►
From YouTube: IETF98-TSVWG-20170327-1710
Description
TSVWG meeting session at IETF98
2017/03/27 1710
A
B
Ok,
this
is
TSP
wgt
transport
area
working
group
they're-
not
three
of
us
up
here,
I'm
david
black
guy
in
the
green
green
sweater
here
is
gori.
Gori,
fairhurst
and
Wes
Eddie
has
joined
us
up
here
more
about
that
in
a
few
slides.
This
is
the
note.
Well,
please
read
it.
It
applies.
Shoot
applies
to
anything
said
or
presented
in
this
session.
We,
let's
see
I,
believe
we
have
a
note
taker
and
a
scribe.
B
B
Are
Raquel
to
also
thank
you.
Thank
you
very
much,
Vincent
quick
reminder
if
you
put
tsv
WG
as
the
as
the
second
element
in
a
as
sorry
as
a
third
film,
a
giraffe
named
your
the
chairs
are
more
likely
to
take
to
take
note
of
it
and
figure
out
whether
whether
us
we
we
ought
to
do
something
about
it
in
the
working
group.
Okay,
with
a
chair
up
here,
welcome
welcome
Wes.
Thank
you
very
much
long
I
get
background.
Clipart
ran
for
a
TD.
You
can't
give
apparently
apparently
apparently,
apparently
apparently
mrs.
B
B
Alright
doctor,
if
you
request
a
standing
reminder,
docking
quality
relies
on
reviews
and
you'll
hear
some
requests
for
docs.
We
need
reviews
on
during
this
session
and
the
Thursday
session,
please
your
Doc's
here
at
least
one
of
the
doc
from
the
working
group.
If
you
want
documents
that
you
care
about
reviewed,
please
do
likewise
some
of
your
documents.
B
Okay,
on
to
the
news
we
have
for
accounting
for
RFC's
published
since
I,
give
97
in
Seoul
network
transport
circuit,
breakers,
80-84,
EP
usage
guidelines,
80
85,
which
our
places
RC
5405,
jeering,
GRE
and
UDP
8086,
which
is
a
working
example
of
UDP
encapsulations,
a
little
more
important
than
just
a
a
weight
of
another
way
to
run
a
giri
and
diffserv
interconnection,
which
is
RC
8100.
You
two
chairs
who've
been
here
while
had
been
very,
very
busy
I'm
Shepherd
of
three
drafts,
author
of
for
us
our
author.
B
The
fourth
gory
is
a
author
of
two,
a
Shepherd
shepherd
for
one
and
the
source
of
the
crucial
insight
that
enabled
the
fourth
one
to
actually
get
actually
get
done
at
all.
So
if
you
wonder
why
not
much
else
has
happened
since
so,
we've
been
kind
of
busy
making
sure
these
happened.
There
are
still
20
drafts
and
Archie
under
q
gift
serve
and
WebRTC
qos
details
and
cap
of
sctp
they're,
both
waiting
for
the
sctp
I
dated
racks
on
the
next
page.
B
We
don't
have
any
IDs
and
ITF
last
caller
post
idea
of
last
call
processing.
Yet
three
IDs
include
that
completed
working
group
left
last
call.
They're
waiting
varies
for
group
chaired
or
author
follow-up
Tom
congestion
feedback
is
mine,
I
talked
to
Shing
Peng
a
yesterday
and
apologized.
The
fact
I
should
have
pay
attention
new
version
submit
in
january,
and
it's
now
March
and
I
still
need
to
pay
attention
to.
It.
I
believe
that
one's
done,
but
I
just
have
to
check
that
it
responds
appropriately
working
with
glass
wall
comments.
A
B
So,
with
a
little
luck,
these
two
drafts
will
move
up
to
at
least
the
first
position.
This
slide
and
perhaps
further
before
the
next
I
ATF
diffserv
to
arguably
802
11
mapping.
I've
talked
to
fred
baker,
who
is
one
of
the
authors
here?
We're
going
to
have
some
work
to
do
on
this
one.
There
were
significant
issues
raised
against
the
draft.
