►
From YouTube: IETF102-MPTCP-20180716-1550
Description
MPTCP meeting session at IETF102
2018/07/16 1550
https://datatracker.ietf.org/meeting/102/proceedings/
C
D
E
A
B
B
B
B
A
B
Right
so
yes,
so
we've
been
working
on
the
Biss
for
a
while.
Now
we
are
in
working
boot
last
call
on
the
Bisson
factor.
In
theory
it's
completed,
but
that's
a
topic
that
we
we
will
come
back
to
and
yet
just
remind
you
that
the
the
best
document
is
to
update
the
existing
RFC.
It
is
on
the
standards
track
rather
than
the
experimental
track
and
that
it
will
obsolete
the
existing
protocol
682
for
because
of
various
changes
in
the
in
the
protocol
in
the
handshake,
etc,
which
mean
that
it
some
is
not
backwards
compatible.
B
We
have
some.
We
had
had
some
work
in
in
the
Charter
on
multi-party
simi
proxies
just
as
a
reminder
that
has
now
shifted
to
the
TCP
M
working
group,
where
there
is
a
working
group
item
on
the
zero
RTT
converter,
which
is
I,
think
pretty
nice.
It's
a
slightly
generalized
version
of
the
multipath
TCP
proxy
I'm.
So
that's
very
good
news.
Please
go
to
TCP
M
to
contribute
to
that
that
work.
The
things
they're
really
important.
B
A
B
A
H
H
G
It
was
basically
two
pieces
we
were.
We
had
some
references
to
experimental
drafts
and
RCS
in
our
68
24
base
draft.
So
we
removed
the
section
3.7
about
the
MP
TCP
experimental
option,
and
this
can
then
be
rather
specified
in
and
separate
RFC
and
the
second
item
is
we
moved
the
references
to
TCP
fast
open
and
how
it
would
interact
with
NP
TCP
to
the
appendix
also.
The
reason
is
because
the
gtfo
RFC
is
currently
only
an
experimental
RFC
at
not
a
standard
struck
RFC.
G
So
that's
why
the
decision
has
been
made
to
move
those
two
parts
either
remove
them
or
move
them
to
the
appendix
then
Phil
launched
the
working
group
last
call
for
the
base.
There
were
quite
a
few
minor
comments
about
editorial
issues
and
then
two
other
bigger
ones.
One
of
the
changes
that
went
into
the
best
version
is
that
we
added
an
echo
flag
into
the
address
option.
A
So
this
is
your
C
from
floor,
and
so
this
is
personal
opinion
that
this
is
interesting
point,
but
other
is
basically
announcement
crying
as
I
have
this
address,
and
then
it's
Sabha
to
utilize
this
address.
So
if
it's
Attica
send
back
to
the
house,
Heidi
address
but
cry
and
has
no
effect,
the
basic
client
doesn't
use
this
address.
So
I,
don't
know
this
could
read
severe
security
issue.
A
That's
what
I'm
thinking
this
because
know
if,
if
a
Kaka
tried
to
utilize
this
information,
akka
attacker
tried
to
initiate
another
hello
and
enjoying
MP
join
to
the
other
s
to
the
Sabha
side,
but
it's
also
neat
also
fray
in
the
steering
rods
token
secret
key.
Otherwise
he
won't
make
it
snow.
I
from
my
point
of
view,
even
this
is
possible
attack,
but
after
that
attack
also
need
to
do
stealing
some
keys.
Otherwise
it
doesn't
work.
That's
my
understanding,
yeah.
G
Yes,
I
agree
like
the.
If
the
attacker
would
do
this
by
setting
the
echo
bit
here
to
one.
The
client
will
just
assume
that
the
above
at
address
has
been
received
by
the
server
which
is
okay.
If
it
got
lost
okay,
he
won't
be
retransmitting,
it
I,
don't
know
if
that's
a
big
deal
and
the
other
case
is
but
that
you
did
I
think
this
rubbish.
He
doesn't
sets
the
echo
bit
right.
