
in
mind.
I'm
happy
to
contribute
to
here
without
even
thinking
about
your
net
I
mean
I
think
this
could
be
a
solution
in
general,
but
I
think
there
is
really.
C
B
C
A
B
C
So
I
guess
everyone
in
the
room
knows
how
penalty
works.
Obviously,
it's
based
on
the
ability
of
nodes
on
the
land
to
be
able
to
send
nld
packets
to
the
Warriors
and
from
the
quarians
to
be
able
to
send
those
queries
to
the
listeners
and
the
barriers
in
that
process.
We
have
an
election
that
is
taking
place
which
elects
among
the
courier's
one
electric
career,
which
is
actually
sending
the
queries
know.
The
proposal
of
this
draft
is
very,
very
simple.
C
C
C
C
So
we
need
the
listeners
to
send
multicast
packet
to
the
warriors.
To
do
so,
we
need
a
multicast
address.
We
also
need
a
multicast
address
to
let
the
courier
sense
which
gas
pockets
to
both
listeners
and
queers
again
across
the
Bo
domain,
and
we
also
need
in
order
to
be
able
to
send
this
multicast
packet.
We
need
the
BF,
our
IDs
of
the
barriers
and
the
listeners.
C
C
When
you
send
a
query,
you
are
going
to
put
as
the
destination
of
the
query
the
multicast
address,
which
has
been
configured
such
that
the
packet
is
received
by
all
the
listeners
end
up
areas
when
you
send
a
report,
it's
the
same
thing,
but
you
put
a
different
destination
address
as
a
source
address.
It
changes
from
LD
because
MLD
you
only
use
the
link
local
address.
C
In
that
case,
we
are
going
to
put
your
VFR
prefix
address,
which
you
already
have,
because
you
are
an
egress
or
an
ingress,
so
the
beer
layer
is
configured
with
such
an
address
and
you
use
that
address
as
the
source
address
of
your
MLD
or
igmp
pockets
such
that
when
the
destination
reaches
the
destination,
the
vertical
destinations
receive
the
report
of
the
query
they
can
know
where
it
comes
from.
We
don't
need
the
hop
by
hop
early
adoption,
because
it's
a
I
mean
it's.
It's
not
useful
anymore,
because
we
use
once
again
multicast
address.
C
And
finally,
we
have
to
enable
the
feature
of
ige,
mp3
and
MLG
v2,
which
is
the
explicit
tracking
such
that
you
can
associate
with
every
half-hour
prefix
address
the
membership.
So
next
slide
please
so
the
first
mapping
is
the
mapping
on
the
left.
You
have
using
MLD
you
Mac,
the
membership
is
the
VFR
prefix.
C
One
final
thing
that
I
didn't
said
is
that
we
I'm
proposing
that
we
don't
try
even
to
support
MLV,
v1
or
I
j.mp/d
too,
because
I
mean
it's
a
completely
new
space.
We
don't
need
to
be
backward
compatible
with
anything
next
slide.
Please
we
have
to
open
issues
right
now,
based
on
discussions
on
the
mailing
list.
One
is
that.
C
You
have
naughty
actually
going
to
forward
that
to
them
to
the
destinations
through
the
gear
domain.
In
order
to
do
so,
I
think
we
could
do
it
with
no
new
messaging
requiring
the
ingress
routers
to
actually
listen
to
the
traffic
within
this
domain,
but
it's
a
little
bit
hacky
I
mean
we
could
make
it
work.
Maybe
may
not
be
a
perfect
solution,
so
it
shows
a
sort
of
that.
We
need
some
election
mechanism
yet.
D
C
Another
mechanism
to
elect
designating
designated
something,
so
maybe
that
also
shows
us
that
we
need
it
to
be.
That's
that's
possibility,
maybe
protecting
MLD
such
that
we
can
reduce,
reuse
it
in
order
in
other
situations.
The
second
output
open
issue
happens
at
the
exit
of
the
domain,
so
the
first
one
was
the
packet
enters
know
at
the
exit.
We
have
a
kind
of
similar
problem
how
to
make
sure
that
the
packet
is
not
sent
twice
throughout
the
same
destination.
C
B
Be
I
I'd
argue
that
the
problems
are
different
and
that
we
said
yet
another
election
mechanism
there's
some
case.
We
already
have
election
meccans.
We
could
potentially
use
this
been
discussion
taking
place
because
these
issues
came
up
on
the
mailing
list
in
it,
as
kind
of
we
had
some
way
to
brainstorm.
A
bit
of
that
today
and
I
wish
you
were
here
for,
but
you
got
to
go
but
I
think
you're
right
on
with
the
downstream
domain.
We
already
have
election
mechanisms
there.
