►
From YouTube: IETF103-MPLS-20181106-1350
Description
MPLS meeting session at IETF103
2018/11/06 1350
https://datatracker.ietf.org/meeting/103/proceedings/
A
A
A
This
is
a
note.
Well,
basically,
I
think
of
it
as
everything
you
say,
do
and
write
in
a
IEF
context.
This
subject
to
the
become
black
now
Rebecca,
it's
subject
to
any
rules
that
apply
to
our
death.
So
if
you
don't
disclose
anything,
you
don't
want
to
disclose.
That's
probably
the
advice:
I
have
okay,
so
the
administrative
information
is.
We
have
an
audio
streaming.
That
means,
even
if
you
shout
loud,
you
probably
have
to
shout
into
the
mic.
A
A
A
A
A
A
We
also
have
had
no
new
artists
ace,
and
that
is
correct.
We
remember
if
this
is
the
second
time
second
meeting
in
a
row
where
we
don't
have
an
URL
see
they
had
one
lost
okay,
so
this
is
the
first
time
that
we
don't
have
a
new
RFC.
On
the
other
hand,
we
have
four
documents
with
the
is
G
and
st.
that
at
least
the
LG
ping
lag.
Multi-Port
trough
should
be
fairly
close
to
approval
and
publication.
B
A
That
is
actually
request
the
publication
for
that
assault.
Yesterday
we
have
new
working
with
documents.
The
newest
is
the
one
at
the
bottom.
The
draft
idea-
amphilocite
assay
encapsulation,
also
approved
the
acceptance
of
working
group
a
little
bit
earlier,
but
actually
posted
this
week.
Okay,
and
we
have
a
number
of
document
that
has
been
updated.
So
the
legend
here
is
that
anything
in
blue
will
be
on
the
agenda.
A
Shouldn't
say
that
the
drawtite
fm+
SSE
is
in
what
group
los
:
it
will
they
end
in
about
a
week
not
a
month
ago,
but
we
fixed
that
the
rest
are
the
ambulance
arm
or
an
ampere
is
next
in
line
to
due
to
working
group
last
call,
and
then,
after
that
we
will
do
the
MPLS
SRO
variety,
so
quite
busy.
Please
review
and
some
comments.
A
A
A
C
Thank
you,
Hodges
Greg
mercy
CT,
so
this
document
that
we
refer
to
it
describes
applicability
of
BFD
demand
mode
for
MPLS
point-to-point
OSP
and
it's
in
BFD
working
group.
Adoption
call
and
call
closes
this
Friday
I'm,
not
sure
about
what
time
zone
working
group
chairs
will
consider.
But
I
appreciate
reviews
and
comments
to
be
the
working
group
and
to
MPLS
working
group,
because
applicability
and
substance
of
the
document
is
for
MPLS
data
plane.
A
A
Okay
and
we
have
a
nama
document
where
we
don't
have
status
reports.
Several
of
them
are
because
we're
not
expecting
one
like
the
bf
be
directed
because
that's
when
the
shares,
so
we
are
still
sorting
out.
This
is
around
that
document.
The
ML
DP
mabe
I
think
it's
still.
What
is
sixth
time
it
gives
too
close
to
expiring
and
we
are
trying
to
push
it,
but
it
seems
like
that's
something
that
doesn't
really
work.
A
A
A
And
there's
some
fixed
points.
The
way
we
don't
give
so
don't
try
to
beat
elementadd,
follow
the
document
structure
and
write
the
technical
invention.
Don't
do
a
document
to
mention
there
are
one
thing
that
is:
I
have
taken
as
a
fixed
one,
that
there
is
only
20
lines
in
a
in
an
abstract.
It's
no
longer
true,
because
I
actually
seen
in
RFC
being
published
with
21
lines
so
but
keep
keep
the
earth
except
shorter
than
to
shoulder
than
20
lines
or
20
lines,
but
it
should
also
include
enough
information
to
actually
be
understandable.
A
This
document
abstract
this
document,
defines
a
new
protocol.
Here.
You
won't
go
through
a
review
by
working
group
shares
or
is
the
you
need
something
I
have
said
before
is
that
we
can,
if
you
think
about
your
next
management
manager
in
the
hierarchy,
write
the
abstract,
so
that
person
actually
understand
what
it's
all
about,
and
you
have
20
lines
to
do
it.
A
Okay,
next
and
then
we
have
the
style
guide.
One
of
the
most
common
comment
we
have
on
any
idea
is
that
okay,
first
time
he
used
an
abbreviation,
you
need
to
expand
it
and
it's
amazing.
Actually
how
often
we
miss
that.
Do
you
look
at
the
abbreviation
list,
because
that
has
two
type
of
entries.
If
you
see
mpls
star,
it
means
that
you
don't
need
to
expand
ambulance.
If
you
see
LSP
without
the
star,
you
must
expand
it.
A
A
Have
discovered
that
I'm
then
I'm
writing
are
starting
to
use
to
ACTU
abbreviations
fairly
frequently
W
DLC
has
been
there
for
some
time.
It
means
working
group
last
fall
and
I
probably
will
start
using
it
more
frequently.
We
also
have
working
group
adoption,
Paul
WG,
ap
I
have
tried
to
expand
it
when
I
use
it
that
I
after
I
know
that
I
start
using
the
WGA
P
as
a
abbreviation
in
males
that
goes
out
to
the
working
group,
so
you
should
be
aware
over
to
what
it
means.
A
So
those
are
not
really
local
to
the
ambulance
and
working
you,
but
this
is
how
we
been
probably
using
it
more
frequently
in
the
future
next
one
so
in
we
have
questions
frequently
asked
about
working,
Googlers
call
and
adoption
polls.