G
Then
the
client
might
be
fooled
into
creating
a
sub
flow
to
this
address,
but
then
the
verification
would
fail
right.
So
yeah.
G
Another
scenario
that
was
changed
in
the
best
version
is,
if
you
nowadays,
you
want
to
announce
an
address
and
immediately
mark
it
as
a
backup
address.
What
you
could
previously
do.
Is
you
send
an
address
with
an
address
ID
X
in
this
case,
and
you
couldn't
basically
follow
this,
then,
with
an
MP
prior
to
downgrade
the
priority
to
back
up
because
the
MP
prior,
you
can
indicate
the
address
ID
and
say
this
one
should
be
backup.
G
Now,
when
the
v9
version
several
versions
ago,
off
the
best
draft,
we
removed
the
address
ID,
because
researchers
pointed
us
to
an
attack
that
would
be
possible
that
an
unperfect,
a
car
could
basically
mark
all
also
sub
flows
as
backup
and
basically
steer
all
that
all
the
traffic
on
his
own
path,
and
so
to
prevent
this
kind
of
an
attack,
we
decided
to
remove
the
address
ID
from
the
MP
prior.
So
now,
if
you
want
to
change
the
priority
of
a
sub
flow,
you
need
to
send
the
MP
prior
on
this
particular
sub
flow.
G
Now
that
our,
if
we
don't
have
the
address
ID
anymore,
you
can
go
to
the
next
slide.
Now
we
can't
do
this
above
anymore
right,
we
can't
say
anymore,
mark
this
particular
address
as
a
backup
right.
We
can't
do
what
what
the
above
line
anymore,
because
there's
no
more
errors,
ID
and
so
the
proposal
that
Phil
pointed
out
to
basically
you
revive
this
particular
scenario
again,
is
to
add
the
backup
bit
into
the
add
address
option
that
way.
G
A
Yeah,
that's
what
I
propose
yeah.
Otherwise,
in
order
to
make
backup
you
have
to
establish
a
connection
in
the
send
mp3-
oh
that's
very
cumbersome
from
my
point
of
view
and
then
it
doesn't
hard
security,
I!
Think
and
that's
why
I
think
this
and
this
this
we
don't
want
to
support
this
case.
We
should
have
this
feature.
That's
my
personal.
If
there
is
another
people
think
another
way,
this
I
agree.
G
B
If
you've
had
time
to
think
about
this
and
you
you
have
a
view
and
you're
in
if
yeah,
if
you
have
a
view
about
this
proposal,
you've
had
enough
time
to
think
about
it.
Maybe
you
could
just
put
your
hand
in
yet
so
I
just
see
if
we
get
a
feel
for
if
there's
enough
people
here
who
know
about
it,
apart
from
the
three
of
us
in
the
front,
does
anybody
else
thoughts
about
this
enough
to
have
a
view
about
it?.
B
B
That's
good
news
and
thank
thank
you
for
implementing
it
and
and
yeah
going
through
the
through
the
process,
which
is
which
won't
be
paying
for
it
released,
and
that's
obviously
important
so
that
people
can
experiment
with
it
and
and
test
it
out.
You
know
to
make
sure
that
we've
got
this
right
before
we
actually
launch
it
on
the
world.
I.
I
G
G
B
Any
more
comments
or
questions
this
point,
so
we've
had
we've
had
a
month
or
so
for
working
group
last
call
and
unfortunately
we
we
didn't
get
comments
apart
from
Yoshi
and
I,
so
we
really
would
like
some
more
eyes
on
this
document
from
people
who
are
potential
users
or
potential
implementers
or
people.
Who've
who've
been
involved
in
the
discussion
over
the
last
few
years.
On
this
document
you
know
I
know:
we've
had
we've
had
an
extensive
discussion
over
quite
a
long
period
on
all
the
bits
that
have
been
added.
B
So
that's
you
know
it's
not
as
ODEs
come
out
of
nowhere.
This
document,
it's
been
reviewed
and
discussed
a
lot
by
the
working
group,
but
really
to
meta
to
move.
This
could
complete.
