►
From YouTube: IETF100-DMM-20171114-0930
Description
DMM meeting session at IETF100
2017/11/14 0930
https://datatracker.ietf.org/meeting/100/proceedings/
A
All
right
welcome
folks,
welcome
to
DMM
working
group
shrink
underbelly
from
Cisco
and
from
re
babaçu
Q.
So
before
we
start
the
meeting
few
things
note
well,
so
this
meeting
is
under
hundred
to
some
idea
of
rules
and
processes.
Please
read
this
note
carefully.
That's
one
thing
and
now
few
other
administrative
stuff,
blue
sheets.
A
A
This
time,
you're
very
pleased
to
announce
that
you
know
you're
announcing
any
award,
which
is
the
note
taker
of
what
yeah
so
the
guys
who
send
the
slides
the
later
you
know
after
the
deadline
right,
so
you
have
four
contenders.
One
is
more
color
wish
and
few
are
this
and
in
the
winner
is
mark
elevation.
Welcome.
They
need
to
take
the
notes
this
time.
B
A
A
Let's
go
the
status
of
each
of
the
working
room
documents.
You
know
one
is
we'll
start
with
the
on-demand
mobility
document.
I'd
believe
Danny
Danny
is
good
you're,
going
to
give
a
presentation
on
that
like
at
least
you're,
going
to
talk
on
the
on-demand
mobility
right
you're,
going
to
give
a
status
update.
A
Yes
right
so
I
think
that's,
but
a
few
things
I
think
the
last
update
was
on
July
30th,
so
we
didn't
see
a
new
version
but
I
believe
you
at
least
you
believe
you
know
all
the
issues
have
been
addressed
right,
but
I
think
you
know
what
is
spending
is
I
know
that
being
started.
Writing
the
the
shepherd
note,
but
we
need
we
didn't
do
the
chair
review
yet
so
I
need
to
do.
He
needs
to
do
that.
So
so
there
is
something
you
know
we
need
to
close.
A
We
apologize
for
the
delay
yeah,
we
apologize
for
the
delay,
but
we
will
take
care
of
that.
That's
one
thing
and
also
we
need
few
more
reviews.
I
know
you
are
quite
active
and
you
try
to
address
all
the
issues,
but
you
need
few
more
videos,
okay,
so
that
we
can
make
sure
the
document
is
in
good
shape
before
before
oratory,
but
in
general
I
think
the
goal
is
to
really
you
know:
hit
working
your
glass
called
by
IKEA
101
next
or
diem
deployment
models.
A
So
this
was
reading
sitting
idle,
but
let
us
seal
to
cover-
and
you
know
it's-
it's
been
quite
active
on
that-
and
he
also
got
some
review
feedback
and
seal
has
been
addressing
this
right.
I
think
few
other
things
that
we
have
identified
based
on
you
know
some
new
directions
and
based
on
some
new
interest
will
possibly
you
know,
we
may
update,
include
few
additional
deployment
models
but
I
think
offline.
Yesterday
we
had
some
discussions.
A
And
overall,
that
needs
to
be
captured.
I
think
that
you
know
based
on
the
progress
of
that,
based
on
the
feedback.
Again,
you
know
we
want
to
you
know.
Hopefully
you
know
by
101,
even
deployment
models.
I
think
we
want
to
I.
Think
we're
still.
You
know
funnel
you
know
here
intentionally.
We
are
slightly
delaying
it
because
we
want
to
the
PC
document,
because
if
PC
was
going
through
a
lot
of
changes,
we
didn't
want
to
just
rush
it
through.
You
know,
because
there's
some
level
of
dependence
it
in
the
documents.
A
You
want
some
consistency,
terminology
descender,
so
we
slightly,
we
I,
think
reasonably
desert
wind
is
in
a
good
shape,
but
but
we
were
slightly
but
but
it's
on
time
now
going
this
CPD
pfpc
documents
again
raising
your
force
from
subtle
song,
Lyle
and
Danny
and
others
and
overall
this
way
lot
of
changes,
but
I
think
today
we're
going
to
get
some
status
update
from
Marc
wonky
on
this,
and
this
I
think
this
is
a.
We
need
to
work
out
a
strategy
because
it's
a
huge
document.
A
You
need
to
have
some
plan
as
how
we
plan
to
get
reviewers
assigned,
because
you
know
it's.
You
know
anything
document
of
this
size
to
get
some
quality
reviews.
It's
not
going
to
be
easy,
so
we
need
to
you
know
it's
racial
and
eighty
cell
and
I
think
also
request
few
people
to
really
spend
some
good
time
and
review
the
document.
C
So
try
to
get
a
search,
trishna
so
try
to
get
at
least
three
good
reviews
from
the
working
group.
So
I'll
do
one
I'll
get
probably
one
from
the
inter
so
like
thats,
at
least
five
before
it
goes
through.
So
but
not
like.
You
know,
spelling
mistakes
and
stuff
like
that.
Like
some
decent
reviews
like,
then
we
can
go
forward
with
it.
A
A
And
this
week
is
mobility,
anchoring
I.
Think
thanks,
adios
I
know
you
provided
some
reviews
and
some
point.
You
are
not
comfortable
with
documents
right.
It
is
the
e
toriel
aspects
right
both
technical
and
editorial.
Now
you
know
I
think
I
just
want
hear
like
what
do
you
think
whether
the
only
good
shape
Park
where
we
stand
with
this
document
so.
D
Carbonaro
so
I
I
checked
yesterday
the
latest
version
and
I
check
how
my
comments
were
address
and
I
rationally
happy
with
with
that,
so
I
ever
even
really
read
again.
The
whole
document
just
focus
on
day
on
the
deef
and
my
thing
is-
is
much
better
than
the
previous
version
that
they
review
so
I
think
it
can
move
along,
but.
A
A
I
think
we'll
ask
that
as
part
of
the
his
object,
so
you
you
after
you
update,
maybe
we
can
ask
for
some
so
Anthony.
That
is
one
thing
and
next
is
I.
Think
this
is
the
maintenance
documents.
Let's
go
this
list.
I
think
you
know.
First,
one
is
the
MRI
ID
I
think
this
was
you
know,
sitting
there
for
a
long
time
and
I
think
really
Suresh
I
had
to
take
a
lot
of
heat
on
this
and
doc
Suresh.