The
first
one
is
that
I
actually
got
from
a
long
time.
Er
former
former
a
D
and
former
IB
member
and
asked
me
a
as
a
co-author,
do
I
need
to
respond,
and
the
author
is
a
co-author
does
not
need
to
respond
to
either
a
working
regards
call
or
a
adoption
Paul.
It
doesn't
change.
A
A
So
one
reason
I
heard
is
that
co-authors
tend
to
respond
to
adoption,
pause
and
working
with
law
school
when
there
is
low
traffic
on
that
on
that
subject.
So
if
you
send
another
mail,
saying
working-clothes
lost
call
on
grossie,
and
so
then
you
increase
the
trafficking,
the
increased
awareness
hours
going
on
and
I
say
you
can
do
that.
It's
not
very
efficient.
It
would
be
much
more
efficient
if
you
actually
thought
about
people
that
has
not
responded.
A
A
I
powerboats,
so
we
do
that
twice
before
adoption,
Paul
and
before
last,
called
authors
and
courses
are
required
or
is
done,
and
you
also
require
to
respond
ahead.
Don
T's
is
not
on
the
list,
but
he
also
required
to
respond.
If
you
know
about
a
IPR
that
related
to
the
in
the
draft
that
particular
raft
and
also
and
like
Andy
should
know
if
someone
writes
in
a
acknowledgement.
Second,
that
we
would
like
to
thank
mr.
malice
for
contributing
text
that
will
trigger
it.
A
A
Okay,
there
is
a
escape
clause,
but
so
far
the
mpls
working
group
has
not
found
it
necessary
to
use
it.
What
we
probably
do
is
actually
send
a
mail
to
the
what
we
planned
to
do.
Last
time
we
were
close
to,
it
was
send
a
mail
to
the
working
group
mailing
list
and
say
this
person
has
not
responded.
Is
it
okay
to
go
ahead
and
doing
the
working
group
adoption
or
conclude
the
work
and
start
work
last
call
anyway.
A
A
And
I
bring
in
their
top
because
we
have
a
mr.
I
have
had
misunderstanding
a
couple
of
times.
We
have
a
preliminary
agenda
that
we
need
to
publish
before
the
working
group
meeting
at
a
certain
date.
We
always
miss
that
very
two
and
three
days,
but
its
preliminary
anyway,
and
then
we
have
a
final
agenda
that
we
need
to
post
at
the
13
dates
about
the
week
before
the
the
meeting.
The
this
is
all
done
on
a
best
effort
basis,
so
the
preliminary
yonder
is
an
approximation
we
list
all
the
request
was
not.
A
A
A
I
still
are
moving,
and
they
should
probably
be
taken
out.
It
also
includes
does
not
include
all
the
document
that
we
could
list
here.
So
it's
again
an
approximation.
What
I
want
you
to
do
or
I
want.
The
working
group
members
to
do
is
actually
take
a
look
at
this
list
and
see
if
you
have
a
document
here,
that
you
posted
like
two
and
a
half
years
ago,
when
you
find
it
they're
still
relevant
I
would
like
to
progress.
D
Hello,
hello,
hello,
a
clay
of
a
comment
of
this
document
list
sure
yeah
actually
on
the
third
row,
Frieden
the
previous,
the
previous
slide,
the
first
slide,
so
you
know
that
Sleeping
Beauty
one,
so
you
have
listed
document
ml,
DP,
multi,
topology
here
I
think
the
no
more
sleeping
our
Dharma
document
it
was
presented
to
ITF
ago.
We
have
reflected
it
Isis
presented
in
London,
okay.
A
C
A
E
F
I
can
see
the
presentation,
okay,
yeah
good
afternoon,
everyone.
Let
me
introduce
myself.
My
name
is
elke
Morrison
I
am
delighted
to
be
part
of
IETF
one
or
three
as
a
remote
participant
thanks
Shay
for
the
opportunity.
A
little
talk
to
you
about
the
new
draft
LDP
identifier
name.
My
object
today
is
to
get
your
point
of
view
on
this
trap
will
be
very
happy
to
receive
your
sessions
and
cards
so
in
a
scaled-up
topology
with
multiple
neighbors.
F
It
might
be
difficult
to
map
label
bindings
received
from
multiple,
lesser
IDs,
having
different
name
space
so
having
so
having
human
under
seven
understandable
needs.
Map
LS
already
would
be
very
helpful
in
reading
through
the
information.
So
the
purpose
of
this
draft
is
to
exchange
LD
pairing
fail
me
through
TP
protocol
itself,
so
the
operators
could
easily
map
the
identifier
names
to
identify
own
name
Oh.
As
we
all
know,
LDP
identifier
is
six
bytes
theta,
four
bytes
is
IP
address
or
towards
the
level
space
which
is
part
of
every
video
header.
F
So
there
are
multiple
ways
to
derive
are
obtained
independently.
Obviously
one
needs
a
static
configuration
at
each
device,
so
other
one
is
a
using
centralized
location
which
maintains
a
LDP.
Id
is
already
to
label
this
ready
to
name
mapping
and
it
has
its
own
advantages
and
disadvantages.
So
one
way
could
be
a
beginner
server
or
Sdn
controller
or
something
a
bear
NMS.
F
So
one
of
the
reasons
would
be
like
reach
ability
of
network
problems
to
reach
the
DNS
or
controller
or
NFS
the
response
time
for
Hart,
and
then,
if
there
is
a
device
so
and
one
more
thing
is
like
operators
could
want
to
have
a
different
name,
mapping
to
the
ID
at
the
runtime
so
and
it
is
easier
to
make
in
this
directly
on
the
router
instead.
So
so
what
we
are
proposing
is
a
protocol
to
carry
a
label
identifier
or
name
mapping
in
a
TLB.