The
working
group
call
last
call
properly,
and
so
we
can
send
it
to
the
ice
gee.
We
really
could
do
with
more
than
just
the
two
chairs
reading
it
so
I'm.
Basically
after
some
volunteers
we're
after
volunteers
to
read
this
and
make
some
comments,
you
don't
at
least
read
parts
of
it.
B
So
people
in
this
room,
you're,
obviously
very
keen
you've
made
it
to
an
end
of
a
long
corridor.
So
I'm
hoping
we
can
get
some
volunteers.
So
this
really
am
now
after
some
individuals
who
are
prepared
to
stick
their
hand
up
and
and
agree
to
review,
all
or
some
of
the
documents
over
the
next
few
weeks.
A
J
B
B
I
Hi,
so
this
is
a
presentation
on
MPCP
operation
in
5g,
on
behalf
of
my
coworkers
revision,
zero
of
the
draft
was
presented
in
the
last
meeting
and
the
feedback
included
the
following,
both
on
the
mailing
list
and
in
the
in
the
room.
First,
the
there
is
an
interest
for
NPT
p5
genes,
mainly,
but
the
proposal
should
be
more
concrete
and
also.
The
second
point
was
that
the
DMM
is
defining
session
continuity,
types
for
IP
addresses,
and
we
should
take
that
into
account
so
between
the
last
right.
I
Between
the
last
meeting
and
today
we
adapted
the
section
on
session
continuity,
and
it
is
now
using
the
session
continuity
types
as
input
to
the
mptp
scheduler.
It
is
possible
also
to
send
the
session
continuity
types
over
an
HTTP
signaling,
so
that
the
way
most
capable
can
also
use
this
information
just
to
clarify
with
the
history.
The
goal
is
to
session
Canadian
wagons
when
you
have
an
IP
address,
which
is
replaced
or
complemented
with
another
IP
address
by
the
network.
Know,
for
example,
when
you
move-
and
so
you
see
you
know,
alternative
details
after
that.
I
I
So
just
a
bit
of
background
on
decision
continuity
types,
so
they
are
defined
in
the
DMM
in
the
Vietnam
draft
on
demand,
mobility
for
IP
addresses
provided
by
by
mobile
networks,
so
we
will
be
using
5g.
There
are
some
letters
and
statements
and
four
types
are
defined.
So
we
have
fixed
session
lasting
non
persistent
and
rescuin
replacements,
so
our
fixed
IP
address
is
valid
for
a
long
time.
I
So,
just
now
this
slide
is
on
the
slide
on
the
session
continuity
type.
It
described.
Basically,
it's
a
modification
of
the
of
the
draft
to
adapt
to
know
using
session
continuity
type
in
MPC
P,
so
in
SSE
mode
1,
which
is
the
5g
mode
with
a
fixed
anchor.
The
initial
IP
address
is
a
session
lasting
IP
address
or
fixed
IP
address,
and
it
can
be,
it
can
be
temporarily
complemented
by
a
non
persistent
IP
address.
I
I
The
initial
non-persistent
IP
address
is
replaced
with
another
non
persistent
IP
address
after
a
short
break,
the
session
continuity
type,
for
example,
in
this
case
now,
I
can
indicate
to
the
MP
TCP
stack
that
a
break
before
make
transition
is
expected.
So
that
may
be
useful.
That
may
or
may
not
be
useful
really.
I
Well,
the
other
one
replaces
the
first
one,
so
it
should
be
a
backup
as
well
so
and
just
to
illustrate
the
type
of
info
that
we
we
need
when
we
think
we
need.
We
need
a
session
continuity
type
on
the
IP
address.
The
new
IP
address
the
fact
also
that
this
IP
address
is
relating
to
an
origin
or
IP
address,
which
it
replaces
and,
of
course
the
you
have
the
continuity
type
of
the
original
IP
address.
I
So
these
three
information
elements
are
good
input
which
can
basically,
which
can
make
it
possible
to
make
a
decision
and
execute
one
of
these
one
of
these
actions
in
SS
symmetry,
which
is
a
distributed
encoder
case
with
the
make
before
break
the
initial
graceful
replacement.