You
know
thanks
for
all
your
forestry
marketing,
bringing
this
to
a
closure.
A
C
Suresh
krisshnan,
so
charlie
posted
the
draft
like
just
before
the
meeting
so
I'm
just
gonna,
make
sure,
like
you
know
he
just
included
you're
deaf,
but
so,
but
probably
there's
something
more.
But
I
can
I'll
take
a
look
at
it
and
make
sure
like
you
know
how
to
stuff,
but
I
the
first
thought:
I,
don't
think
it
will
address
everything
because,
like
you
know,
there's
something
you
needed
doing
something
Charlie
needed
to
so
only
differs
the
one
like
you
propose,
so
that's
I
think
we
can
solve.
A
Thanks
thanks
again
Suresh
I
know
this
is
this
game
stirred
some
reports,
so
I
think
I
think
this
will
should
be
able
to
closure
and
also
few
other
good
point.
Good
things
like
the
remembering
draft
is
now
published,
its
RFC,
8191
and
I.
Think
congratulations
to
authors
right
and
that's
one
thing
and
same
with
Mac.
You
know
configuration
parameters,
8127
right.
A
This
is
good
this
thing,
but
there
was
one
errata
that
was
noted
by
Verizon
and
I
think
it
will
be
posted
very
soon,
but
very
minor
thing,
but
I
think
some
clarification
is
needed,
so
8127
will
have
that
discussion
and
one
more
thing
the
multihoming
draft
is
also
up
put
so
I
think
ad
approve
it
recently
and
it's
in
the
RSC
attack.
You.
C
A
A
Work
has
happening,
I
think
the
it's
from
Sprint
like
mommy's,
you
know
they're
talked
about,
but
in
general
that's,
some
good
amount
of
courts
are
going
into
that
and
next
we
move
on
to
some
new
topic
around
mobile
user,
plain
motivation,
goals
we'll
have
Marco
presently
and
later
Satoru
son
will
present
this
track
and
some
some
good
amount
of
discussions
you
know,
went
through
on
sm64
mobile
user,
plain
document.
I.
Think
we'll
probably
not
pick
that
up.
A
You
know
as
for
adoption
and
the
other
documents
or
the
anchoring
Anthony
change
is
going
to
give
an
update
on
that
and
Danny
was
going
to
talk
about
the
socket
extensions
and
seeing
on
the
deployment,
considerations
and
fewer
the
new
topics
they
got
or
the
topics
that
have
been
discussed
in
the
past,
but
I
mean
we're
going
to
revisit
them.
What
is
the
VM
solutions
using
by
Carlos?
A
That's
one
thing
and
mobility
capability,
negotiation
by
cheongha,
hack
and
and
one
Wi-Fi
uses
for
you
know,
Devon
protocol
well,
Wi-Fi
uses
vibration
and,
lastly,
I
think
needs
to
be
extensions
on
demand
or
again,
I
think
this
is
also
an
extreme.
You
can
change
order,
but
that's
that's
not
like
that.
I
think.
Next
we
will
go
to
the
first
topic
agenda
topic,
which
is
Marcus
FPC.
B
F
Since
last
ITF
from
many
meetings
site
meanings,
we
had
at
ITF
ninety
nine
lessons
learned
feedback
receive
we
publish
in
between
last
and
this
ITF
for
eight
and
nine,
so
they
are
out
and
ready
for
review.
So
between
last
and
this
ITF,
we
had
almost
weekly
be
sometimes
be
weekly
calls
scheduled
allowed
and
we
came
together
to
really
improve
the
drought.
Just
to
add
to
what
we
said
before.
The
draft
didn't
receive
changes,
content
wise.
We
try
to
improve
the
description
of
the
information
model.
F
F
So
we
revised
mainly
the
core
document
structure.
Core
document
structure
means
the
core
part
introduction
how
to
read
the
information
model
and
the
operational
part,
and
in
the
back
we
find
the
young
descriptions
we
added
to
ease
readability,
a
section
on
model
conventions,
as
well
as
on
the
notation
that
we
agreed
on
and
thought
it's
very
valuable
to
read
and
understand
information
model
and
ease
of
final
implementation.
F
According
to
this
model,
we
improved
the
model
and
that
components
the
scriptures
and
we
aligned
it
throughout
the
he
meant
to
make
it
consistent
from
the
beginning
to
the
end.
In
the
end,
thanks
to
some
improvements
in
the
models
itself,
we
could
reduce
the
document
size.
Also,
we
removed
some
of
the
messaging
in
the
operational
part,
because
existing
RFC's
can
be
used
to
accomplish
that.
For
example,
cloning
of
templates,
that's
something
we
removed
and
the
most
fundamental
parts
are
still
and
yeah
we're
nine
is
out,
and
we
are
happy
to
receive
your
comments.
F
So
we
treat
only
abstract
aeroplane
nodes
here
and
the
information
data
models
have
pointers
to
physical
resources.
So
each
data
plane
no
needs
to
be
realized
in
the
physical
infrastructure
by
viscous,
which
is
software
switches
whatever.
So
here
the
model
is
flexible,
but
this
figure
just
shows
the
mapping
on
agent
level
from
the
abstract,
adapting
node
to
one
or
multiple
physical
resources
representing
the
data
plane.
F
All
right,
this
is,
after
some
forward-backward
our
current
match
and
conclusion
on
the
most
suitable
course
structure
for
the
information
and
data
model.
So
all
the
sub
structures
which
we
configure
and
modify
throughout
a
lifecycle
of
a
registration
are
under
mobility,
so
this
is
five
structures
that
we
define
in
the
graph
and
treat
by
means
of
the
configuration
which
is
the
paula
g,
representing
what
I
showed
before,
and
the
data
plane
out
association
and
supported
interfaces
in
between
a
general
policy
container
and
by
reference
to
this
policy,
we
have
configurable
policy
and
mobility
context.
F
Pre-Configured
characteristics
of
the
data
plane
and
changes
are
only
made
in
case
of
reconfiguration,
so
that's
more
than
long-term
term
track
off.
They
depend
on
configuration
and
mobility.
Context
is
mainly
what
this
group
is
about,
which
is
handling
the
data
plane
node
for
the
mobility
management
states
and
transactions.