B
We
can
probably
hijacked
in
some
way,
but
in
terms
of
into
the
network
like
what
was
put
out
if
it
was
a
BGP.
That
policy
can
elect
that.
We
lack
that
here.
So
here's
a
case
where
we
may
have
to
do
something
different,
but
I,
think
the
I
think
into
the
network
and
out
of
the
network
are
two
different
cases
and
the
topology
out
of
the
network
and
the
topology
in
the
network
actually
can
be
different
and
changed
the
case
as
well.
Ok
and.
B
A
Roberta
mañana
Cisco
I
just
have
an
eye
level
calm
and,
as
you
said,
the
document
is
applicable
to
both
MLD
an
igmp
v3
for
clarity.
In
my
opinion,
it
would
be
better
to
reflect
that
in
the
tight
or
just
to
make
clear
that
is
applicable
to
both
ipv6
and
ipv4
I
know
you
come
from
neta
that
exactly
six
focus.
E
A
F
F
May
be
for
clarity
if
you
can
explain
whether
you
want
to
totally
replace
MLD
or
you
still
have
a
hosts
connected
thermal
d.
So
your
BM
LD
listeners
do
that.
Do
that
trans
protocol,
translation
between
MLD
MLD
and
B
MLD?
And
if
we
look
at
this
the
way
I
look
at,
is
you
extending
the
deal
multicast
domain
towards
the
access
side?
F
Now
you
push
this
down
and
it
really
replace
this
line
with
the
beer.
You
really,
the
question
is:
will
your
host
will
will
your
host
and
hosted
an
igmp
and
we
need
to
translate
those
messages
forward
them
for
the
beer
and
do
the
leader
election
or
were
we
talking
about
no
igmp
normally
at
all,
and
everything
is
built
like
I
think
there
can
be
slight
differences?
How.
C
F
Unraveling
of
him
probably
would
have
the
igmp
animal
d
coming
and
we're
talking
about
taking
those
moving
them
into
the
beer
messages
sending
to
the
DMV
MLD
wearers
round
up
some
lightweight
protocols
and
back
that
to
me
would
make
sense-
and
that's
probably
the
way,
and
that
would
also
answer
your
question
on
a
slide
below,
and
we
kind
of
touched
on
this
with
with
ice
it
on
one
of
your
last
slides
weather.
Where
should
we
do
this
work
shouldn't
be
extensions?
F
I
would
argue
no,
so
it
should
be
in
beer
as
a
and
not
extensions
by
melody
people
don't
just
say
how
we're
going
to
treat
their
messages
and
do
and
do
you
know
probably
still
half
of
the
extension
protocols,
but
it
may
be
easier
because
you
don't
want
to
enable
MLD
or
anything
clothes
on
the
anywhere
in
the
gear
domain
like
I.
Don't
want
to
have
any
of
that.
F
F
That
would
be
just
silly,
so
that's
why
I
say
I
would
probably
you
know,
borrow
as
much
as
we
can
to
say:
borrow
steal
as
much
formal.
The
agent
p
can't
get
the
common
election
in
here.
Do
it
cleaning
beer
keep
em
al
di
jump
you
you
know
either
either
at
the
add
some
hosts
or
potentially
not
even
have
it
at
all.
You
can.
You
can
imagine
how's
that
are
sending
the
message
right
away
at
some
point.
C
B
Pierre,
so
we
just
kind
of
Rapa.
Somebody
starts
if
she's
going
to
keep
coming
up
today,
but
the
duel
home
nature
in
and
out
our
issues
that
have
been
kind
of
cropped
up
with
this
discussion
and
I
think
we
need
to
focus
that
on
the
list
or
either
with
the
design
team,
because
I'm
agree
completely.
Why
rebuild
it?
Why
have
a
protocol
dependency?
You
know
that
this
problem
exists.
B
G
Okay,
so
I'm
going
to
give
me
an
update
on
the
architecture,
and
the
mpls
document
should
be
quite
quick
now
there
are
a
few
changes
to
the
mpls
encapsulation
document,
as
you
may
have
seen
so
remove
the
few
fields
around.
We
added
the
only
one
and
we
gave
OEM
a
few
bits
to
play
with
so
I.
Don't
think
nothing
Lee,
drastic,
nothing
rly
that
changes
the
architecture
in
that
sense,
so,
yes,
and
of
course,
on
sometimes
patients
and
directions
to
the
architecture.
G
Draft
next
select
now,
first
of
all,
a
few
modifications
or
translations
to
what
a
beer
domain
is
right.
You
know
we
consider
the
bear
domain
to
be
where
your
bear
protocol
messages,
your
updates,
and
that
is
basically
confined
to
the
igp
flooding
scope.
So
it
basically
follows
your
HP
flooding
school.
So
we
made
that
more
more
clear
in
the
document.
Also.