I
will
be
worked
following
up
with
Fred
and
and
the
other
author
Tim
offline
as
a
very
good
chance.
B
This
one's
going
to
need
a
second
burger
blast,
call
as
significant
group,
basically
the
mappings
in
this
graph
of
fine.
It's
everything
else
that
we
had
some
problems,
starting
with
something
starting
with
security.
Okay,
the
East.
An
experimentation
draft
is
likely
to
go
to
work
in
group
last
call
soon,
and
then
we
have
five
more
drafts
in
the
working
group,
diffserv,
lower
effort,
/
hot
behavior,
ec
encapsulation
guidelines,
EC
and
I
capsule
gather
for
tunnels
at
you,
shim
headers,
sctp,
net
draft
and
sctp
arata,
and
issues
draft
all
right.
B
Alright,
so
a
teacher
of
G
agenda
we're
rocket
we're
going
to
show
you
the
agenda
in
in
a
minute
we've
gone
through.
The
note.
Well,
we've
done
through
done.
Dakin
accomplishment
to
status
milestones
view
is
coming
up,
the
next
slide
drafts
and
then
Thursday
we'll
come
back
with
note.
Well,
maybe
a
few
notes.
New
chairs
and
more
drafts
longer
session
is
on
Thursday.
B
Okay,
milestone
review.
Here's
the
eye
chart:
okay,
we've
been
sufficiently
busy
getting
those
four
RFC's
out
that
we've
had
a
three
drafts
that
are
now
behind
on
milestones:
the
first
to
tell
Jason
feedback
and
sctp
new
data
chunk
the
best
guess
between
gory
and
myself
is
that
will
get
the
publication
requests
for
those
submitted
in
April
next
month.
We
we
just
have
to
review
them.
The
best
guesses
both
are
both
are
ready
to
go
the
Airdrie
on
eleven
draft.
Second,
diffserv
a
311
mapping,
draft
best
guess
on
that
is
sep
tember.
B
We
think
in
second
group
last
call
as
likely
major
made
major
revisions,
so
we're
going
to
push
that
at
this
with
hope,
get
it
done,
get
it
done
sooner.
Sctp,
net
draft,
no,
that's
not
getting
submitted
TI
ESG
next
month,
exercise
and
very
wishful
thinking.
December
is
more
realistic,
is
more
realistic
date
based
on
talking
the
author,
who,
among
other
things,
is,
is
going
to
be
working
on
making
sure
the
sctp
new
data
chunky
idea
draft
moves
we're
going
to
talk
about
the
milestone
for
lower
effort
of
phb
draft.
B
As
part
of
the
discussion
of
that
draft,
the
main
milestone
is
not
necessarily
unreal
list.
We
need
to
have
a
bigger
scope,
discussion
about
a
deployability
transition
and
things
like
that
and
when
we
see
our
discussion
goes
on
a
better
idea
of
milestone
to
sign
to
that
one.
The
experimentation
draft
needs
a
new
milestone,
also
we're
wearing
the
the
Hat
of
author
of
that
draft
I'm
going
to
suggest
June.
We
will
need
work
of
last
call
it
shortly,
but
I'm
we've
done
without
revising
it.
B
Maybe
June
before
we
realize
what's
happened
to
us
the
condition
of
vacation
draft,
which
is
back
up
here
in
April
and
the
property
association
of
occasion
across
a
pipe
eternally
separated
by
a
shim
draft.
We
bundle
additional
vacation,
capsulation
graft
move.
September
the
other
hand.
We
have
no
milestone
for
encouraging
draft
authors
to
use
shorter
draft
titles
and
the
sctp
Arad.
An
issues.
Draft
September
still
seems
like
a
good
time
frame
for
that.
This
is
a
heads
up
on
milestones
and
essence:
we've
got
two
drafts
were
likely
to
be
sending
our
RFC
publication
fairly.
B
Alright,
we
have
three
IDs
looking
for
adoption.
We
discussed
these
in
Seoul
and
I
think
the
direction
is
to
add
optimum.
A
West
will
be
talking
a
bit
to
these.