F
Last
one
at
least,
we
have
a
substructure
about
monitors,
so
thinking
about
non-functional
requirements,
not
only
traffic
steering
is
important,
but
monitoring
the
data
plane,
for
example,
to
detect
traffic
for
a
mobile
node,
which
is
dormant
or
in
this
continuous
reception
mode,
and
also
to
collect
chargeable
events,
so
collect
traffic
information
to
accomplish
charging.
So
this
is
what
monitors
is
about.
F
A
few
details
about
the
information
model,
so
I
said
already.
We
added
a
sectional
notation
and
conventions
to
ease
understanding
the
information
model
and,
following
up
on
that
said,
it's
an
attribute
which
you
can
see
frequently
in
different
contexts,
either
applying
to
a
single
rule
or
to
a
whole
data
plane.
Node,
so
I
said
we
have
a
general
policy
structure
which
can
serve
as
references
and
individual
sub
structures
like
mobility
context,
for
example,
can
refer
to
an
existing
policy
and
apply
settings
to
add
or
change
existing
properties
per
the
actual
mobility
context.
F
So
mobility
context
is
the
main
substructure
for
mobility
management
states
on
the
data
plane
and
its
size.
Optimized
and
yeah
traffic
treatment
can
be
handled
by
two
options:
either
references
to
the
policy
substructure
and
apply
settings
to
it
or
by
an
embedded
rule.
So
we
define
the
format
of
an
embedded
rule
which
also
has
traffic
descriptor
and
traffic
treatment
actions,
which
is
then
embedded
to
the
rows.
So
here
a
client
is
flexible
in
how
the
entries
are
being
made
made
in
mobility
context
next
one.
F
So
that's
it
about
the
information
model.
We
also
defined
the
operational
and
messaging
part.
So
here
we
cleaned
up
a
bit
on.
Inconsistency
is
because
minor
modifications
and
restructuring
has
been
done
on
the
policy
and
information
model,
so
we
aligned
the
operations
and
messaging
with
the
changes
in
the
information
model
and
some
features
have
been
reduced,
such
as,
for
example,
cloning
of
policy
templates,
because
we
found
that
this
can
be
done
by
an
accomplished
by
existing
specifications
like,
for
example,
touch
mechanisms
of
HTTP.
F
So
last,
but
not
least,
that
also
helped
to
reduce
the
size
of
the
core
part
and
all
in
all,
we
came
up
with
a
shorter
document
and
with
an
improved
structure.
So
we
are
confident
that
it's
easier
to
read
and
understand
and
provide
feedback
to
authors,
and
we
are
happy
to
discuss
your
feedback
on
the
mailing
list.
Also.
We
would
like
to
go
through
some
examples.
First,
through
East
understanding
and
the
cross
check,
if
all
features
are
supported
by
a
PC
and
second
that's
to
be
discuss.
F
Yeah
going
forward,
I
mean
we're
at
version,
9
I
think
going
forward
shouldn't
take
too
long,
so
we
would
like
to
have
an
efficient
review
phase
and
we
need
to
clean
up
a
few
remaining
issues
in
the
back
of
the
document
that
we
have
in
mind,
it's
mainly
editorial
and
for
the
core
part
we
would
like
to
get
feedback
from
from
all
of
you
and
I
said,
we
should
move
the
discussion
on
the
clarity,
but
also
on
the
examples,
the
mailing
list
and
we're
happy
to
receive
your
feedback.
Thank
you.
Any
questions.
G
F
So
the
policy
is
main
structure
is
independent
of
whether
it's
configurable
mobility
context
policy.
So
it's
a
general
description
of
more
aggregated
policies
and
to
enforce
these
policies
either
the
configurable
policies
or
long-term
policy
can
reference
to
these
policies
and
enforce
our
mobility.
Context
can
also
refer
to
a
policy
in
this
poly,
a
policy
description
in
this
policy
tree
and
then
also
apply
settings
or
use
embedded
policies.
So
it's
just
serving
as
reference
for
either
configure
policy
or
mobility
context.
G
F
A
Anything
else,
thank
you.
Any
other
questions
comments,
so
I
think
so.
The
next
steps,
as
we
discussed
I,
think,
are
very
advised
right,
I
think
even
how
to
sign
up
view
so
as
you're
going
to
sign
up
view
or
a
yes
right,
okay,
so
marker.
When
should
we
start
that
process,
they
think
we
should
be
for
some
more
time
what
engage
folks
right
away
so.
F
I
think
for
the
for
the
reviews
at
any
time,
I
think
you
can
start
looking
at
the
core
part
now
and
since
we
things
and
we'll
definitely
go
for
a
word
and
chairman
with
the
minor
fixes.
So
if
you
have
comments
on
the
core
part
already
based
and
we're
in
nine,
we
take
this
into
account,
which
then
improves.
Of
course,
the
version
10
already
I.
F
The
of
course
I
mean
I
took
basically
lots
slide
from
last
time
and
there
is
progress
on
it.
So
the
first
implement
Jason's
print
work
on
is
the
multi
VPN
agent
support.
Multi
DPN
means
the
agent
runs
on
top
of
an
Essene
controller.
In
that
case,
it's
an
open-source
contribution
to
the
opendaylight
following
the
FPC
specification
and
also
since
some
time,
a
contribution
of
the
FPC
agent
on
us
using
the
honors
flow
store
as
being
progressed,
and
mainly
this
is
being
done
in
the
mobile
cord
project
for
Sprint
also
contributes.
F
Right,
the
next
presentation
is
actually
introducing
a
topic
and
I've
been
asked
to
have
a
general
bootstrapping
of
the
discussion
here.
So
this
is
neither
a
solution
or
positioning,
but
since
there
is
more
visibility
and
more
traction
on,
so
how
should
the
mobile
data
plane
look
like
in
the
future?
F
So
this
is
just
fostering
the
discussion
and
just
as
a
kind
of
teaser
I
mean
it's,
it's
clear
that
the
next-generation
mobile
will
have
to
support
diverse,
set
off
end
devices
from
IOT
from
mobile
internet
using
smartphones
and
mission-critical
kind
of
support.