Ip
address
is
replaced
with
another
graceful
replacement
address
and
there
is
an
overlap
period.
I
So
it's
important
for
us
to
learn
about
it
so,
and
this
type
is
also
also
enables
performing
a
graceful
migration
of
traffic
from
the
first
to
the
second
IP
address,
and
maybe,
as
importantly,
you
know,
it
enables
releasing
the
resources
on
the
first
IP
addresses,
so
that
the
network
can
actually
use
them
for
something
else.
The
remote
pier
can
implement
this
behavior
as
well.
I
If
the
type
is
transmitted
over
mptp,
signaling
and
also
the
type
is
also
useful
to
track
and
update
the
backup
address,
so
I
mean
as
a
conclusion,
so
the
session
continuity
type,
we
think,
is
a
useful
input
to
operate.
In
fact,
G
scenarios
you
know
just
to
basically
to
track
what's
happening
to
this,
to
this
IP
addresses,
for
example,
to
the
cellular
IP
addresses,
which
is
associated
with
the
session
continuity
mode,
and
it
enables
distinguishing
between
cases
and
also
it
offers
flexibility
for
future
evolution.
I
So
now
a
few
words
on
the
other
section,
so
dual
connectivity
and
immediately
some
co-authors,
so
that's
basically
Ken
and
his
colleagues
at
an
ICT
I've
analyzed.
The
3gpp
see
especially
specification
sorry
with
the
focus
on
the
issue
of
MPCP
of
virtual
connectivity
and
as
a
result
about
paper.
It's
an
empirical
analysis
framework
for
multipath
communications.
I
I
What
are
the
next
steps
so
about
the
session
connectivity?
The
continuity
section?
We
would
like
to
collect
an
address
feedback
from
the
PCB
working
group
on
the
recent
change
to
use
the
DMM
defined
session
continuity
type
well,
so
we
also
sent
the
notice
to
the
DMM
group
in
case
they
have
some
comments,
then
also
we
plan
to
do
some
experimentation
using
simulation
or
emulation
because
of
the
lack
of
availability
of
G
of
Phi
G
system
now
about
dual
connectivity
about
the
the
section
on
the
all
connectivity.
I
Our
next
step
is
to
prototype
MPC,
P
overall
connectivity
and
to
identify
impacts
on
MPT,
CP
and
3gpp
standards.
So
now
about
working
with
eruptions.
I
know
it's
a
bit
early,
but
it's
because
there
will
be
some
discussion
on
the
new
charter
on
possible
new
work.
So
maybe
it's
still
a
good
moment
to
to
present
this
to
say
that
you
know
would
like
to
know
if
the
working
group
is
interested
in
discussing
or
adopting
this
work
at
some
point,
the
topic
could
be
really
MPTP
adaptation
and
optimization
for
4G.
I
B
I
I
I
think
that
maybe
first
there
locally,
the
the
stack
then
PHP
stack
will
very
likely
need
to.
You
know
be
aware
of
this
session,
cata
media
type
and
the
relationship
between
an
address
and
its
origin.
Ip
address.
We
know
this,
but
we
think
that
in
s,
asymmetry
to
enable
some
optimization
to
release
early
the
first,
you
know
to
release
resources
associated
with
the
first
IP
address
as
early
as
possible.
I
It
would
be
beneficial
to
send
this
information
over
the
year,
but
basically
to
send
it
over
the
protocol.
For
example,
that
could
be
a
message
saying
when
it's
it's
a
bit
similar
to
the
backup
bit
you
know
like,
but
it
could
be
a
separate
message
or
a
part
of
the
add
address,
but
it
could
eat
something
that
as
the
session
continuity
type
and
if
it's
a
separate
message
we'll
need
the
index
to
the
to
the
address.
B
B
I
B
B
J
I
Think
you
know
I
think
that
there
may
be
the
the
thing.