These
will
come
up
in
come
up
in
the
session.
In
a
session
on
Thursday,
a
TSP
WG
is
the
venue
that
is
intended
for
for
work
on
these
drafts
and
will
pick
these
from
pick
these
up
on
Thursday.
B
We're
actually
in
the
middle
of
item
one,
although
you've
just
seen,
item
2,
which
is
we've,
talked
through
the
the
summary
of
the
four
drafts
with
which
are
milestones.
We
will
pick
up
the
elie
phb
draft
under
item
3.
This
is
the
rest-
agenda.
These
four
drafts,
easy
experimentation,
the
low
effort,
Birkhoff
behavior
and
the
two
ec
and
tunnel
encapsulation
drafts.
B
Everything
else
is
thursday.
They're
the
the
chairs
up
front
and
some
of
the
and
the
TCM
p.m.
chair
has
been
talking
about
what
to
do
with
the
RTO.
Consider
draft
I
will
say
no
more
other
than
we
will
have.
We
hope
have
something
intelligent
at
coherent
and
roughly
in
LA
and
and
hopefully
that
will
get
consensus
to
say
on
Thursday
3
3
s.u
TP
drafts,
the
most
important
of
which
is
SCTV
end
date
of
the
first
one,
which
we
hope
to
publish
shortly.
B
B
Vincent
has
a
couple
of
drafts
on
a
on
for
Derek
correction
framework,
and
then
we
have
the
Alvarez
drafts,
which
will
be
the
the
Bob
Briscoe
and
Wes
Eddie
show
anything
give
us.
And
finally,
we
have
yes
sitting
down
under
item
7
Ingemar
Johansson
ask
for
a
I
believe:
that's
a
congestion
control.
Our
presentation
well,.
B
Okay,
anybody
want
to
bash
the
agenda
going
once
going
twice:
considerate
bashed.
We
may
do
it
again,
the
start
of
the
session
on
on
Thursday,
okay.
So
oh
yeah,
one
more
thing,
sorry
for
all
speakers
and
it
looks
like
I'm
up.
First,
we
have
this
standing
reminder:
there's
a
pink
box
on
the
floor,
you're
supposed
to
stand
in
it.
This
is
what
happens
if
you
don't.
B
I'll,
second,
that
so
I'll
step
into
the
pink
box,
and
that
means
I'm.
Now
the
standing
here
is
the
author.
The
experimentation
graph.
This
track
was
a
dot
by
work
group.
After
after
some
discussion,
there
were
a
few
comments
during
adoption
pigment,
one
from
a
Karen
Nielsen,
the
dash
0
1
version
of
it
reflects
that
comment.
I
know
of
nothing
else,
needs
you
done
to
this
draft,
so
I
think
we're
it's
basically
ready
for
working
group
last
call.
A
B
D
I
reviewed
the
document
and
I
believe
I
sent
some
comments
somewhere,
not
sure
if
I
send
it
to
the
readiness
or
to
you
whatever,
but
my
high-level
comment
was
that
the
document
were
very
much
sweets
like
we
do
this,
because
we
want
to
do
this
experiment
and
I
think,
that's
wrong.
What
we
should
say
is
we
do
this,
because
it's
the
right
thing
to
do
and
by
the
way,
this
open
up
some
experiment,
space,
I,
don't
think
it's
about
the
content.
It's
really
about
the
wording
a
little
bit.
A
It's
a
PS
and
I
guess
the
document
should
last
and
be
be
seen
as
inspiring
in
future
times
when
we're
not
just
enabling
something,
it's
actually
part
of
the
spec.
So
is
that
what
you're
saying
you
change
the
wording,
so
it
actually
states
what
we
really
want
to
do?
No
route,
rather
than
the
intention
of
enabling
the
experiment
says
ill.
We
want
to
make
these
changes
because
it's
the
right
thing
to
do
and
by
the
way
that
allows
these
experiments
afterwards.
Yes,.
B
Ok,
well,
yes,
we
work
offline
on
wording
of
the
draft
is
guilty
as
charged
and
of
the
most
part.