Iot,
so
different
kind
of
devices
are
to
be
supported
with
different
requirements,
or
you
may
not
have
mobile
IOT
devices
but
stationary,
but
you
still
may
need
mechanisms
for
paging
not
to
locate
the
device
but
to
wake
it
up
from
this
continuous
reception,
more
than
policy
mode.
F
So
the
network
seems
to
be
required
to
or
to
be
tailored
to
serve
these
diverse
requirements.
And
so
the
question
mark
for
Carius
is
how
to
efficiently
support
these
different
devices
and
their
customers
from
the
vertical
industry
and
so
how
to
handle
the
heterogeneity,
how
to
manage
resources,
how
to
treat
costs
and
keep
them
at
a
reasonable
level
and
how
to
accomplish
network
management.
Some
of
the
answers
are
given
by
the
support
of
network
slicing.
F
So
here
on
the
data
plane,
people
were
asking
if
something
needs
to
be
done
compared
to
what
the
current
generation
of
mobile
communication
has.
If
we
go
to
the
next
slide,
we
just
see
of
airplane
concept,
work
in
today's
networks
or
the
word
packet
system,
which
is
considered
to
be.
The
fourth
generation,
has
basically
a
plane
control
plane
for
mobility,
management
and
policy
controller
and
then
on
the
data
plane,
its
nodes,
which
have
actually
both
the
control
plane
and
data
plane
color
period.
F
So,
according
to
work,
we
also
doing
it
in
DMM,
which
is
FPC
work.
We
separate
the
control
plane
from
the
data
plane,
which
is
already
a
good
step
towards
efficient
support
in
next
generation,
but
basically
it's
a
single
node,
the
packet
data
network
gateway
that
has
to
perform
many
tasks
for
mobile
terminals
subscription
and
associated
traffic,
so
the
IP
address
assigned
to
the
device
is
typically
anchored
at
this
central
gateway.
F
Tunnels
are
being
maintained
to
map
the
appear
dress
in
traffic
to
a
locator
where
the
mobile
device
is
currently
attached
and
then
is
being
forwarded.
But
non-functional
parts
are
compeers
well,
for
example,
for
mirroring
qsr
enforcement
and
monitoring
of
chargeable
events
and
tragedy
when
reporting
so
typically,
tunnels
are
being
used
in
between
gateways
and
mobile,
its
location
and
people
were
asking
if
this
is
appropriate
for
all
different
cases
that
need
to
be
supported
in
the
future.
F
So
this
is
basically
the
last
slide
whether
this
follow-up
slide,
because
we
say
okay,
as
today's
data
plane
sufficient
or
do
we
need
some
modifications.
Some
simplification
and
I
heard
the
word.
Simplification
on
the
data
plane
for
next-gen
very
often
so
these
are
actually
a
few
statements,
and
this
is
just
input
for
further
discussion
here
and
on
the
mailing
list,
and
there
are
a
couple
of
candidates
out
there
being
discussed
as
an
alternative
to
today's
hundred
base.
F
F
Not
associating
the
many
states
with
the
data
plane
like
GDP,
for
example,
having
money
many
channel
and
points
and
concatenated
forwarding
agent,
then
per
packet
overhead
counts,
so
that's
different
views
from
from
different
sites,
so
some
people
say
overhead
per
packet
doesn't
count
some
say
we
should
save
and
reduce
the
overhead
on
a
per
packet
and
if
we
do
encapsulation
in
different
tunnel
protocols,
so
the
per
packet
overhead.
This
is
pretty
big.
I
H
That
that's
interesting
No,
so
when
mobile
ipv6
was
being
done,
it
had,
you
might
say,
a
pretty
simple
data
plane
and
there
were
no
food
shortage
of
people
bringing
that
to
3gpp
and
in
fact
they
did
put
in
P
map
is
sort
of
an
alternative,
but
people
actually
putting
the
network's
out
there.
Their
idea
was
well.
This
is
supposed
to
be
the
evolved
package
system
APs
and
are
we
evolving
from
well?
H
We
know
a
lot
about
how
to
do
gtp,
and
so
the
evolution
basically
wanted
to
continue
the
use
of
gtp,
because
there's
a
lot
of
operational,
knobs
and
ways
of
making
things
work
under
gtp
that
didn't
directly
map
into
payment,
and
then
we
had
a
succession
of
P
med
documents
on
how
to
control
things
by
using
AP
ends
and
all
this
kind
of
stuff.
So
it
wasn't
like
the
IETF
wasn't
trying
to
satisfy
the
needs
of
3gpp
it's.
There
was
a
lot
of
work
on
that.
H
So
what
it
really
boils
down
to
to
me
is:
do
they
really
want
a
simplified
data?
Plane
I
mean
gtp
is
not
simple
and
worse
than
that,
gtp
C
is
drastically
not
simple.
So
I'm,
not
quite
sure,
I
mean
if
we
had
the
goal
of
making
a
simplified
data
plane
and
actually
even
had
a
provably,
most
simple
data
plane.
What
difference
would
that
make.
F
I,
don't
want
to
position
a
particular
solution
here,
but
various
aspects
need
to
be
into
account
and
the
slide
that
comes
after
this.
It's
not
the
last
one.
I
have
a
follow-up
slide,
which
actually
says
simplification
has
its
limits.
So
there
are
a
couple
of
things
in
mobile
operator
networks
that
need
to
be
considered,
including
charge
of
event,
monitoring,
qsr,
so
I.
Think
a
plain
IP
in
IP
network
may
provide
the
required
level
of
simplicity,
but
not
support
all
features,
so
I
think
we
should
not
per
se,
say
what
we
have
now.
F
I
think
there
are
people
pushing
for
study,
also
in
3gpp,
which
claims
to
be
the
5g
standardization
agreement.
So
I
think
we
can
do
some
work
to
at
least
prove
that
different
data
plane
with
different
characteristics,
simple
or
whatever-
is
more
suitable
for
next
generation,
and
then
we
have
good
arguments.
F
H
A
So,
just
to
maybe
a
comment
right
essentially
to
Charlie's
comment
that
you
know
why
you
know.
In
the
past
we
failed
with
respect
to
pushing
IETF.
You
know
protocols
or
whatever
I.
Think
one
thing
is,
if
you
look
at
you
know
the
gtp
had
you
know
from
2g
to
3G.
There
was
a
history
and
for
whatever
reasons
for
legacy,
hand
or
reasons
there
to
you,
know
that
protocols
alright.