Is
it's
a
twin
a
it
enables
optimizing
when
you
are
in
SSC
mode,
three
to
shoot,
enable
optimization
of
the
transition
and
also
it
should
enable
realism.
The
first
I
mean
resources
of
the
first
IP
address
early,
because
you
just
make
a
one
small
point.
It
seems
to
me
the
SSC
mode.
3
was
kind
of
defined
for
applications
like
a
browser
or
something
like
that.
So
you
open
you
open
a
small
or
short
session.
I
You
know
here's
the
first
IP
address,
then
we
change
you.
The
new
IP
address
you
still
keep
using
for
the
existing
session,
the
first
IP
address.
But
when
you
open
a
new
session
in
your
IP
session,
you
use
the
new
IP
address
and
since
that's
relatively
short,
a
positions
when
you
stop
using
the
first
IP
address
and
you
move
on.
I
But
here,
if
you
are
in
PCP-
and
you
want
to
use
all
your
links
as
optimally
as
possible,
you
will
keep
using
that
first
IP
address
as
long
as
you
can,
and
maybe
the
it's
not
really
as
the
network
operator
would
prefer
you
to
use
that
IP
address,
because
it
provided
a
new
one
that
it
wants
you
to
use.
So
some
form
of
you
see
it.
It's
not
like
major
thing.
It's
not
like.
It
won't
work.
If
you
don't
do
it,
but
you
won't
realize
the
first.
J
I
Okay,
I
mean
I'm
any
some
of
your
points.
I'm
sorry
about
this
I
need
to
think
more
but
yeah.
My
understanding
is
that
the
both
sides
would
need
to
know
that
this
is.
We
are
on
SSI
mode,
three,
so
that
you
actually
stop
sending
packets
on
on
the
first
phone
on
the
path
using
the
first
IP
address,
so
you
need
to
send
to
stop
using
packet
to
stop
sending
the
packets.
Otherwise,
you
need
to
wait
for
the
network
to
just
bring
down
the
IP
address.
I.
A
K
My
way,
valleys,
I'm
working
our
auto
body
in
the
fact,
is
that
nuts
for
our
Stan
Stan
Stan
Stan
for
man
standing.
The
fact
is
such
an
continuity.
Yes,
is
working
on
the
IP
address
that
you
say:
Abbott
resin
layer,
because
you
para
you're
talking
about
MPT,
CPO
I,
do
not
know
you
when
you
doin
this
work,
maybe
you
introduce
some
deepen.
A
pataiiputra
did
not
appeal.
K
I
just
have,
but
is
okay,
but
Adams,
and
confuse
that
for
our
selling
continuity
will
only
for
Apple
layer,
then
for
all
the
TCP
layer
or
UDP
layer
and
what
or
the
other
layer
or
quickly
who
can
support.
But
at
you
you're
not
only
talking
about
MP
TCP,
my
concerns,
maybe
you
should've,
proposed
your
proposal
in
the
in
IP.
K
I
You
know,
to
a
certain
extent
the
PGP
understanding
its
acting
and
it's
between
the
application
and
the
IP
layer,
so
it's
actually
acting
as
a
user
of
the
IP
layer.
So
it's
it.
It
won't
try
to
use
the
different
IP
addresses,
typically
as
funny
as
it
can
and
but
that
causes
it
I
mean
to
me,
I
mean
I.
I
Think
we
identified
the
possible
issue
in
asesino
3
that
you
keep
using
all
the
IP
addresses
as
much
as
you
can,
which
a
normal
application
like
a
browser.
When
will
not
do
because
a
browser
knows
I
mean
in
will
be
associated
with
assessing
node
3,
so
any
application
which
is
associated
with
as
a
single
tree,
we
all
actually
I
mean
SS
me
ma
tree-
is
designed
for
a
certain
type
of
applications
and
MPTP
does
not
behave
like
this
type
of
applications.
K
Man,
essentially
the
app
your
proposed
3
tab
or
fourth
tab.
Oh
I've,
just
I've
read
this
is
very
helpful
for
the
SE
model,
one
two
three,
but
massaging
my
concerns
that
this
type
of
use
shape
in
general
for
all
the
plant
water
protocol
is
not
a
related
for
UDP,
quick
or
TCP
or
CTP
its
general
of
habit.