I
left
the
ration
rationale
for
the
experiments
to
the
underlying
paragraphs
and
sound
like
vehicle
and
just
nothing
to
a
rationale
to
explain
why
these
changes
ought
to
be
made
in
principle,
I'll,
try
to
work
I'll,
try
to
work
wording,
the
drop
line.
Yes,.
C
A
A
A
And
this
is
about
the
law,
a
pit
/
hot
behavior,
the
hot
behavior
that
we're
going
to
define
that
will
enable
scavenger
class
or
a
lower
priority
service
through
the
internet
and
we're
going
to
talk
about
the
core
point
in
dscp
land
so
forward.
Please-
and
the
goal
of
this
is
to
obsolete
the
recommendations
on
use
of
cs
one
for
this
class
and
we've
done
a
number
of
updates
from
zero
zero,
two
zero
one
and
largely
result
of
feedback
from
the
working
group.
A
One
of
the
things
we
added
was
the
requirement
that
congestion
control
should
be
used
with
the
PHP.
So
a
scavenger
class
and
application
actually
has
a
congestion
control
appropriate
to
that
and
that's
really
useful.
If
the
dscp
doesn't
survive
the
path
or
you
don't
receive
the
/
hot
behavior
expect.
So
that's
a
good
transport
recommendation.
I
think,
and
rudy
good
guide
also
make
suggestions
about
what
to
do
about
the
dropping
precedence
for
the
scavenger
class
traffic
and
and
that
wording
has
been
revised.
A
It
brings
us
back
to
the
recommendations
in
section
5
and
we
don't
recommend
dscp
bleaching.
That's
where
you
reset
the
diff
circle,
point
20
and
but
there's
two
different
semantics
for
Ellie.
What
to
do
and
whether
to
remarket
to
be
is
okay
or
whether
to
only
transmit
it
and
if
the
pH
be
supported
and.
A
We
could
use
to
s2
DSC
PS
that
might
be
useful.
That's
not
currently
in
the
draft
and
you're
a
number
of
things
here.
The
working
group
can
make
comments
on
will
come
later
to
which
DSC
PS,
we
might
think,
are
best
for
this
service,
but
maybe
we
want
more
than
one
as
a
possibility.
The
current
suggestion
is
just
to
use
one
and
let's
go
next
line.
A
0
angles,
I'm
end
up
with
a
very
strange
core
point
and
so
to
get
all
this
right
and
text
has
changed
but
and
yeah
read:
read
section
60,
explicitly:
wats
changes,
respect
to
459
point
okay,
and
then
we
have
some
comments
from
David
black.
First,
we
need
better
distinction
between
networks
in
general
and
networks
with
full
support
of
ally,
and
we
probably
have
work
to
do.
This
is
not
a
document.
That's
finish.
It's
a
document
that
needs
people
to
read
it.
The
must
remark,
requirement
and
needs
to
be
discussed
in
section
5.
B
I
think
what
I
do
is,
if
you're
going
to
bring
up
your
your
slides
from
map
Margie
or
something
like
it
later.
Let's
do
it
in
that
context,
so
this
is.
This
is
actually
a
transition
concern,
it's
fairly
easy
to
put
musts
on
new
usage.
This
is,
must
on
deployed
running
code,
which
is
a
little
interesting
yeah.
A
Okay
and
what
whatever
we
do,
we
got
to
get
the
eye
on
the
considerations
correct
and
all
these
things
will
be
incorporated
into
the
next
version.
What's
next
slide,
Oh
discussion
of
ecn
in
combination
with
Ellie
has
been
suggested.
That
could
be
an
interesting
one
to
run
I,
don't
think
we
currently
have
any
text
on
list,
and
so
should
any
le
trafic
we
dropped
prior
to
ecn
marks
for
any
other
PHP
and.
A
Something
the
working
group
might
have
an
opinion
on
and
should
le
be
able
to
support,
ecn
or
not
how
how
does
the
less
than
best
effort,
lower
3rd
class,
actually
interact
with
ecn
marks
and
different
people
have
different
opinions.