G
When
you,
when
you
create
that
flooding
scope
with
your
rabbits
in
the
middle,
if
you
go
from
one
VFR
to
another,
it's
possible
that
you're
actually
path
actually
leads
you
to
a
router
which
is
not
BF
are
capable
right,
because
the
underlay
is
really
your
IGP
on
the
lake,
so
it
doesn't
really
have
the
information
call.
You
know.
You
know
that
the
next
node
is
not
be
for
a
capable,
but
it
does
no
proceed.
Is
to
go
around
it.
Your
self
start
right.
So
you
need
to
know
that
you
know
it.
G
It
will
use
all
the
routers
or
other.
You
know
in
an
HP
domain,
whether
they
not
capable
enough
right.
So
you
may
get
stuck.
That's
an
old
speak
about
that
more
later
on
it
also
regarding
subdomain.
So
it's
the
BFI
are
that
determines
which
subdomain
you're
gonna
store
it
a
packet
e
right
now.
Subdomain
is
strictly
ties
to
the
underlay
to
the
igp
that
you
look
upset
so
you're
not
allowed
to
change
subdomain
in
flight
right.
G
G
So
also
what
I
said
is
the
the
sub
domain
is
strictly
type
to
the
to
the
igp.
Alright.
So
normally
we
just
use
the
IGP
that
is
there
for
unicast.
If
you,
you
look
as
point
you
to
a
router,
it's
not
BF
are
capable.
There
are
certain
procedures
you
can
do,
but
simply
too
serious
note
you
can
do
anything.
You
cannot
send
a
packet,
so
you're
stuck
so
you
drop
the
packet
or
you
can
try
to
hop
over
it
right.
G
You
can
tunnel
through
it
and
and
well
we
don't
know
we
don't
really
want
to
do
a
lot
around
rerunning
around
it,
because
if
you
REBOUND
around
it
leave
another
igp
topology
need
to
do
that
right
or
specific
SPF
calculations
to
find
another
path
to
go
around
it.
We
totally
attempt
to
do
that.
So
the
best
thing
is
to
sort
of
tunnel
through
relief.
If
you
can
that
also,
it
may
actually
hit
a
router
which
is
difficult.
Maybe
it
didn't
advertise
label
for
the
rights
of
the
main
light
was
the
right
history.
G
G
Also,
to
make
it
more
clear
in
the
architecture
we
refined,
Inquisition
and
disposition
bits
in
length,
that
imposition
is
a
pretty
sim
length
that
the
English
router
puts
on
to
the
Packers.
This
position
length
is
what
you
used
to
forward
packets
right.
So,
although
the
other
routers
in
the
bath
vehicles.
G
They
are
it's
up
to
the
implementation
right,
but
you
do
like
to
support
two
different
missing
legs
at
the
same
time
for
migration
purposes,
but
so,
if
you
want
to
move
from
256
to
512
later
on,
it's
very
useful
that
you
can
support
512
your
network
side
by
side
next
to
256
and
none
on
the
ingress
you
cut
over
once
you
determine
that
your
next
able
to
further
back
alright
so
for
migration
purchase.
You
like
to
support
two
different
bit
string
lengths
at
the
same
time
for
this
position.
G
If
you
know
the
shortest
path,
leads
you
to
that
route,
but
it's
also
possible
that
the
drug
wasn't
configured
for
for
a
beer
at
all.
Right
again,
you
cannot
detect
that
now.
So
what
do
you
do
if
you
dare
detect
an
inconsistency?
Well,
we
think
the
best
thing
to
do
is
just
login
error
message
that
you
know
there's
a
potential
inconsistency,
but
you
still
want
to
send
a
packet.
G
It
might
be
because
the
packet
can
still
go
to
places
read
minutes
ago
and
the
minute
we
can
hit
that
router,
which
has
two
different
bits
in
length
at
all
right.
So
it's
still
you
want
to
do
a
best-effort
basis
could
live
the
packet,
even
though
you
could
see
that
there's
a
different
bits
in
length
supported
in
that
subdomain.
G
That
next
next
time,
so
as
I
said
so,
if
you
encounter
an
inconsistency,
you
can
drop
the
packet
length
or
tunnel
through
the
router
using
procedures
similar
to
if
the
router
was
in
traffic
it
for
VFR
route
or
for
birla.
So
we
we
don't
really
want
to
do
any
heroic
efforts
to
be
around
around
routers.
That
happening
is
configured
to
miss
provisions
I,
because
that
would
require
us
to
change
the
HP
to
underlay
new
special
FPS
calculations.