Tab
I
mean
if
your
proposal
from
here
it
means
it
only
use
the
per
MPCP
paper.
Ue
o
is
used
by
all
kind
of
UE.
This
my
concern.
A
A
K
K
Tell
Marta,
for
example,
you
can
access
Wi-Fi
and
the
faculty
in
selectivity
can
use
an
TCP
to
submit
the
the
one,
the
IP
fellow
from
different
access
in
such
a
case,
the
enhance
the
enhancement,
the
video
access
and
increase
the
done
in
the
data
also
people,
also
in
in
the
last
acetyl
meeting.
People
also
proposed,
and
please
eternity
to
enhancement
the
communication
in
two,
because
there
are
two
paths,
so
one
man
passes,
failure
and
that
have
can
can
be
survived
with
the
communication.
K
G
I
have
one
question
with
the
dual
connectivity
that
you
mentioned:
where,
basically
you
have
you
mention
it
is
in
the
draft
quickly.
Look.
You
have
two
cases
where
they
mention
where
your
ones
is
where
you
have
different
IP
addresses
on
the
two
bearers.
The
other
case
is
where
you
have
to
say
my
piano
is
under
both
arrows
right.
G
So,
if
you
have
to
say
my
PRS
on
both
barrels,
you
basically
when
you're
sending
you
need
to
kind
of
indicate
which
bearer
the
subfloor
belongs
to
the
question
is:
how
can
the
base
station
know
to
which
bear
or
discipler
belongs
to?
Because
the
base
station
is
not
really
aware
of
MB
TCP
port,
any
kind
of
signaling
or
I,
don't
know.
I
What
I
know
about
this-
and
it's
not
I-
can
know
a
lot
about
this,
but
my
understanding
is
at
lower
layer,
it's
really
at
the
layer
tool
that
between
the
UE
and
the
base
station,
but
these
two
links
are
basically,
you
have
different
PD
new
sessions.
Usually
you
should
have
different
musicians,
I
think
not.
I
This
case
it
might
be
transparent,
it
may
be
transparent
to
you
know
when
it's
transparent
to
the
UE.
Usually
it's
one
right
now
so
here
you
know
my
understanding
that
Canyon
disconnects
they
they
are.
Then
they
need
to
define
a
way
to
expose
this,
as
you
are
saying,
the
application
layer
and
that
that
is
not
something
that
is
done
right
now.
So
yeah.
G
And
it
it
can
be
exposed
on
the
device
side,
but
the
tricky
part
is
then
to
expose
the
same
thing
on
the
base
station
side,
because
when
the
return
traffic
is
coming
back
right,
it
needs
to
know
over
which
bearer
to
send
which
sub
flow
sort.
It
becomes
a
little
bit
more
complicated
on
the
downstream
link,
I
think
if
there's
no
specific
signaling
or
if
we
want
to
avoid
them
to
do
deep
packet
inspection.
That's.
G
K
Yeah,
for
that,
your
connection
is
a
layer
2
technology
in
the
faculty
system,
in
such
case
only
full
of
one
IP
address
for
the
one
up
he
flow
DiSpirito
to
that
the
packet
have
a
flow
path
to
put
even
the
connection
this
that
your
connection,
so
this
layer
to
technology
that
doesn't
like
at
appears
Kaname
solicit
that
they,
the
the
base
station
and
not
a
know
any
happy.
Fellow
human.
A
Hello,
this
is
your
serum
for
yeah,
it's
interesting,
and
so
you
said
you
have
three
SSC
mode
and
then,
in
my
understanding,
if
you
see
mode
is
determined
inside
I
was
in
attack
my
cry.
A
Mp
TCP
application
can
only
know
okay
I
mean
if
they
see
more
diversity
more
to.
That
means,
if
you
know
it's,
a
MPD's
application
had
started
and
then
by
using
some
special
API,
and
it
can
know
the
current
SEC
mode
and
the
it
said.