F
So
this
draft
interviews
isn't
blue.
A
little
I
need
Appel,
name
plv
as
an
optional
parameter
as
part
of
LDP
holo-message.
Similarly,
they
put
address
t
lv,
o
EP
identifier
name,
will
clear
encoded
as
a
ASCII
string,
which
is
the
swinging,
will
be
mapped
to
both
r
LS
r
ID
I
mean
IP.
For
address
plus
label
space
together,
we
can
give
a
one
string.
G
F
Can
be
said,
what
will
be
the
string,
so
we
just
stopped
paying
clearly
so
I
mean
is
a
simple
proposal
with
a
introducing
and
new
optional
TLB
in
a
hollow
message.
So
we
would
like
to
hear
from
you
whether
Urbina
is
this
useful
or
like
this
extension
is
a
right
way
right
approach
or
need
further
sessions
to
improve
the
document.
F
H
Echoing
Louie
yeah,
thank
you
for
the
explanation.
Thank
you
in
particular
for
that
chunk
of
text
you
put
in
about
duplicate
names,
because
when
I
started
reading
the
document
I
was
going.
Oh
my
god
duplicates,
oh
my
god,
and
then
I
got
down
to
the
bit
at
the
bottom.
That
says
what
you
do
if
two
LS
ours
use
the
same
name.
So
that's
cool
I'd
like
you
to
spend
some
time
thinking
about
internationalization,
okay,
I,
don't
yeah!
You
currently
say
us-ascii
and
well,
where
I
come
from.
H
A
D
D
Next
slide:
peace,
yeah
I'm
on
the
next
slide,
already
Hyun's
room:
okay,
yeah
it
takes
time
okay.
So
what
is
coming
up
right?
So
so
we
have
been
presenting
LDP
number
of
DP
yang
model
drafts,
almost
every
ITF
and
update
to
the
drafts
at
least
last
ITF
at
Monte.
We
skipped
it
because
there
was
not
much
activity,
so
we
are
back
to
present
and
present
both
of
them
in
this
idea.
So
LVP
dropped
just
two
remaining,
never
even
very
quick
history.
We
presented
this
in
Singapore.
D
We
updated
and
refine
a
draft
a
bit
and
present
it
again
in
London
in
London
view
calls
for
the
working
group
last
call
and
call
was
initiated
thanks
to
Cheers
in
July
2018
right
after
the
Montreal
idea,
and
we
also
received
the
working
group
last
call
review
comments
from
Tom
fresh.
They
were
very
useful
and
we
had.
Where
does
those
comments
and
posted
the
new
revision
5
before
Bangkok
and
I'm?
Just
gonna?
D
So
while
the
next
slide
is
coming
actually
let
me
I
have
my
own
version
too.
So
I
can
keep
speaking
in
yeah,
so
so,
basically
as
a
part
of
revision
5,
there
are
two
two
main
things
we
did
a
couple
of
main
things.
One
was
working
to
blast
all
comments
we
received
from
Tom,
yes,
those
almost
all
of
them.
I
dropped,
an
email
to
Tom
and
the
working
group
to
take
a
look.
D
Some
of
the
comments
from
Tom
are
pretty
generic
and
not
specific
to
this
model,
and
he
has
noted
that
some
of
those
misses
as
well
as
you
know,
missus
as
well
as
some
of
the
the
incorrect
use
is
holocausts.
The
many
routing
area
yang
models,
so
I
have
a
slide
on
that
later
on.
We
can
go
through
that
and
lower
has
asked
me
to
maybe
touch
upon
some
of
those
generic
comments
which
is
useful
for
many
other
as
well.
D
D
Okay,
so
so
I'll
go
through
general
comments
that
tom
has
in
the
draft
I'm
just
summarizing
some
of
them.
So
one
main
comment,
I
think
as
observed
was
that
specification
of
the
common
language
that
we
have?
It
is
still
using
old
boilerplate.
We
were
not
referencing,
the
RFC
174
and
the
actual
language
which
I
have
highlighted
in
italic
later
on.
D
So
this
is
the
actual
text
that
we
have
to
add
in
specification
of
the
requirement,
so
I
found
it
all
across
my
yang
another
documents
in
ITF,
so
anything
that
I've
touched
on
bacod
I
have
updated,
but
I
notice
and
I
looked
at
around.
All
of
the
documents
are
still
using.
All
boilerplate
is
gone,
so
I
think
this
is.
This
is
in
general,
true
for
many
documents
in
working
area,
so
we
need
to
fix
that.
D
D
Publication
I
think
it's
time
that
a
doctor
should
have
a
proper
RFC
editor
notes
so
that
to
make
the
job
easier
to
replace
the
text
like,
for
example,
G
number
text,
llamo
dealer
file
names,
a
revision
of
the
revision,
the
date
of
the
revisions
of
the
module
with
the
deals
of
the
publication
and
whatnot,
so
in
general
model,
got
yanked
as
well.
As
our
document
is
missing
those
notes,
and
then
we
did
not
have
any
revision
or
version
statement
at
all
in
our
yang.
D
So
by
default
it
becomes
yang
one
dot,
o
all
the
we
were
complaining
to
young
1.11.
So
these
were
few
comments
and
excite
this
right
and
then
few
more
comments
that,
in
our
imported
modules
in
over
yang,
there
was
no
reference
statement
and
then,
of
course,
since
res
reference
was
missing,
we
were
also
missing
the
respective
RFC's
or
the
document
from
the
Normandy
reference
list.
We
were
using
tree
diagram,
but
not
referencing.
Instead
of
see
a
security
consideration,
section
was
not
conforming
to
RFC.