Well,
there's
now,
with
the
separation
of
control
and
user
plain,
there
is
an
opportunity
for
to
IETF
to
really
you
know,
apply
some
newer
mechanisms.
A
Now
the
3G
we
can
keep
all
the
you
know,
signaling
complexity
in
the
in
the
either
whatever
entities
right,
but
at
least
use
a
plane.
Can
we
normalize
it?
Can
we
look
at
it
as
a
very
normalized
data
plane
where
it's
not
access
Pacific
right?
Can
it
be
like
a
used
for
a
common
for
access?
You
know
Wi-Fi,
why
doesn't
matter,
but
you
know
one
common
data
plane.
Where
will
we
get
rid
of
the
tunnels
and
we
leverage
all
the
new
capabilities
of
the
you
know?
Sr
v6
are
those
kind
of
new
stuff.
H
H
A
J
Certain
much
softer
Bank,
let
me
just
share
the
just
machine
status
in
the
CDP
P
I,
including
soft
Bank,
and
some
supporting
company
proposed
to
start
start.
A
study
work
about
your
ipv6,
only
solution
to
the
next-generation
use,
a
prayer,
but
the
rest
City
for
meeting
this
is
John
was
postponed
to
next
time.
So
next
city
for
meeting
will
be
held
on
end
of
the
this
month,
so
just
pretty
just
share
that
status
and
please
bring
your
opinion
to
the
CP
PCT
who,
if
you
are
delegate
how
about
contact.
Thank
you.
F
B
F
That's
something
many
many
sites
want
to
have
so
enable
anchor
allocation
and
steer
the
traffic
in
the
over
core
network
in
an
efficient
way
without
changing
the
IP
address
assigned
to
the
user
equipment,
so
that
kind
of
floating
anchor.
It's
something
to
be
investigated.
There's
another
comment:
yeah.
E
Dave
allen,
ericsson,
just
a
quick
observation
and
when
you
focus
on
trying
to
re
architect,
things
I'm
sort
of
noticing
that
a
lot
of
this
seems
to
only
focus
on
one
end
of
the
network,
which
means
like
you
really
need
to
deconstruct
all
of
the
things
you're
trying
to
achieve
here,
because
you're
trying
to
support
virtualization
right
now.
It's
anchored
mobility,
intercarrier
roaming
and
then,
when
you
go
into
the
ran,
there's
a
whole
lot
of
things
going
on
that
nobody
seems
to
be
paying
attention
to
such
as
dual
access,
coordinated,
multi-point,
X
to
handover,
etc.
E
F
Agree
I
mean
I
said
on
the
next
slide.
Maybe
we
have
the
chance
to
have
a
look
at
it
later.
Simplification
has
its
minute
its
limit,
so
I
think
from
the
motivation
to
change
it.
Aeroplanes.
We
also
should
take
into
account
where,
as
the
demand
according
two
or
three
two
BP
and
the
data
plane,
and
actually
not
it,
ignore
this,
and
but
what
would
you
sell
for
for
different
excesses?
I
mean
if
we
look
at
today,
we
have
gtp
you
or
just
mainly
use
for
cellular.
F
So
we
also
heard
about
the
need
to
actually
have
more
control
on
how
data
plane
is
being
routed
throughout
the
transport
network,
independent
of
the
device
IP
address
and
I
think
there
is
one
proposal
being
discussed
subsequently
from
control
and
optimize
operation
between
control
and
data
plane.
So
to
optimize,
really
the
costs
in
between
controlling
they
are
planning
to
set
up,
update
and
teardown
states
required
in
the
data
plane
during
mobility
transactions.
F
Just
a
conclusion,
so
still
many
non-functional
features
are
needed,
and
here
we
should
really
collaborate
with
other
organizations,
so
still
traffic
classification
and
queue
as
mapping
is
required.
So
in
gtp
this
is
being
done
once
on
the
Gateway
period
way
and
then
the
follow-up,
uncoordinated
gateways.
It
just
take
this
virtual
tunnel
identifier,
which
is
a
tunnel
and
point
ID
GDP
and
map
it
to
an
outgoing
tunnel
identifier.
So
they
don't
do
traffic
classification
again.
F
If
we
don't
do
this,
the
Gateway,
because
we
want
to
have
plain
IP
routing
the
most
simplistic
data
plane
we
can
imagine
somewhere.
You
need
to
classify,
for
example,
to
support
compatibility
with
legacy
radio.
So
today,
the
base
station
and
satellite
expects
a
tunnel
with
the
tunnel
endpoint
ID
coming
in,
which
is
then
mapped
with
the
outgoing
radio
error.
So
here
these
things
cannot
be
ignored
when
really
looking
at
the
end-to-end
aspects
of
the
data
plane
support
of
non-lp
data.
So
we
have
known
@p
data.
F
We
cannot
use
the
IP
based
transport
network,
so
that's
to
be
considered.
We
can
use
it,
but
something
needs
to
be
added
to
make
it
routable
and
yeah
means
for
charging
event
monitoring
the
polling
needs
to
be
done.
In
particular,
we
have
more
diverse
and
distributed
data
plane
note
so
where
to
collect
and
from
which
notes
to
report
the
monitored
events
and
support
of
dama
devices.
F
So
some
of
the
proposals
being
discussed
through
the
data
plane
as
either
in
the
fire
locator
separated
so
in
support
of
dormant
devices
which
are
in
discontinuous
reception
mode,
the
locator
may
be
outdated
so
how
to
treat
this
in
collaboration
with
a
separate
control
plane.
Last
but
not
least,
I
mean
compatibility
with
IP
before
transport
question
mark.
So
the
question
is:
if
we
still,
this
is
still
a
requirement
to
be
met
or
if
we
can
assume
that
people
6
on
the
data
plane
and
private
IP
address,
support.
F
Important
is
if
we
have
more
distributed
data
plane
for
some
proposals
being
discussed,
so
how
to
do
cures
in
terms
of
mirroring
right.
So
typically,
today
you
have
a
pair
APN
maximum
bit
rate
or
per
user
maximum
bit
rate.
So
if
we
have
traffic
coming
from
different
sites,
I
don't
want
to
enter
the
anchor
less
data
plane
discussion,
because
what
anchor
loves
means
write
in
somewhere.