If
you
are
SSC
mode
one,
then
there
is
not
so
many
single
student
for
this
MP
TCP
application.
I
But
it's
it's
only
for
one
IP
address.
So
basically,
if
you
are
using
a
system
mode
one
and
you
are
Wi-Fi
also,
you
know
it's
basically
like
4G,
but
it's
just
the
application
is
associated
with
a
system
with
two
or
three
that
basically
you
you
gain
something.
You
know
you
because
you
have
shorter
path,
but
you
have
to
work
more
for
it.
So
you
so.
A
I
A
K
K
A
I
E
A
Let's
say
it
cross
the
media,
but
I
probably
I
see
some
still
some
energy
in
the
working
group.
So
vara
I've
tried
to
figure
it
out,
but
we
can
do
in
short
time
and
wrong
town
and
then
some
previously
we
see
lots
of
interesting
presentation
in
the
previous
meeting.
So
I
tried
to
summarize
some
of
the
nice
proposal
on
the
you
know,
working
group
meeting
on
the
on
the
it
a
mailing
list
as
well
and
so
I
have
listed
several
topics
here.
A
For
example,
api's
we
have
several,
you
know,
presentations
about
api's
and
then
also
about
some
feature
extension.
We
can
have
some
security
feature
experiment
by
using
Tiaras,
or
we
see
some
proposal
and
then
these
days
no
tcp
ink
has
been
deployed
and
then
and
previously
original.
If
an
sctp
bug
STP
ink
working
group
has
me
establish
one
of
the
application
for
se,
tippy
ink
working
group
is
to
support
and
participate.
A
That's
one
way
and
also
we
can
think
about
packet
scheduling
our
quoi
spec
does
not
talk
about
packet,
scheduling
and
right
now
it's
live
to
the
implementation.
But
if
someone
come
up
with
very
nice
packet
scheduling
mechanism,
it
might
be
better
to
standardize
it
that's
another
possible
idea,
and
then
there
are
some
suggestion
that
we
should
interact
with
working
groups.
More
and
DMM
is
one
of
the
good
example
you
know
by
you
know
we
might
be.
You
know,
find
a
nice
way
to
utilize.
A
Mp
TCP
on
5g
network
may
be
promising
idea
and
block
ops,
there's
some
segmented
routing
proposals.
Maybe
they
might
utilize
MP,
TCP
and
everything
working
group.
We
can
utilize
some
Marsh
pass
information
and
then
we
might
be
able
to
utilize
MP
TCP
here
and
there
are
several.
There
may
be
several
way
to
collaborate
each
other,
but
I
would
like
to
see
you
know
what
we
can
do,
especially
in
short
term
and
then
also
we
can
see
something
in
long
term.
A
F
A
Is
no
time
right,
but
right
now
we
don't
have
any
working
group
item
and
then,
if
we
don't
have
working
right
then-
and
there
is
no
strong
reason
have
a
meeting-
probably
so
we
might
skip
one
or
two
meetings
until
it
all
possible.
Promising
working
group
item
has
proposed
yeah.
We
need
to
think
about
it.
If
there
is
very
promising
idea
and
they
win,
you
know
there
are
so
many
people
start
feeling
this.
It's
a
good
idea
to
discuss
and
then
we
can
have
a
meeting
and
all
right.
A
L
B
I
mean
that's
a
good
question,
so
this
is
this
was
a
collection
of
ideas
that
people
have
raised
about
topics
that
people
might
be
interested
in
working
on.
So
in
order
to
threat
to
become
a
working
group
item,
it
basically
needs
a
community
of
interest,
so
it
needs
more
than
one
person
to
say
I'm
interested
in
this
topic,
so
you
need
several
people
to
work
on
it.
At
least
you
know
you
need
people
to
work
on
it,
but
you
also
need
people
who
say
if
somebody
does
work
on
it,
I
would
use
it.
B
So
it
needs
a
community
of
people
rather
than
just
being
one
individual
who's
interested
in
the
topic
and
will
be
the
only
person
who
would
use
it
so
that
that's
basically
the
selection
criteria.