G
Roboti,
this
and
I
think
you
can
just
consider
it
as
a
Miss
configuration
like
if
you
we
have
experience
from
that
in
the
past
as
well
like
if
you
have
a
beam
router
which
is
on
shortest
path
to
the
source,
but
that
router
was
not
going
to
leap
in
they
don't
get
there
right
or
if
you're
mpls
interfaces,
not
neighbor
friend,
pls
you're
stuck
right
so
with
lot
of
Orientals
in
the
works
also
to
do
the
deckers
misconfigurations.
So
we
don't
want
to
do
anything
really
to
three
rounds
around
misconfigurations.
D
A
H
Bf
yards
and
like
mentioned
the
previous
presentation
as
well,
that
beard
carries
traffic
within
only
one
igp
domain,
but
a
meeting
also
supports
details
that
are
segmented.
So,
if
you
want
to
go
use
segment,
two
towns
you
need
to
have
the
abs
are
sore
ABR
support
or
act
as
BF
I
arms.
So
you
could
have
each
town
do
beer.
H
H
We
have
two
mechanisms
to
identify
the
exact
set
of
nodes
that
we
need
to
follow.
The
packet
clip
one
is
based
on
the
leaf
info
required
flag,
that's
defined,
and
the
other
mechanism
that
we
have
added
to
the
draft
is
using
the
method
if
thats
infused
in
extracting
using
the
leaf
info
required
per
flow
plan,
I'll
explain
more
as
we
go
forward
next
slide.
H
Also
on
the
EMG
is
the
MPLS
upstream
label
we've
added
text
to
see
how
it'd
be
useful
in
the
segmented
case
as
well,
any
unique
MPLS
label
is
assigned
to
each
see
flow
and
this
is
used
by
the
be
fer
to
be
to
identify
the
sea
float
which
the
packet
belongs,
and
this
is
useful
when
you're
forwarding
across
segments
x-lite,
please
so
for
expressive
tracking
to
identify
the
set
of
egress.
Please
there
are
two
choices
available.
A
H
Blasted
back
Sheila
spiced
now
for
multiple
floors,
you'll
have
multiple
s.
Penzeys
what'll,
RPF
de
does
is
reduce
the
scale
of
eskom's
ease.
It
does
by
sending
only
a
wild
card
as
flimsy
the
wild
card
has
lindsay,
will
have
the
Li
RPF
flag,
set
the
BF
ers
on
receiving
the
wild
card.
S
tansy
will
respond
with
one
or
more
leaf
ferries
that
correspond
to
that.
So
you
it
doesn't
need
an
LA
are
to
be
set
on.
H
Each
see
flow,
seafloor,
suspends
e,
so
you
just
send
one
wild
card
ask
lindsay
and
the
BF
eos
response
at
all
this
one
or
morally
fading.
Now.
One
restriction
to
this
mechanism
is
that
it
cannot
be
used
for
segmented
towels.
The
reason
is
that
for
four
segments
two
towers
we
may
have
to
follow
the
packet
to
the
next
segment
and
to
determine
the
town
to
which
we
follow
the
packet
on.
H
We
need
to
know
the
label
or
we
need
to
know
other
context,
and
if
we
are
using
the
explicit
tracking
mechanism
using
the
lrp,
a
flat
then
get
only
sending
one
while
Costanzi
and
there
is
no
unique
labels
for
each
flow
in
the
network,
so
you
cannot
really
identify
which
the
outgoing
thongs
so
the
restriction
currently
is
that
you
cannot
use
it
for
segmented
details
excite
me.
So
this
is
pretty
much
what
I
explained.
H
The
big
advantage
of
using
the
lrp
a
flag
is
a
significant
reduction
in
the
S
Pen
z80
routes
needed
you
right
need
only
one
as
star
star
S,
Pen
CID
out.
This
also
means
that
you
don't
need
you
don't
really
need
any
eye.
Pins
ease
in
the
network
using
a
star
surest
NZ
similar
to
a
partition
model,
is
sufficient
for
beard
it's
like,
so
we
also
had
a
text
to
modify
and
capsulation.
H
You
this
Islamic
educational,
bakit
type
in
the
area,
that's
a
match
for
transmission
and
to
create
the
beer
header.
So
if
it,
the
SMG
ad,
has
an
ally.
Oflag
set
the
Lees
Ferry
routes
are
the
ones
which
identify
their
skis
and
the
Lees
Ferry
routes
have
the
route
key,
which
is
identical
to
the
nlra
of
the
S
Pen
zeroed
out.
If
we
use
the
new
mechanism
of
La
RPF
flag,
then
the
leaf
ad
matching
routes
are
not
identical
to
the
NLRA
of
the
SMC.
H
They
are
derived
from
the
NLRA
that
is
mentioned
in
section
5.2,
but
basically
how
it
works
is
we
modify
the
route
distinguisher
when
we
respond
to
the
leaf
a
DS,
so
the
route
distinguisher
is
derived
from
the
S
Pen
Z
and
we
add
some
values
to
the
distinguish.