D
Sixty
eighty
seven
base
and
Iannetta
consideration
section
was
either
missing
in
some
cases
or
in
LDP
cases
it
was
actually
missing
and
we
have
to
add
the
fully
under
consideration
section.
So
these
are
all
comments
on
the
work
and
all
the
eight
over
l,
DP
p--
working
of
last
call
and
as
you
can
notice,
that
most
of
them
are
pretty
much
applicable
to
any
other
yang
which
may
or
may
not
have
you
know,
which
may
or
may
not
have
mister
some
of
those.
A
D
So
so
so
I
think
the
previous
slide
I
tried
to
cover
what
tom
has
you
know
captured
and,
of
course
we
addressed
all
those
comments
in
over
revision.
Five,
we
added
power
references
at
a
loss
of
RFC,
editor
notes,
but
it
you
know,
treat
diagram
references
executed.
Consideration
section
is
much
in
house
now.
He
might
still
be
lacking.
D
Some
of
you
know
some
of
the
text
that
might
you
know
we
might
have
to
might
have
to
revisit
an
update
as
part
of
the
security
review,
but
right
notes
is
in
much
much
better
shape
than
the
or
original
security.
Second
section
e
NR
section
was
totally
missing,
so
that
has
been
updated
and
it
looks
good
and
we
added
proper
references
to
the
to
the
latest
revision
of
the
the
normative
references,
informative
references,
you're
using
and
next
like
piece
so
JSON
example.
D
This
was
not
part
of
the
review
comment
from
Tom,
but
this
was
a
review
committee
ml
de
Quito
aft
only
review
by
AC,
so
we
have
added
there.
So
we
also
kind
of
applied
that
to
LEP
draft
as
an
example.
This
is
just
an
example,
but
we
have
more
more
just
an
example
in
our
or
document,
so
this
just
shows
that
you
know
how
how
you
see
a
purely
state
different
parameters
of
the
peer.
D
This
is
so
we
have
added
this
as
an
appendix
in
the
document
I
just
for
a
reference
and
I
think
this
is
a.
This
is
useful
information
to
have
in
a
yang
model
to
people
to
visualize.
You
know
how
how
the
yang
will
look
like
and
last
slide,
please.
So
as
a
next
step,
I
think
we
have
addressed
on
the
working
to
last
call
comment.
D
We
haven't
seen
an
explicit
act
from
Tom
or
others
that
looks
good,
but
at
this
point
in
time,
authors
believe
that
I
think
this
document
is
ready
to
ready
to
next
ready
to
go
to
next
step,
which
is
basically
publishing.
Unless
there
is
more
review
comment,
we
would
be
happy
to
at
the
stores
and
RESPA
and
in
eurovision,
but
for
now
from
oversight,
I
think
they're
waiting
as
an
extract
for
the
publication.
A
I
E
D
I'm
not
on
the
history
slide.
Oh
okay,
you
already
there
okay!
So
so,
let's
go
to
the
second
part
of
this
presentation,
so
it's
ml,
DP
yang!
It's
also
division.
Five,
so
LD
PNM,
LDP
yang,
are
going
hand
in
hand
with
respect
over
of
grace
and
revision,
so
just
again
reminding
everyone
history
about
Emily
P
recent
history.
These
documents
have
been
around
for
a
good
four
five
years
now,
at
least
since
Hawaii.
If
I
remember
correctly.
So
recent
history
is
that
we
posted
ml
DP
revision,
three
IDF
in
Singapore.
D
We
received
yang
dr.ali
review
by
AC
around
December
time
and
we
posted
sorry
I
I,
don't
see
that
slide
anymore,
but
I
can
continue.
Okay,
I
can
see
how
we
both
shared
a
vision
for
before
Montreal,
where
you
guys
still
hear
me.
Yes,
yeah.
The
slide
somehow
is
gone.
What
I
was
seeing,
but
I
can
continue
on
my
own
version:
I
guess:
okay,
yeah!
So
right!
So
so
we
posted
the
revision
for,
but
this
was
a
minor
update.
Somehow
we
have
missed
AC
email
and
we
did
not
update
it
until
later.
D
So
we
posted
a
revision
5
right
before
ITF
103
in
Bangkok,
and
we
addressed
basic
comments
and
actually
on
that
front,
I
replied
in
detail
to
AC
comment
and
took
comment
by
comment
and
reply
to
AC
and
I.
Think
I
had
seen
acknowledgement
from
AC
and
taking
and
acknowledging
that
the
changes
I
made
and
the
AC
has
done
a
wonderful
job.
He
was
very
detailed.
He
helped
improve
the
document
quite
a
lot.
D
D
So,
as
I
said,
we
basically
division.
5A
addresses
review
comments
from
AC.
If
you
do
the
diff
between
four
and
five,
you
see
quite
a
bit
of
changes
in
the
text
as
well
as
in
the
dot
yank
file
and
I
also
alluded
in
LDP
presentation
that
we
also
applied
comments
generate
comments
from
Tom
that
were
that
were
LEP
on
the
LDP
document.
D
D
So,
basically,
as
in
terms
of
what
we
address
young
actors,
we
revised
the
security
consideration
section
add
a
JSON
example
for
configuration
operational
state
over
the
script.
Descriptions
were
not
very
clear,
so
we
in
yangyang
module.
We
clarified
those.
We
added
missing
automatic
references,
renamed
some
of
the
the
Leafs
which
were
causing
some
confusion
to
the
reader
and
also
clean
up
some
grouping
as
as
suggested
by
AC,
and
we
also
took
loss
of
a
C's,
updated
text
and
applied
to
the
document
next
slide.
Please,
yes,
and
then
the
second
part
of
the
changes
are
we.