That's
been
discussed
on
the
list,
but
nothing
really
is
come
out
of
that.
So
that's
something
that
still
needs
to
be
talked
about.
A
And
then
there's
a
choice
of
which
dscp
to
use
for
this
new
class.
We
needed
the
SCP
to
market
traffic
and
there
were
two
proposals.
Originally
proposal
number
one
was
used
dscp
for
paul's
number
two
was
wind
used
dscp
to
and
of
these
two
dscp
21.
The
beauty
contest
with
that
still
has
problems,
and
one
good
thing
about
dscp
to
is
its
allocated
from
the
pool
ending
with
zero,
which
is
my
anna
allocated,
google
for
standards
action
drafts.
A
So
when
this
becomes
our
AC
anna
commit
that
allocation
and
it
shouldn't
be
bleached
by
a
toss
bleacher,
because
our
ta's
lecture
that
removes
or
zeros
the
first
three
bits
of
the
core
point
will
make
these
three
bits
0
when
they
already
were
0
and
therefore
the
cold
point
survived.
So
it's
got
good
survivability
and
is
there
a
better
choice?
Well,
then,
maybe
because
this
bleaching
that
I
mentioned
causes
other
problems
to
other
traffic,
which
means
that
other
the
scps
currently
get
remarked
to
the
value
to
as
they
pass
through
the
insula.
A
This
is
kind
of
a
little
bit
of
a
nightmare
and
the
me
at
least
it
means,
and
something
good
to
measure.
So
I'm
going
to
talk
in
mahadji
about
some
measurements,
we've
done
to
try
and
start
off
figuring
out
what
happens
when
you
use
dscp
to
and
does
it
get
through.
I
can
tell
you
immediately.
The
answer
is
yes,
just
do
you
receive
other
traffic
with
dscp
too,
apart
from
the
ones
which
try
to
set
it?
B
A
B
A
G
Like
to
suggest
the
the
solution
that
I
propose
would
propose,
it
would
be
to
assume
that
too
is
the
right
answer
and
then
try
and
inventory
the
amount
of
harm.
That's
going
to
happen
from
things
that
don't
aren't
compatible
with
that
I
I'm.
Sorry,
I
didn't
see
that
this
draft
was
on
the
agenda
or
I
would
have
read
it,
because
this
is
something
that
this
is
actually
one
of
the
things
I
think
is
a
huge
missed
opportunity
for
the
IETF
and
one
of
the
things
that
would
buy
us
a
huge.
F
G
Of
additional
capacity,
we
have
all
of
these
things
in
our
pockets,
which
are
often
using
prime
data
to
update
when
you
should
be
updating
at
three
in
the
morning
and
the
total
load
just
on
that,
especially
in
the
southern
hemisphere,
is
massive
and
and
doing
it
right
would
make
a
huge
difference.
I'm
11,
minor
technical
question,
which
I
wasn't
paying
attention
to
and
I
would
have
read.
Is
there
a
spec
that
the
passive
open
loops
back
the
bit
the
marking?
Oh,
that.
A
A
G
Understand
that's
one
of
the
weird
applying
in
the
real
world,
but
but
the
fact
is,
you
have
to
you
have
to
start
somewhere.
This
is
a
multi-party
deployment
and
the
particular
concern
is,
for
instance,
for
uploading
software.
If
I
touch
the
Installer
I
wanted
to
do
it
in
the
foreground.
If
it
runs
on
a
timer
at
three
in
the
morning,
I
want
to
do
it
in
the
background
and
so
to
the
application
is
the
only
thing
that
has
a
scope
for
knowing
well
how
that
bit
should
be
set.
Yeah.
B
G
B
Either
the
TC
p.m.
chairs
here
Michael
Scharf,
no
okay,
because
Matt.
My
reaction
to
that
question
is:
if
I
passive
open,
we
mean
TCP
passive,
open,
that's
going
to
wind
up
as
a
tcp
em
topic.
If
we
mean
a
quick,
quick,
whatever
quick
is
going
is,
is
going
to
do,
is
kind
of
what
it's
is
going
to
wind
up
there.