F
You
need
to
have
a
note
where
the
IP
address
of
the
user
equipment
is
so
politically
correct,
but
you
have
other
nodes
that
play
a
role
in
enforcing
policies
for
that
mobile
node,
so
where
to
do
mirroring
right,
I
mean
it
needs
to
be
set,
or
the
middle
layer
needs
to
be
split
between
multiple
data
payment.
That's
to
be
discussed,
that's
pretty
simplistic
in
today's
architecture,
where
all
user
traffic
needs
to
work,
a
single
gateway.
F
K
I'm
Prakash
from
Cisco
I
have
a
couple
of
questions.
One
question
and
one
comments
so
I
think
the
previous
discussion
on
gtp
I
don't
call
this
as
a
problem.
It
needs
an
evolution
because
gtp
is
more
about
the
anchoring
to
the
Gateway
and
it's
kind
of
a
being
a
user
plain
separately
at
the
edge.
It
needs
an
evolution.
So
I
think
if
we
address
it
in
the
next,
you
will
listen,
it
will
go
a
little
bit
smooth
with
the
3gpp
and
we'll
be
able
to
get
more
support.
So
that
is
my
my
comments.
K
F
We
progress
concepts
are
aware,
though,
which
are
not
available
in
the
form
of
a
draft,
so
there
is
work
out.
I
think
what
could
be
accomplished
in
the
IETF
is
a
problem
statement
of
requirements,
analysis
which
then
can
go
into
a
kind
of
informational
document.
I
think
there
is
work
plans
right,
I,
think.
K
E
A
L
Comments
as
hell
I
think
that
is
very
comes
to
me
that
interesting.
If
you
have
any
solution,
then
you
can
go.
But
one
idea
is
that
maybe
that
people
in
here
may
tend
to
agree
and
at
some
point
you
just
raised,
but
something
that
I,
just
as
something
comes
to
my
mind,
that
something
can
be
resolved
by
nowadays
there's
some
technology
enablers.
L
You
just
point
out
the
problems,
but
a
little
bit
that
comes
to
me
that
little
bit
high
levels,
but
there
are
some
more
examples
that
some
you
know
cases
figure
that,
like
that
more
simplification
of
the
network
designs
in
a
very
you
know,
high
levels.
Nowadays,
the
nettle,
slicing
I
separate
that
those
Callanetics
and
list
of
things
will
be
decorating.
L
It
nettles
right
so
and
it'll
be
very
no
clear
to
you
know
that
is
the
net
management
and
orchestration,
but
you
you're,
focusing
on
that
how
we
can
decorate
each
data
plane
right,
for
example,
and
also
that
there
are
many
solution
comes
from
that
also.
You
know
focusing
on
that.
One
of
the
problem
you
focus
now
about
the
separation
of
the
locator
and
IP.
Addressing
so
maybe
a
little
bit
more
clarification
will
be
helpful
for
us
to
focus
yes,
I.
F
Think
what
you
said
at
the
beginning,
I'm
with
you
and
I,
think
it's
a
prerequisite
to
analyze
the
shortcomings
of
today's
data
plane.
We
don't
want
to
do
something
new,
just
to
be
different
right,
so
we
should
really
have
good
idea
why
today's
architecture
is
suboptimal
and
in
different
aspects.
Thank.
J
Every
morning
Saturn
Masterman
Softbank-
this
is
the
update
of
the
service
form
over
is
a
friend
draft
from
me,
ITA
99.
We
got
to
New
York,
oh
no,
and
a
damn
boy
from
Bell
Canada.
J
So
let
me
share
the
V
feedback
from
the
after
DF
ITF
99,
the
main
feedback
from
many
people
it
whatever
the
system
impact,
so
I
figure
out
many
people
really
care
about
the
impact
to
existing
system,
because
I
think
the
many
operate
had
been
the
paid
huge
resource
and
time
to
and
effort
to
build
up.
It
was
our
current
network.
So
when
we
introduce
new
something
technology,
our
solution
to
mobile
is
a
brain
that
cause
a
lot
to
the
work.
Road
to
things
are
rebuilt.
So
that's
what
that's!
What
that's
why
they
really
care
about?
J
This
is
member,
so
the
idea
of
the
impact
will
be
expected
like
control,
plane,
protocol,
a
gtp
she
and
P
me,
and
also
the
current
radio
access
system,
as
Mark
mentioned
before
that
the
they
just
the
cross
connecting
specific
tid
to
specifically
do
channel.
So
that's
a
bit
agnostic
about
ipiria.
So
when
we
decide
when
we
introduce
new
user
friend
solution
demo,
how
do
they
adopt
that
change?
J
J
What's
the
benefit
of
these
are
basics,
and
Ebola
is
a
prayer
I
took
later
so
other
thing
is
the:
if
that's
it's
new
and
very
benefit,
but
isn't
that
in
possible
with
the
currently
supreme
protocol,
so
I,
don't
think
it'll
be
posted,
but
I
think
I
need
your
review,
whether
it
is
achievable
or
not,
by
existing
protocol
and
also
so
the
the
question
is
like:
isn't
it
so
basics,
it's
just
another
autonomy
protocol,
but
actually
the
naming
has
show
that
segment
routing
its
routing.
E
J
Actually
not
tunnel
but
I
think
the
turning
it
point
to
explicitly
the
remote
and
the
point
that
I
think
that's
the
one
kind
of
woody
sauce
working
it's
so,
but
beyond
that
segment
routing,
it's
also
one
of
the
sort
in
technology.
So
then
the
signal
rod
is
our
basic,
can
point
the
mud
crew,
avoiding
point
beyond
that.
Just
on
the
remote
and
end
point.
J
So
the
good
thing
is:
nobody
asked
how
I
saw
physics
works
or
mobile
is
a
plane.
I
saw
I.
Think
I
did
it
to
explain
how
it's
a
bisque
works
for
that
existing
typical
mobile
using
is
a
parent
role,
because
it's
really
just
simple
timing:
it
and
the
endpoint
are
toner
and
the
point
and
then
just
pour
the
packet
based
on
the
caller
ID
or
some
APN
bind
to
the
dot
Tonya.
J
The
first
I
introduced
the
basic
mode
user
prayer.