D
E
D
As
a
next
step
for
for
this
document
for
MLBPA
you're,
looking
for
any
further
reviews
of
the
document
on
this,
this
other
reviews
and
and
then
there's
some
some
more
revisions
we
have
to
update
but
I
think
we
believe
that
says
you
have
applied
very
detailed
comments
from
AC
on
the
document
as
well
as
actively
applied.
Only
people
capable
assed
call
comment
on
MLD
PDF
as
well.
D
Over
Emily,
P,
yanked
off
is
also
in
very
good
shape
and
authors
and
co-authors
believe
that
we
should
be
ready
for
ever
king
of
past
call
in
this
document,
as
well,
I
believe
up
to
chairs
to
discuss
this
either
now
or
on
the
mailing
list,
but
but
this
is
acting
it
from
all
sides.
This
is
an
extra
for
this
document.
B
A
So
next
step
is
probably
you've,
seen
IPR
poll
and
run
being
run
through
the
working
group.
Discord
process
and
I
would
like
to
say
something.
It's
a
pity
that
Tom
patch
is
not
here
when
he's
normally
not
here,
so
he's
not
surprised,
but
I
would
like
to
say
that
we
very
much
appreciate
the
reviews
that
Tom
patch
is
done
across
the
working
routing
area
and
a
number
of
gang
models
in
actually
work
to
kind
of
make
creating
confirmation.
J
Good
afternoon
you
can
hear
me
well,
okay,
my
name
is
tarick
and
I'm,
giving
you
a
quick
update
on
the
two
yang
drafts.
The
first
one
is
the
MPLS
base
yang
draft
the
puts
the
cornerstone
for
other
augmentations
of
MPLS
protocols
and
the
second
one
is
for
MPLS
static,
Ellis
fees,
which
is
augmenting
the
MPLS
phase,
so
I
don't
have
many
slides,
but
as
I
go
through,
I
will
describe
what
we've
done
in
terms
of
the
updates
to
the
drafts
and
then
I'll
close
with
some
next
steps.
J
So,
okay,
the
drafts
have
undergone
a
yang
doctor
review
and
we
got
quite
a
few
of
comments.
We
did
clean
up.
The
document
in
general
meet
editorial
changes
and
some
myths
that
were
pointed
out
by
the
by
Tom
patch.
Basically,
I
did
mention
his
name
in
the
next
slide,
but
it
was
found
that
he
reviewed
our
yang
model
as
well
and
the
drafts,
and
we
did
proactive
job
of
you
know.
We.
J
We
did
notice
that
the
LDP
model
and
the
LDP
model
they're
being
reviewed,
and
there
were
generic
comments
as
Camryn
was
saying
that
are
applicable
to
pretty
much
all
the
ank
models
that
are
being
published
at
IETF.
So
we
took
care
of
those
comments
that
are
applicable
to
our
drafts
and
we
addressed
them.
Although
they
were
not
called
out,
it's
basically
the
ones
that
were
called
out
and
we
did
put
references
where
it
was
missing
for
imported
modules.
J
J
One
detail:
we
did
import
this
common
type
that
is
defined
in
their
routing
rerouting
common
types.
Mpls
label
stack,
so
we
found
no
reason
to
redefine
it
in
our
model
and
we're
reusing
that
pretty
much
the
the
what
Cameron
presented
the
generic
comments,
where
applicable
to
our
model
I'm
not
going
to
go
through
them,
but
we
took
care
of
those.
J
A
G
A
Actually,
who
who
requested
the
gang
doc
to
review
Ian
I,
was
my
ok.
It's
such
a
long
time
ago
that
I've
forgot
about
it.
Ok,
so
those
four
documents
are
basically
80%
of
the
gang
models
that
we
can,
as
they
have
in
the
ambulance
work.
If
we
also
have
a
document
that
is
well
LS,
beeping
gang
and
the
authors
of
that
document
has
actually
requested
work
group
adoption,
so
we
will
start
that
process
well
shortly
after
this
meeting
we
have
many
such
pros.
A
B
B
B
With
SBR
links
between
ASPRS
and
the
central,
a
PE
traff
defines
different
types
of
suits
to
be
able
to
use.
You
know,
send
traffic
on
these
links
and
choose
either
load-balancing
or
single
links
or
set
of
load
balancing
links.
So
when
there's
a
deployment
that
that
uses
this
EPE
sits,
if
there
is
a
there
is
a
operational
requirement
that
to
be
able
to
do
ping
and
traceroute
across
these
is
these
links,
and
these
validates
the
control
plane
and
data
plane
for
these
sets.
So
it's
useful
to
have
this
cross.
B
Yes
for
localization
in
case
you
know,
there
are
problems
in
in
control,
plane
and
forwarding
plane
programming
for
these
sets
so
that
there
are
we
trying
to
see
two
use
cases.
One
use
case
may
be
the
two:
a
SS
belong
to
a
single
operator
and
they
have
been
segregated,
for
because
of
you,
know,
easier
and
operational
and
maintenance
reasons,
and
there
could
be
another
use
case
where
the
two
guesses
belong
to
two
different,
completely
two
different
customers.
So.
B
So
what
we
really
need
is
the
fixed
stack
definition
for
EPA
six.
So
we
have
the
spear
no
suit
and
pH
since
he
said
so
for
OAM
to
work.
We
need
fixed
AK
definitions,
so
the
definition
looks
like
so
we
have
new
target
fixed
at
TLV
definition.
So
this
one
defines
the
tech
stack
for
both
note
said
and
I
Jason
C
said
so.
The
information
to
be
validated
is
pretty
similar
for
both
no
sets
and
adjacency
said
so.
I
thought,
probably
you
know
having
just
one
for
them
is
good
enough.
B
So
we
have
the
localized
number
remote
ace
number
locally
at
the
face
address
remote
interface
address
and
advertising
bgp
router,
ID
and
receiving
BGP
route
right.