Indistinct
distinguish
the
fact
that
these
leaf
a
DS
are
in
response
to
their
lair
PF
flag,
and
you
can
use
that
to
identify
the
usps
that
we
need
to
send
the
packet
excite.
Please
again
for
D
capsulation.
H
When
the
beer
packet
reaches
the
egress,
be
using
small
typo
length,
the
it
should
be.
The
talk
label,
the
top
MPLS
label,
to
tell
you
that
you
are
in
ESP
and
then
the
u.s.
PE
kit
does
the
D
capsulation
by
delivering
the
bfr
prefix
and
the
payload.
The
top
label
is
the
upstream
assign
label
in
the
payload.
Now
the
bfr
prefix
is
the
context
in
which
the
MPLS
of
the
upstream
assign
label
is
interpreted
and
the
VFR
prefix
corresponds
to
the
bri
ID
and
the
subdomain
in
the
header
next
slide.
B
I
I
D
E
D
D
I
In
contention,
that's
the
part
of
beer
attribute
must
support
a
policy
to
enable
or
disable
the
creation
of
the
beer
attribute
and
the
attachment
to
specific
EGP
roads.
A
implantation
make
disabled
Oh
disable.
The
Christian
appear
attribute
pensively
a
unless
it
configure
a
BP
speaker
must
only
attend
the
local,
a
creek
kayden.
They
are
tribute
to
a
BP
update
message
on
which
at
least
one
the
pfr
practice
it's
contained
in
the
AR
I
feel.
I
Restrictions
on
sending
and
receiving
fear
information
implication.
Let's
support
beer
attribute,
must
support
a
/
EPP
session
policy
that
indicate
whether
attribute
is
labeled
all
disabled.
For
you
on
that
the
beer
attribute
must
not
be
sent
on
any
EVPs,
for
which
the
session
policy
is
not
config-if.
I
A
beer
attribute
is
saved
on
a
PDP
session,
on
which
the
session
policy
note
can
fake.
The
information
must
be
treated
at
a
it's.
What
are
recognized
launches
chance
to
contribute
to
prevent
the
pair
attribute
from
leaking
out
of
the
beer
domain?
Each
PP
router
on
the
pier
domain
must
part
of
outbound
wrote
announcement
policy.
Such
a
policy
must
be
disabled
by
default.
I
Deployment
consideration
assume
that
the
beer
domain
is
a
mini
treaty
domain
which
are
composed
of
multiple
guesses
user
burn,
pier
domain
in
others
narrow
in
outsides
of
these
documents.
Since
the
beer
attribute
is
optional,
transitive
peso
tribute
now
pfr
PP,
speaker,
crews
do
otherwise
receive
them,
wrote
with
beer
a
tribute
to
continue.
I
This
is
beneficial
for
the
incremental
deployment
appear,
wear
PPE
speaker
could
tunnel
appear
package
to
the
EF.
Ye
are
directly
if
the
pp
next
hub
is
a
noun
d
upon
a
BP
speaker
is
allowed
to
tunnel
a
beer
packet
to
the
BP.
Next
hope
it
is
true.
Pp
speakers
are
not
directly
connected,
such
as
in
them.
G
B
So,
in
light
of
that
information,
without
having
read
the
draft,
what's
the
sense
of
the
room
of
adoption
I
all
right,
please
read
it.
There
will
be
a
call
to
the
list
for
adoption
with
a
two
week
window
and
it's
not
just
what
you
don't
like
about
it.
It's
accepted.
So
let's
hear
what
you
like,
if
you
don't
like,
just
get
some
discussion
there.
Ladies
going
to
look
to
see
this
actually
some
turn
the
list
that
people
have
chimed
in
right.
We're
not
just
looking
for
knows
so.
Yes,
those
are
just
as
important.
Thank.
C
A
B
B
J
J
J
Yes,
so
if
you
have
a
lot
of
beer,
you
need
more
pet
freedom
being
just
on
the
shortest
path.
Is
not
good
enough.
Next
start
reminder:
what
was
this
about
so
in
beer
te,
the
bit
positions
are
not
only
the
leaves
that
need
to
receive
the
traffic,
but
also
a
Jason
sees
through
which
we
want
to
have
the
three
go,
so
we
basically
populate
in
the
forwarding
table
of
each
Beauty
note
only
those
adjacencies
that
we
want
to
follow
it
all
on
that
one,
and
that
gives
really
a
very
simple
forwarding
engine.
J
J
Jason
sees
and
I'm
getting
to
the
relationship
other
technologies
later
on,
so
the
real
trick,
and
that's
where
some
of
the
complexity
and
beauty
comes
from-
is
all
the
extensions
of
giving
different
type
of
adjacencies
to
the
big
position
so
that
we
can
actually
get
away
with
as
few
big
positions
as
possible
because
they're
limited
property
with
maybe
2056
being
what
we
would
like
to
target
without
a
size.