This
means
that
there's
no
advance
of
future.
It's
it's
almost
or
a
completely
same
with
the
existing
user
frame
network,
because
Adam,
as
I
mentioned
before
so
many
people
who
care
about
system
impact
to
this
network,
so
I'm
trying
to
so
then
I
tried
to
show
no.
No,
it's
it's
it's
possible
to
deploying
as
well
basics.
J
There
is
no
impact
to
the
rest
part
of
the
mobile
mobility
management
system,
so
that
means
it's
just
people
can
be
in
safe,
feel
sane
safe
to
consider
the
new
use
a
premium
solution
in
the
future.
So
but
the
no
impact
is
my
personal
opinion,
so
I
think
we
need
your
review
with
the
no
impact
concept
is
true
or
not
so
even
I
introduced
basic
mode
with
no
change
of
existing
control
parent
function.
J
It
doesn't
mean
I
surrender
to
keep
existing
architectures
I
read
you
want
to
introduce
new
flavor
or
functionality
to
future
use
a
pram
networking,
so
first
introduce
srt-6
without
any
change
the
rest
part
of
the
mobility
system.
That's
enable
us
gradually
gradually
migrate
from
basic
to
more
advanced
user
play
mode
using
SR
basics.
Then
next,
it's
introducing
a
use
case
of
a
state
to
disinter
working
with
legacy
access.
The
remark
mentioned
before
the
and
I
also
mentioned
that
the
existing
lady
access
is
not
familiar
with
the
the
other
as
user
plain
portable.
J
So
what
if
we
can
conserve
the
existing
legacy
used
up
frame
protocol
to
access
network,
so
it
makes
us
easy
to
migrate.
The
user
plan
protocol
except
I
radio
access
network,
so
the
and
also
the
status,
is
a
key
not
to
introduce
any
impact
or
change
to
the
last
part
of
the
mobility
management
like
control
plan.
So
Searchers
means
that
the
existing
mobile
control
crane
doesn't
need
to
care
about
whether
there
is
the
interworking
node.
So
that's
node
does
not
require
any
awareness
of
the
mobility
state.
J
That's
the
beauty
of
the
status
to
make
sure
that,
inter,
inter
working
with
the
existing
of
this
network,
so
this
is
that's
why
it
may
one
my
main
point
to
introduce
aggregate
mode
so
in
a
grid
mode,
is
a
plane,
node
and
assign
the
segment
ID.
Instead
of
the
toner
and
the
point
and
t
ID
to
the
present
use
of
function
in
service
policy
basis
not
so
basis,
that
means
the
multiple
mobility
session
can
be
created
into
the
one
or
multiple
subs
policy
segment
to
be
handled
or
the
mobility.
J
That
means
that
we
can
integrate
the
other
ones
of
future
beyond
the
just
manage
a
mobility.
Currently
we
deploy
in
that
future
also
loaded
with
the
mobility
toner
like
SGI,
is
a
praise
to
put
that
future
charging
Delta
meeting,
not
happy
so
a
future
for
that
our
customer,
but
we
need
that
currently,
but
at
the
other
side,
things
also
deployed
in
the
Gi
Gi
Gi
Ram
area.
So
when
we
introduce
aggregate
mode,
that's
automatic
feature
can
be
integrated
within
the
mobility
path.
Mobility
management
paths
note
the
outside
of
the
mobility
tunnel.
J
So
this
picture
shown
how
we
can
conserve
the
existing
control
plane
protocol
to
know
to
to
no
impact.
So
let
me
explain
a
bit
so.
First,
some
mobile,
no
mobility
management
node.
They
see
the
tunnel
initiation
at
on
a
modification
with
the
tunnel
endpoint
others,
a
chrome
Chrome,
which
is
only
rated
Phi,
0
X,
1,
2,
3,
4,
5,
6,
7,
8
and
then
contour
entity
in
the
null
received.
J
That
message,
then,
that
entity
notice
that
parameter
with
the
a
Krong
Krong
with
direct
1
2
3
4,
5,
6
7
8
to
the
user
per
entity
that
use
a
plain
entity
generate
the
segment
ID
based
on
the
Thunder
endpoint
address
and
so
on,
divided
fire
into
the
one
single
128-bit
airspace.
So
then,
the
side
segment
ID
had
been
formed
like
a
gronckle
one
to
c4
groans,
five,
six,
seven,
eight,
that's
a
way
to
keep
the
control
plane
other.
J
It
is
even
we
introduced
as
a
v6
with
the
basic
mode
that
I
think
there's
no
impact
to
the
continent.
So
then,
once
these
ideas
introduce
afib
and
they
know
the
deceived,
the
pocket
are
distinct
to
the
decon
crown
from
the
associates
of
the
S
control
node
at
the
fifth
table,
to
adapt
and
then
for
the
node.
Who
are
the
pocket
like
this,
the
the
node
leader
I
the
destination,
address
to
the
a
chrome
chrome.
J
What
two
three
four
five
one,
two
three,
four,
five,
six,
seven
eight,
which
is
received
by
the
end
of
the
end
of
the
movie
teamily
node,
but
original
decision
others
had
been
kept
in
the
segment
hit.
Exchanger
hitter
as
the
signet
ID
is
zero.
So
then,
once
even
the
receiving
node
to
to
pop
the
up
the
SL
hitter
to
and
then
back
the
writing
back
to
the
original
decision,
the
crong
crong
at
the
node
can't
find
that
original
destination
from
the
pocket.
J
So
that
means
that
there's
no
need
to
introduce
another
kind
of
the
mapping
table
to
rewriting
the
address
to
the
from
the
intermediate
anchor
from
the
original
destination
address.
So
that's
the
that's
kind
of
the
concept
of
a
lorry
in
the
pocket
so
that
no
need
to
additional
state
or
table
to
do
that.
The
other
one
with
the
advantage
base
our
basics.
J
So
this
is
picture
showed
how
stateless
interworking
segments
are
the
interworking
function.
Working,
let's
see
the
when
the
pocket
comes
from
the
legacies,
for
example,
radio
access
network
to
the
service
network,
the
pocket
had
the
ipv4
source
on
this
edge
under
others,
and
also
the
tundra.
Id
may
be
expected
the
started
to
get
rings.
So
that's
and
the
packet
payload
is
ipv6.