That's
the
definition
of
the
effect
for
here
no
set
energy
since
I
said
so.
It
supports
both
ipv4
and
ipv6
and
and
the
the
length
would
decide
what
exactly
is
being
used.
Whether
it's
for
ITV,
486,
Sapir
set
said,
which
defines
load
balancing
require
I
mean
see,
the
label
would
load
balance
across
multiple
names
belonging
to
different
aliases.
The
the
links
could
belong
to
could
be
going
to
different
aliases.
B
So
the
there
is.
This
number
of
elements
in
the
set,
the
the
link
representation-
is
exactly
the
same
as
before,
but
you
could
represent
those
links
multiple
times
in
this
target
effect,
so
validation.
So
there
are
two
kinds
of
validations
that
may
be
necessary.
One
where
we
can
you
know
operator
can
allow
the
cross
is
validation
case
where
the
Clausius
validation
is
not
allowed,
because
the
OEM
packet
is
not
allowed
to
reach
the
other
side,
so
that
can
be
that
can
be
controlled
by
a
configuration.
B
A
A
Yeah
can
I
ask
how
many
has
actually
read
this
draft.
There
are
a
handful
or
little
bit
more,
how
many
of
you
who
had
read
it
has
read.
It
I
think
it's
ready
to
start
things
like
MPLS
r/t
review
and
the
general
process
to
make
didn't
go
to
working
group
adoption
about
the
same
number,
a
little
bit
less.
A
B
The
problem
that
we
are
looking
at
is
we
have
to
trace
a
srte
path.
So,
let's,
let's
assume
the
path
is
represented.
Using
two
labels
on
the
label:
stack
1,
0,
0,
4
and
1
0
0
5.
So
we
have
this
network
with
R
1,
2,
R
5
and
then
with
the
intent
is
to
send
the
traffic
from
R
1
to
R
4
and
then
our
4
to
our
5.
So
those
are
represented
using
two
loads,
so
it
1
0,
0,
4
and
1
0
0
5.
B
So
when
when
MPLS
trace
or
procedure
has
to
be
supported
for
this
label
stack
so
based
on
RFC
8
287,
it
suggests
that
the
when
you
start
tracing
this
this
SRT.
If
this
tunnel,
you
have
to
start
with
all
the
outer
label,
the
topmost
to
start
with
1,
and
then
they
all
the
inner
label
set
to
0.
So
so
the
step
one
we
set,
1
0,
0,
4,
2,
1
and
then
1
0,
0,
5,
2,
0
and
then
packet
is
sent
to
R
2.
B
So
on
our
to
the
TTI
check,
fails
for
the
out
PL.
So
the
packet
packet
goes
to
R.
2
and
then
our
two
cents,
an
and
pls
eco
reply,
which
says
so
so
based
on
what
is
programmed
on
R
2,
it
says
I
am
the
transit
for
1
0,
0
4,
so
the
next
step
when
1.
This
is
a
echo
reply
that
r2
is
transit.
Then
it
increments,
the
the
TTL
for
1
0,
0
4.
So
now,
in
the
second
step
it
becomes.
B
Okay,
if
the
second
step,
the
TTL
becomes
two
four
one:
zero,
zero
four
label
and
then
packet
is
sent
out
from
our
one,
so
our
to
reduce
decrement
the
TTL.
It
goes
to
our
three
and
our
3
again
decrement
the
TTL
which
which
makes
the
TTL
for
label
one
zero
zero.
Four
two
zero.
So
our
three
gets
the
packet
and
then
it
sends
an
eco
reply.
So
our
please
be
PHP
for
label
one
zero,
zero!
Four,
so
it
sends
the
echo
reply
as
I
am
the
transit.
B
So
the
next
step
are
one
increments,
the
TTL
of
one
zero,
zero,
four,
two
three
and
then
against.
Thanks
back
the
packet
on
the
wire
and
then
now
it
in
this
step,
it
reaches
r4,
so
it
reaches
r3.
So
r3
is
popping
the
labels
or
one
zero.
Zero
four
is
removed
from
the
label
stack
and
then
packet
is
sent
to
r4
when
it
comes
on
at
four
there's
only
one
label:
one
zero,
zero
five
and
the
TTL
here
is
0,
because
the
initially
the
TTL
was
set
to
zero
four
nine
one:
zero
zero
five.
B
Let's
assume
that
you
know
packet
gets
to
are
for
our
four
processors
and
then
see
what
what
what
will
be
the
problem,
even
if
our
four
processes,
so
our
flow
process,
processes
the
packet-
and
it
says
I-
am
the
egress
for
this
fact,
which
is
holdout
4.4.4
and
then
sends
that
return
code
to
r1
in
the
MPLS
echo
reply.
So
our
well
now
implements
the
layer.
The
TTL
of
one
zero,
zero
five,
two
one
and
then
it
starts
tracing
and
sets
the
outer.
B
So
it
has
reached
the
egress
for
that
that
tunnel
represented
by
label
one
zero
zero
four,
so
it
such
that
255
and
then
it
starts
sending
the
packet.
So
in
this
case,
when
the
packet
reaches
our
four
it's
again,
because
the
TTL
is
one
so
it
again
expires
on
our
force.
The
packet
gets
punted
to
our
four,
but
our
expectation
was
package
should
reach
our
find
in
this
step.
So
that's
the
problem
in
short,
pipe
model,
the
short
pipe
not
aware
the
TTL
is
not
getting
coffee
and
the.
H
I
do
a
question
about
the
short
pipe
before
you
go
on
to
the
other
ones.
On
the
on
the
on
the
three
step,
three
that
one!
No,
then
you
have
to
step.
Three
is
the
next
step,
three
that
one?