So
there
has
to
be
a
group
of
people
who
who
want
the
work,
want
the
output
and
they're
prepared
to
work
on
it.
They
don't
have
to
be
completely
overlap,
doesn't
have
to
be
exactly
the
same
people
in
each
set
yeah.
If.
L
J
Meuk
live
in
respondent
lady.
So
if
this
working
group
has
reached
the
end
of
the
Charter,
then
what
we
usually
do
with
close
working
group
right,
I'm,
not
saying
that
you
cannot
like
reach
order
and
figure
out
new
work
items.
But
if
there's
like
nothing
to
work
on,
the
working
group
will
be
closed.
That
does
not
mean
that
you
cannot
work
on
NP,
t
CP
anymore,
I.
J
Think
before
closing
the
working
group
I
would
like
to
talk
to
the
T
CPM
working
group
and
see
if
they
can
reach
other
and
explicitly
include
any
maintenance
work
on
em
V
TCP
in
their
charter.
Because
then
you
also
have
like
you
have
one
working
group.
What
the
whole
key
CP
community
can
actually
talk,
which
I
think
would
be
very
valuable,
yeah
and
like,
and
even
if
we
close
a
working
group
we
can
so
leave
like
mainly,
must
open.
So
you
can
have
further
discussions.
There.
J
But
I
mean
like
even
if
we
integrated
into
TCP
M,
if,
like
a
lot
of
new
work,
shows
up,
there's
also
a
way
to
whatever
T
CPM
could
get
two
sessions
per
ITF
and
have
one
more
focused
on
MBT
CP
and
one
more
focus.
Another
tcp
question:
I
mean
that's
kind
of
what
we
do
as
a
transport
area
or
can
group
a
transport
working
group,
for
example,
that
they
very
often
have
like
one
sessions.
J
G
Christophe
Apple
I,
think,
what's
probably
going
on
at
the
moon,
is
like
people
are
busy,
deploying
it
and
finishing
up
use
cases
that
already
have
been
identified
in
the
past
and
pushing
to
get
MP
TCP
out
there
in
the
implementations
and
so
on.
So
that's
maybe
why
working
like
the
activity
in
the
working
world
has
calmed
down
a
little
bit
and
also
feel
like?
G
Sometimes
them
please
be
working
group
is
highly
location,
dependent
where
the
meeting
is
it's
more
or
less
activity,
because
the
people
who
are
working
on
MPSP
are
mostly
centered
in
in
the
European
location
and
so
personally,
I
think
it
would
be
okay
to
like
wait
a
bit
and
see
where
it
goes,
because
people
are
busy
deploying
it
and
there's
a
lot
of
activity
going
on
in
mp
TCP.
Even
if
it's
not
visible
at
the
IETF,
there
are
billions
of
devices
out
there
that
are
using
it
every
day
and
they
are
more
coming.
A
G
G
J
G
I,
don't
think
they
are
going
to
be
well
kind
of
depends
personally.
I
have
a
long
list
of
things
that
I
would
like
to
change,
but
that's
not
going
to
happen
because
we
need
to
talk
to
servers
so
I,
don't
think,
there's
going
to
personally
I,
don't
think
there
are
gonna,
be
lots
of
huge
changes,
at
least
not
in
the
foreseeable
future.
A
long-term,
attractive,
well,
I
know:
I
have
a
list
of
things
that
are
wrong,
but
that's
a
different
story.
J
My
main
concern
with
that
group
is
that
a
lot
of
TCP
experts
don't
show
up
in
this
group
anymore,
and
we
have
seen
this
with
the
proxy
discussion
that
those
people
suddenly
then
show
up
again
and
have
different
opinions
and
so
on.
So
I
think
it
would
be
valuable
for
any
kind
of
MB
TCP
extensions
to
go
into
the
TCP
and
again
I.
Don't
want
to
stop
anybody
from
working
on
this
protocol
because
it's
a
good
protocol.
It's
really
nice
to
see
that
it's
taking
off.