Next.
K
A
J
Thank
you
very
much
and
yeah
there
are
you
going
back
to
the
work,
then,
with
his
input,
so
the
Z
reserved
person
read
a
little
bit
like
there
is
only
controllers,
the
universe
and
nothing
about
CLI.
So
what
was
really
meant
to
be
implied?
Is
that
and
I
think
that's
also
what
we're
doing
in
our
photo
tightly.
The
actual
configuration
of
beer
to
eat
on
the
individual
devices
is
through
CLI,
whether
or
not
we
need
to
standardize.
J
There
is
an
open
question
and
then
basically
a
controller
would
go
through
the
CLI
first
and
once
we
basically
have
enough
experience
with
that,
we
can
think
about
the
yang
model
to
basically
make
it
more
ITF
process
compliant
and
for
networks
that
are
mission-specific
limited
scale
it
with
its
eyes,
like
I,
can
even
see
some
IPTV
deployments,
maybe
menu
a
configuration
would
be
good
enough.
Where
so,
that's
experience
to
be
collected
through
experimentation.
J
So
then
I
got
the
feedback,
they
said
identify
what
about
it
said,
identify
you
didn't,
mention
it
and
then
just
fell
over
in
0,
because
it's
not
the
problem
of
using
it
and
the
heck.
Yes,
okay
with
the
set
identifier
in
the
same
way
as
in
here
you
basically
per
set
identifier.
You
have
a
different
your
teeth
or
would
enable
with
the
different
semantics
of
the
bits.
So
pretty
much
the
same
thing.
J
The
trick
is
rather
how
I'm,
here
to
e
to
avoid
that
you
need
in
every
set
identifier
reallocate
a
bit
for
intermediate
mas,
so
that's
kind
of
the
intelligent
logic
of
trying
to
figure
out.
Oh
I
have
one
set
identifier
for
let's
say
this
part
of
the
network
and
that's
basically,
where
allocate
bids
for
the
interviewee
hops
that
I
go
through
and
then
I
have
another
set
of
a
higher
when
I'm
doing
replication
into
another
part
of
the
network,
and
there
is
certainly
a
lot
of
future
text
on
you
know
possible.
G
J
It
could
be,
the
header
itself
could
be
in
the
encapsulation,
and
it
seems
that
the
most
easy
starting
point
is
with
the
MPLS
encapsulation
we're
using
a
dedicated
label
to
identify
the
ear
packet
and
we
would
use
another
unique
label
value
to
identify
the
RTP
packet
and
I
think
we
would
need
to
check
which
other
encapsulations
are
relevant.
If
you
worked
on
and
figure
out
what
distinction
or
the
detective
litre
bottle,
name
prepare
as
well.
A
J
Entire
thing
was
ecmp
right:
why
do
we
have
a
separate
ecmp
make
it
isn't
in
here
to
eat
over
the
mechanism
in
fear?
And
so
it's
hard
to
put
in
into
all
the
details
of
a
2
to
the
best
of
my
understanding
is
I,
couldn't
figure
out
how
to
make
the
EC
NP
in
fear
work
for
VTE
because
of
the
right
way
it
is
tied
into
the
beer
of
igg
integration?
Maybe
that's
just
a
short
coming
on
my
side
right.
J
Selection
for
pakistan,
the
flow
having
the
same
entropy
but
then
be
able
by
setting
up
explicit
a
list
of
the
alternative
adjacencies
in
the
GE
forwarding
table
and
being
able
to
do
that
differently
on
multiple
hops,
to
very
explicitly
engineer
the
ecmp
choices
that
are
made
up
by
hop
for
traffic
on,
and
so
I
think
that
is
going
beyond
the
flexibility
that
I
would
have
in
their
viewer.
But
any
feedback-
and
you
know
review
on
that
section.
More
than
welcome,
but
now
that
was
more
a
high-level
question.
L
J
G
Yeah
I
think
what
you
mean
is
how
to
identify
key
traffic
from
normal
traffic
and
ideas
to
add
associated,
been
a
different
beer
label
right.
So
we
allocate
a
different
label
and
advertise
it
around
tensely.
So
it's
not
a
globally
unique
or
the
domain
by
unique
label
for
traffic
engineering,
but
you
can
do
the
same
as
you
do
today
for
shortest
path.
You
just
announce
a
different
label
that
you
announcing
the
igp
and
Sega's
just
traffic
engineering.
J
Because
we
had,
you
know
much
more
easily
and
we're
getting
to
that
inside
further
down
to
get
across
non-beer
capable
Hawks
so
and
if
the
label
is
something
we
need
to
send
through,
the
igp
will
need
to
figure
out
if
we're
reducing
the
incremental
adoption
options
with
this
approach.