Yes,
why
do
you
you've
established
the
top
label
TTL
span
because
you've
got
to
the
end
of
it.
So
you
know
the
the
TTL
for
the
top
label
that
will
always
work.
B
H
I
guess
not,
except
now
you
have
to
actually
change
two
things
in
your
sauce
packet
rather
than
just
one
thing.
So
if
you,
if
you
look
at
the
previous
slide
of
also
up
step
3
there,
the
the
setting
the
TTL
to
three,
you
gets
you
all
the
way
that
you
needed
to
go.
So
there's
no
need
ever
to
change
that
three
again
in
any
of
your
processing.
H
B
B
So
the
problem
with
uniform
model,
so
we
have
first
step
same
no
difference,
comes
to
r2
and
then
gets
an
eco
reply.
Step
two
comes
to.
R3
gets
an
eco
reply,
same
no
difference,
step
three
where,
after
he,
pops
the
label
and
r4
gets
the
packet,
so
uniform
model
generally
copies
the
label
from
top
label
to
bottom
label.
B
But
in
this
case
it
does
not
get
copied,
as
we
see
the
top
label
that
one
zero
zero
four
has
a
TTL
of
one
when
it
is
just
just
just
to
be
parked
and
the
bottom
level
one
zero
zero
five
has
the
label
zero,
so
generally
MPLS
TTL
copy
mechanisms
from
top
label
to
bottom
label.
They
do
not
copy
the
outer
tunnel
TTL
to
the
inner
TTL
if
the
inner
TTL
is
lower.
This
is
this
is
done
to
prevent
you
know
packets,
getting
looped.
B
B
So
all
we
need,
as
in
a
solution
is,
is
for
PHP
router
to
say
that
I
am
PHP
and
then,
when,
when
ingress
knows
that
it
has
reached
the
PHP,
it
has
to
increment
the
inner
label
TTL
to
1
instead
of
waiting
till
the
packet
reaches
the
egress.
So
let's
see
what
happens
when
we
do
that.
So
in
this
the
step
one
remains
same.
No
change,
step,
two
means
change.
The
only
change
asked
is
are
three
will
inform
our
one
that
it
is.
B
It
is
PHP
because
it's
popping
the
label
and
then
when,
when
it
does
that
our
one
will
increment
the
TTL
of
the
next
label,
and
and
so
when,
when
the
packet
reaches
our
three
the,
even
though
the
top
label
is
popped,
we
have
the
bottom
label
with
with
the
TTL
of
one.
So
it
it
goes
to
our
four
and
then
gets
then
gets
punted
up
to
tre
and
then
our
four
says
I
am
egress
and
then,
when
it's
a
secret,
seven
will
increment
the
next,
the
next
label
TTL
to
two.
So
so
it
all.
B
So
they
there
are,
there
could
have
been
two
possibilities.
How
our
three
would
inform
that
it
is
PHP,
so
one
would
be
one
could
be.
The
return
called
introducing
a
new
return
code,
but
that
would
make
it
backward
incompatible.
So
is.
If
some
are
one
does
not
understand
that
return
code.
Then
it
would
not
it
the
behavior,
you
know
it.
It
would
cause
in
throw
up
issues.
So
so
the
proposal
is
to
put
the
in
the
da's
flats
in
the
DD
map.
B
So
these
are
some
of
the
alternate
approaches
evaluated
to
solve
this
problem,
so
PHP
router
itself
sending
egress
written
code
instead
of
new
code.
It's
PHP
itself
sends
return,
hold
that's
one
of
alternate
approach,
evaluated
and
then
setting
all
inner
labels
to
1
instead
of
0
and
then
setting
all
inner
labels
to
255
and
then
setting
inner
label
to
2
on
reaching
egress
those
other
coaches.
B
So
setting
as
a
the
PHP
router
itself,
sending
egress
written
code
that
you
know
the
tunnel
is
not
actually
being
traced
in
the
egress,
so
you
know
the
intent
is
to
make
sure
in
the
entire
MPLS
path
is
fine,
so
you
have
to
be
able
to
verify
it.
Control/Play
and
data
train
till
the
actually
dress
and
then
PHP
itself
sending
egress
from
it
does
not
validate
the
last
hop
and
it
also
does
not
effectively
detect
the
problems
like
artoo
had
programmed
a
pop
operation
for
one
zero,
zero.
B
Four
and
if
I
have
four
had
something
programmed
miss
program,
something
all
those
cases
will
get
missed.
If
we
do
this
so
this
this
solution
was
discarded
and
also
you
know,
the
intent
was
when
we
created
the
label
stack
one:
zero,
zero,
four
in
one
zero,
zero.
Five,
the
intent
of
the
operator
was
to
figure
out
that
that
traffic
should
hit
1
0
0
for
our
four
and
then
go
to
our
five
and
and
if
you
do
this,
that
that
won't
be
guaranteed.
B
So
this
is
not
the
best
way
to
do
it,
and
next
option
evaluated
was
the
setting
inner
labels
to
one.
So
this
this
works.
Fine,
when
you
know
the
the
MPLS
path,
the
asahi
path
is
the
all
the
nodes
on
the
path
are
programmed
correctly
so
setting
inner
label
to
begin
with.
If
we
set
the
inner
label
to
one,
this
mechanism
works
fine,
but
the
problem
is
if
there
is
something
wrong
programmed.
For
example,
R
2
is
prematurely
popping
the
label
for
one
zero
zero.
Four,
then
it
may
not
detect
the
this
problem
immediately.
B
So,
for
example,
in
the
first
step,
the
one
has
on
one:
zero:
zero
for
n
1,
zero,
zero
five,
both
cetera
TL
2
1-
and
in
this
step
to
the
by
mistake.