So
how.
G
G
B
D
Me
I
could
have
been
I
mean.
Actually
you
should
know
better
than
I
do
but
I
mean
the
measurements
you
did
where
you
were
just
using
trace
road
and
you
get
if
you
would
get
back
icmp.
You
have.
The
information
you've
been
asking
for.
Unfortunately,
I
see
if
he
doesn't
work
very
well,
but
like
that's
what
you
want
yeah.
B
A
A
A
Older
me,
okay,
we
should
written
this
draft
quite
some
considerable
time
ago
and
something's
changed
significantly,
and
people
are
really
deploying
networks
that
can
carry
dscp
s
through
them.
There's
a
lot
of
work
to
get
that
fully
deployed
knows
a
good
opportunity
to
actually
write
this
draft
and
get
it
out
there.
So
please
comment
please
discussed
on
the
mailing
list
and.
B
Pending
discussion
and
map
RG,
we
need
to
sort
of
hold
needs
to
hold
a
finger
to
finger
to
the
wind.
That's
not
blowing
this
room
and
figure
out
what
the
right
milestone
date
is
I'm,
pretty
sure
base
and
discussion
for
right
now.
The
current
may
milestone
date,
for
this
is
is
wonderful,
is
wonderfully
premature,
I'm
inclined
to
go
for
it
to
go
for
September,
in
the
hope
that
another
meeting
cycle
will
help
what's
going
on
here.
B
B
B
A
Right
I
need
to
go
pee.
Animal
is
hang
on
a
minute
here,
while
David's
getting
things
set
up
and
this
documents
being
around
for
a
while.
So
we
will
hear
an
update
on
it
and
it's
one
of
the
pair
of
documents
so
we'll
consider
its
progress
with
respect
to
that,
but
we
got
sides
working
on
yeah,
okay.
This
is
where
you
find
the
slides
they're
here
and
if
you
click
on
this
and.
F
H
The
main
purpose
of
the
draft
was
to
catch
systemic
ecn,
layering
problems,
and
this
work
has
been
laced
with
iti,
Tripoli
and
3gpp
and
has
got
positive
responses.
All
of
this
is
being
updated
in
the
draft.
The
latest
updates,
I
think,
we're
with
relation
to
the
protocols,
goo
and
Geneva,
and
also
a
whole
bunch
of
other
protocols,
trill
VX,
lan,
and
so
on.
I
mean
basically,
the
mechanisms
of
feet
forward
and
up
and
feed
up
and
forward
they've
all
been
outlined
and
completed
in
this
draft.
H
The
last
set
of
comments
I
think
that
were
addressed,
one
was
about
a
master,
a
shirt,
that's
been
addressing
the
draft
and,
lastly,
it's
the
RFC
6040
update
xem
that
has
some
dependencies
to
this.
That
needs
to
be
welcomed.
That's
what
gori
was
referring
to
so
go
to
the
next
page.
Here,
I
think
I
got
that
draft
wrong.
It's
not
the
troll
draft,
but
it's
the
RFC
6014
xem.
That
should
be
there.
That
has
some
dependencies
on
it
and
penis
should
both
put
procedure.
The
working
group
class
altogether
are:
can
this
be
handled
separately?
A
A
Okay,
I
have
one
other
observation,
and
this
particular
draft,
then
that
John
presented
here
and
hasn't
really
changed
that
the
last
update
is
simply
to
reply
references
with
newer
references.
So
if
we
heart
reviewers
for
this
draft,
we
could
get
the
reviews
in
early
against
this
draft
prior
to
the
working
group
last
call
together.
Does
anybody
wish
to
volunteer
to
review
this.
B
B
I
B
A
I
I'm
not
really
much
to
say
I
I
just
didn't
get
around
to
updating
it,
that
there's
some
to
do
bits
in
it,
which
have
basically
just
got
to
give
specific
text
that
updates
each
of
the
other
PS.
Is
you
know
at
the
moment
it
just
says
a
general
statement
of
how
they
should
be
updated
and
I
was
asked
to
give
specific
text
and
I
haven't
done
that
do.