Hitter
next
up
all
right-
and
this
is
basically
really
where
this
guess
too
so-
I'm.
J
Your
statement,
routing
so
technology
and
I
hope
I'm
politically
correct
here
with
this
terminology,
otherwise
come
and
get
me
so
the
way
I
always
understand.
Second
growling,
it's
a
live
web
rotational,
RSVP,
I,
and
so
then
the
question
is:
what
would
be
the
you
know:
multicast
equivalent
for
segment
routing
today
is
the
unique
past
technology
and
we
actually
embarked
on
trying
to
find
the
best
answer,
and
you
know
we
hope
your
ge
is
the
best
ender
there,
because
in
the
end,
the
you
know,
fundamental
concept
on
which
we
are
trying
to
do
here.
J
So
obviously
replication
not
being
the
subject
in
unicast.
That's
a
novel
part,
and
so
you
can
see
easily
how
beer
te
applies
to.
You
know
that
they
specifically
IPTV
bigger
contribution
network
examples
which,
in
the
past,
have
been
very
strong
on
suggesting
they're,
usually
rsvp-te
as
the
technology
to
support
their
traffic
engineering
needs
like
for
once.
They
are
options
for
recovery.
J
So
the
error
are,
there
is
beauty
specific
at
this
point
in
time
is
very
powerful,
but
also
to
a
good
degree
complex,
and
we
just
edit
text
to
say
that,
because
of
the
fact
that,
through
the
saying
router
Jason
sees
that
give
us
the
you
know
similar
functionality
to
segment
routing
I'd,
like
you
know,
steer
through
some
few
hops
in
the
network
and
everything
in
the
middle.
They
don't
care
about.
The
same
adjacencies
also
give
us
the
ability
to
just
leverage
normal
mpls
all
right.
J
So
if
we
do
an
MPLS
encapsulation
for
beer
to
e,
then
a
rounded
adjacency
could
simply
be
an
MPLS,
protected
adjacency
and
we're
with
normal
mpls
mechanism.
You
have
set
up
Lee
protection
tunnel
and
hawala
br2
II
can
use
those
adjacencies
and
I
think
that's
pretty
much
the
end
of
the
update
for
version
01
each
other
now.
So,
basically,
just
a
few
things
that
were
on
the
mind
for
for
the
next
rounds,
but
please
search
I
mean
on
the
list
on
anything
else.
B
B
So
all
right
we're
going
to
get
since
the
room.
Who's
read
the
draft
okay,
who
thinks
this
should
be
a
doctors,
are
working
group
item
who
feels
this
is
not
in
charter
love.
It
all
right,
looks
good,
we'll,
take
it
to
list
same
game
two
weeks
and
then
give
your
ya
na
or
yay
or
nay,
but
either
way
thanks.
L
L
So
that's
why
you'll
find
in
the
list
a
lot
of
stress,
unicast,
OEM
capabilities
in
as
continuity
check
as
well
as
performance
measurement,
but
at
the
same
time
there
could
be
implications
that
are
interested
in
x
direction
of
OEM,
and
we
believe
that
this
tool
set
tools
has
to
be
developed
as
well,
and
so
thus
we're
marking
these
requirements
as
a
must.
Whether
somebody
will
choose
to
use
them
or
not.
That
will
be
their
decision.
L
L
B
L
B
Come
to
the
conclusion,
not
understanding
why
we've
done
we've
done.
We
have
to
go
back
through
that
again,
so
that's
a
good
point
having
that
documented
somewhere
could
help
so
who's
read
this
draft
all
right,
all
right.
Those
who
read
I
think
this
is
something
worth
adopting.
Yeah
clear.
This
is
a
process
doc.
This
is
an
important
part
of
moving
stuff
forward.
Excellent.
Thank
you.
Okay,.
L
L
You
can
definitely
find
a
lot
of
parallels
with
functionality
with
a
mpls
LSP
paint,
but
because
mpls
LSP
thing
is
specific
to
mpls
data
plane.
So
we
decided
to
decouple
the
functionality
and
naked
more
is
a
control
protocol
that
can
be
put
into
any
encapsulation
in
this
case
into
beer
encapsulation.
So
what
was
updated
are
we
specified
the
version?
We
specify
the
message
types
I
will
specify
what
the
protocol
meeting
is
next
slide.
Please,
and
then
we
clarified
some
procedures
and
when
they're
transmitting
require
receiving
reply
and
transmitting.
E
L
M
Basically,
a
bfr
could
get
a
TTL
exploration
for
a
beer
ping
packet
and
then
it
will
send
a
response
back
to
the
BFI
are
and
that's
how
Chris
router
works.