Let's
say
by
mistake:
our
two
removes
the
label:
1
0,
0
4
in
this
case
packet
reach
three
and
then
defect
stack
still
contains
4.4.4,
our
four,
which
is
a
valid
effect,
and
it's
a
transit
for
that
fact,
and
whereas
the
incoming
label
map
is
one
zero,
zero
five,
so
the
the
factors
are
validated.
B
The
incoming
label
map
is
validated,
and
if
there
is
no
validation
across
these,
then
it
won't
detect
I
mean
it
will
detect
ultimately,
but
not
in
the
first
step,
so
it
will
go
on
like
after
he
says,
I
am
transit.
Car
one
will
just
increment
the
increment
the
TTL
and
then
keep
incrementing
the
TTL
to
reach
r3
and
then
and
then
keep
getting
transit.
So.
B
The
another
option
that
we
evaluated
was
setting
all
the
label
below
the
stack
to
255.
This
mechanism
obviously
doesn't
work
well
because
once
you
reached
r3,
your
label
stack
would
contain
255,
so
you
wouldn't
even
receive
the
unity,
will
heat
to
the
router
r4
to
trace
that
MPLS
path.
So
this
won't
work
and
the
last
option
is
the
so
when,
when
the
egress
is
reached
instead
of
setting
the
inner
label
TTL
to
one
set
it
to
two,
so
this
also
works
well
in.
B
If
the
MPLS
path
is
programmed
correctly,
so
it
would
reach
our
four
and
then
once
our
four
is
done,
the
next
step.
It
would
reach
our
faith,
but
it
has
the
same
issue
that
that
is
described
for
same
as
the
setting
inner
labels
to
one
so
so
tracing
the
society
paths.
There
may
be
already
implementations
that
are
doing
one
of
these,
like
setting
label
to
the
the
mechanism
described
in
this
slide,
or
this
mechanism
described
in
this
slide
because
it
kind
of
works.
B
G
B
B
K
B
B
B
So
not
necessarily
so
so
why
I
am
trying
to
say
is
mechanism
can
be
incremented
deployed?
Incrementally,
like
you
upgrade
you
don't
have
to
take.
You
don't
have
to
worry
about.
You
know.
Some
load
is
upgraded.
Some
node
is
not
upgraded,
so
there's
going
to
be
in
therapy
shoes
right.
If
you
you
are
sending
something
in
D
s
map,
let's
say
you've
upgraded
the
some
of
your
core
routers
and
then
headend
is
not
upgraded.
It's
just
fine.
It
just
works.
How
it
works
today
like
if
you
have
deployed
already
deployed
srte
in
your
network.
G
In
respectively,
so
the
the
the
major
question
I
have
right,
so
we
can
actually
discuss
offline
about
this.
My
major
concern
is
your:
why
don't
we
just
use
the
TTL
label
for
the
the
very
next
label-
TTL
equal
to
one,
not
all
of
the
labels,
but
just
the
very
next
level
TTL
equal
to
one,
why
it
doesn't
work.
G
B
In
this
case,
the
effect
is
valid
right
when
it
comes
see
if
the
speck
validation
and
as
incoming
label
validation,
so
incoming
label
is
different
from
what
is
feck,
and
this
is
if
your
in
your
network,
if
it's
a
peer
shapiro,
this
will
always
happen
right.
The
incoming
label
is
different
and
your
fact
is
different
because
you
have
already
hopped
the
label
right,
so
so.
B
G
B
B
For
binding
sites,
there
is
no
difference
right.
It
would
be
same
like
how
so
so
the
way
the
binding
site
is
supposed
to
work
is
you
are
going
to
self
extract
change
from
that
point
towards
your.
So
when
you
discover
that
you
know
this
label
is
stitched
to
a
set
of
different
labels,
you
signed
up
extract
change,
but
the
TTL
value
it
just
same
as
in
8
0
to
9
right
I
mean
you
just
continue.
Incrementing
the
let's.
B
So
so
there
are
different
modes,
defining
it
0
to
9.
There
are
there's
one
more
where
you
can
trace
the
contained
LSP
when
you
are
doing
the
effect
stack
I
mean
when
you
are
when
the
fixed
at
change
is
happening.
There's
another
mode
waiting
you
can
skip
tracing
and
you
can
trace
both
options
and
then
RFC
l0
to
line
defines
how
exactly
the
tracing
works
right.
So
the
TTL
for
the
TTL
value.
If
you
want
to
press
it,
it
would
just
have
to
increment
for
the
bindings,
if,
like
you
start
from
but
binding.
D
B
Swap
operation
for
the
TTL
values
right
so
I
think
we
copy
from
this
label
and
then
reduce
it
by
one
and
that's
how
it
gets
for
the
swap
labels.
So
generally,
if
you,
if
you
just
just
forget
binding
site
for
a
movement,
how
does
this
TTL
get
copied
from
flick?
Are
to
there's
a
stack
like
this
fat
right?
It's
going
from
like
you
have
1
0,
0
2
it
swaps
to
1,
0,
0
2,
and
when
it
does
that
it
decrements
that
utl
by
one,
I
think,
that's
the
same
way.
B
B
B
L
L
Jennifer
coming
from
adrene
adrene
is
at
rest
on
the
extensions
removal
we
could
discuss.
Regarding
the
first
presentation.
It's
some
student
zipping
Angela
to
be
fair.
There
is
another
draft
from
Zafar
which
is
a
very
seminal
which
is
in
spring
original
or
so,
and
nebulous
a
Dada
in
a
different
slot.
So
next
wanna
so
clearly
buzz
draft
are
redundant.
A
A
If
there
is
a
break
soon,
we
should
be
on
this
floor,
I,
think
and
then
okay,
so
tomorrow.