But
there
is
one
issue:
next
pink
drink,
the
beer
package:
it
can
be
tunneled
across
a
nose,
for
example,
some
intermediate
routers
cannot
support
via
year.
In
that
case,
especially
if
the
econo
uses
a
uniform
low,
40,
gr
or
processing,
then
we
will
have
an
issue
because
the
tunnel
ingress
will
copy
the
TTL
from
the
payload
and
decrease
it
at
every.
M
So
the
bigger
the
dr
exploration
could
happen
on
that
tunnel.
Transitive
router,
which
does
not
support
here
at
that
time.
The
kind
of
behavior
is
that
ngos
I
simpiy
processing
was
simply
drop
that
man's
message.
There
is
no
indication
law,
it's
like,
so
the
proposal
is
that
some
can
creek
park
in
oak
creek
yeah.
So
they
sit.
M
M
M
M
Far
will
hit
the
collect,
some
deer,
specific
information
and
also
the
panel
hobby
information
on
where
the
excursion
happens
and
then
send
them.
Jenna's
send
a
bed
here,
ping
echo
reply
message
back
to
the
BFI
are,
that
is
to
be
document
separately,
for
example,
in
that
beer,
team
and
draft.
This
paragraph
is
only
about
how
the
mps
processing
works
next
slide.
M
M
L
M
In
this
diagram,
PFR
one
who
so
that
we
are
packaged
into
in
Tibetan,
oh
so
we
need
put
the
tag
into
the
tunnel,
it
copies
the
GTL
from
a
beer
payloads
and
put
it
into
the
label
for
the
tunnel
itself
and
then,
when
the
when
the
packet
gets
through
a
noun
VFR
one.
At
that
time,
chea
could
expire
because.
K
M
A
K
M
L
M
M
L
M
M
L
M
L
M
M
L
M
A
L
D
L
H
M
B
Independent
of
the
content,
what
we
have
precedence
or
so
far
is
solidified
the
technology
here
then
take
it
off
to
the
actually
appropriate
working
group
for
just
collaboration
and
we'll
decide
where
it
moved
on
from
there.
But
it
sounds
like
that.
Some
discussion
needs
to
take
place.
So
let's
do
that,
let's
bring
it
out
to
the
list
get
that
going
and
then
we
can
talk
about
an
adoption
boat.
Yes,
thank
you.
J
A
D
J
A
B
E
Each
one,
this
module
arguments,
looting,
instance,
sorry
and
with
with
with
beer
contenders,
the
beer
containers
contains
the
beer
conveys
the
fact
outer
defense
out,
saholic,
sorry
I'm
surface
and
the
beer
containers
defense.
All
the
configuration
parameters
related
to
the
to
the
beer.
The
beer
coffee
configurations
contains
the
global
routing
instance
configurations
and
the
parameter
and
the
interface
configurations.
E
The
global
routing
instance
configure
computer
resistance
includes
the
transport
type
address-family
mm
and
the
mat
etiology.
Oh
yes,
next,
please
control
a
complaint
configuration
this
module
of
agreement,
routing
protocols,
it's
a
support.
Is
this
an
ospf
as
controller
for
beer?
But
yes,
next
original
state
structure,
it
defense,
Multi,
multi,
topology,
identifier,
sub
domain
and
other
information,
such
as
fit
in
tax
routing
table
and
the
built-in
tax
for
the
table?
Yes,
next
notifications,
a
it
in
this
draft,
it
defies
three
fifty
cents,
PFR
identifier,
collision,
a
PFR,
zero
and
the
sub
domain
I
the
fireplaces.
A
B
A
J
And
by
the
way
you
know,
I
mean
I.
You
saw
my
remark
right.
I'd
I'd
be
happy
if
I
don't
have
to
do
the
yang
staff
at,
which
is
why
I
love
to
be
able
to
just
stick
to
see
a
lie,
but
obviously
to
to
be
full
solution.
I
want
the
yang
also
for
the
beer
te.
So
if
I
can
ever
explain
to
you,
the
teesta
then
maybe
find
some
cycle
said
that
as
well.
Thank.
A
B
Questions
comments
that
was
fast,
that
was
our
last
freeze.
Oh
I'm,
we
had
eight
stacked
up
and
I
was
hoping
at
a
little
more
dynamic,
but
a
guy
said
with
the
first
one
had
the
most
was
talk.
I
think
our
biggest
action
item
from
here
is
a
focus
on
a
design
team,
potentially
even
an
interim
meeting,
I
think
to
to
nail
down
the
redundant
issues,
ingress
and
egress,
so
just
as
kind
of
show
of
hands
who
would
be
interested
in
a
an
interim
meeting
just
to
focus
on
those
architecture.
Issues
not
gonna,
raise
your
hand.