J
We
had
soon
the
pocket
had
the
source
and
essential
others,
so
the
ones
the
tetris
interworking
no
received
that
pocket
and
then
that
no
do
the
like
this.
The
the
popped
out,
the
outer
ipv4,
the
Condor
header
and
the
also
the
generate
the
segment
ID
from
the
receiving
pocket
of
the
ipv4
addresses
and
toner
ID
like
this.
J
So
here
so
the
the
definition
alice
is
the
second
points,
the
segment
which
consists
of
the
locate
of
the
next
receiving
node
and
then
destination
address
of
the
tunnel
ipv4
and
there's
also
the
spur
of
the
tundra
pocket
and
then
donor
ID.
So
even
that
three
parameter
is
started
to
shrew
bit
rings,
it's
just
ninety-six
bit.
We
can
please
we
can.
It's
remained
still
so
started
to
beat
space,
which
is
really
easy
to
get
from
your
nick
organization.
A
picnic
ripe
eating
some
some
others.
J
And
also
the
opposite
direction
from
the
SR
basic
to
existing
ipv4
legacy,
IP
access
network,
the
the
the
opposite
operation
we
performed
like
thundering,
ipv4,
header
and
thunder
ID-
is
extracted
from
the
receiving
segment
ID
and
then
the
node
generates
the
outer
header
and
then
a
the
push
that
header
to
the
pocket
and
then
kwatak
to
the
legacy
IP
network.
That's
how
it
works.
J
Ok,
so
to
do
this
I
trying
to
define
the
pseudocode
to
do
that,
endo,
TM
and
in
this
function
of
the
eternal
mapping-
and
you
can
find
the
a
pseudo
call
in
the
draft
period
in
the
TM
work
for
the
SR
risks
and
to
the
legacy
direction
and
TT
map
is
transplanted.
The
behavior
of
team
up
means
like
a
tunnel
mapping,
so
that
works
for
the
legacy
to
the
sf-86
direction.
So
please
check
the
draft
how
it
that
works
to
do
to
achieve
that
status,
interworking
function.
J
So
this
is
the
list
of
the
work
in
progress
we
have.
We
need
QoS
and
accounting
future
in
the
services.
I
think
that
would
be
easy
to
just
map
the
segment
ID
to
represent
the
QoS
policy
and
also
the
accounting
policy,
but
how
we
define
the
bit
space
to
means
specific
QoS
policy
or
accounting
policy.
Is
it
still
question?
J
So
next
thing
is
n,
2,
n
segments,
routing
and
networks
rising
the
because
it's
RV
6,
it's
a
part
with
the
ipv6
specification,
so
latest
rhynocs
can,
for
example,
have
a
socket
interface
to
push
the
segment
header
to
the
pocket
from
the
that
means.
Application
is
able
to
do
to
this
designate
specific
thrice
or
Liesel's
in
network,
so
that
enable
asked
to
fulfill
the
requirement
to
the
5g
application
based
red
fox,
facing
something
like
that.
J
So
we
need
to
study
more
to
data
possibility
and
go
work
with
that
and
at
the
point,
in
the
application
segment.
Header
2
mapping
to
the
network
resource,
so
ipv4
support
I.
Think
to
make
us
enable
easy
to
migrate
from
existing
to
is
a
v6
at
the
v6
to
be
required.
But
we
already
have
a
bunch
of
the
ipv6
transition
solution
like
a
map,
remap
T,
not
6,
4,
4,
6,
4,
X
rod
and
the
aside
I.
Have
you
seen
the
big
discussion
in
that
basic
sobs
wrist?
J
So
you
may
use
not
6
for
SSID
or
ladies
or
ipv6.
Only
network
now
works
to
bring
ipv4
application.
So
I
don't
mean
we
can
keep
the
ipv6
only
use
a
plane
to
carry
the
ipv4
application
so
lasting
the
cooperation
its
I
mentioned
before
in
the
mic,
and
we
supporting
company
trying
to
propose
a
new
study
item
to
the
3gpp.
J
J
So
excuse
ride,
show
the
how
basic
mode
works,
but
it
is
pretty
much
similar
with
the
previous
meetings
right.
So
we
define
just
access
point
node.
They
add
to
Lankan
old,
like
I
said
when
gateway
essa
gateway
and,
as
you
see
a
kernel,
a
picot
way
as
well.
That's
a
map,
the
existing
just
existing
and
it's
a
risk
and
the
function
there's
no
need
to
introduce
new
function
for
the
basic
mode
with.
A
Shreya,
sorry,
son,
so
one
question
is
like
you
know:
where
is
the?
Is
there
an
assumption
there
there's
a
v6,
the
beginning,
the
starting
point
of
a
service?
Is
it
that
you
node,
B
or
or
the
access
point
or
is
it
da
the
first
operator,
like
you
know,
in
both
the
cases
you
know
you
know
what
is
the
impact
the
service
extinction?
J
J
L
J
Point
because
many
other
working
group
working
on
that
future
SFC
working
group
how
time
to
figure
out
how
to
change
the
services
in
the
past
and
also
the
spring
working
group,
also,
the
home
working
group
within
segments
working
trying
to
introduce
segment
function
with
to
fit
the
kind
of
the
proven
ability
with
your
network.
So
this
work
is
just
liberabit,
the
dot
effort
and
the
solution
to
the
mobile
user,
praying
so
I
I
think
we
can
liberate
that
solution.
J
L
How
have
you
checked
that
how
it
can
address
the
concern
that
raised
by
you
know
that
the
Marco-
and
he
said
it
just
they
don't
show
that
what
is
requirement
and
in
the
Carmen,
because
he
not
afford
her
better
next
generation
data
playing
networks.
That,
in
terms
of
that
I,
wondered
that
how
much
it
can
fully
address
this
concern.
Second,
is
about
contraband
inside
I
mean
how
will
program
this
falling
behavior,
how
it
can
be
aligned
with
a
PC
for
applicability
and
already
seen
here,
yeah.
J
Thank
you,
I
put
it
with
you,
but
my
habit.
It's
almost
like
our
ITF,
like
things
just
trying
to
publish
the
solution
or
idea
the
protocol
extensions,
then
maybe
some
people
find
out
what's
the
problem.
What's
the
applicability
for
that
solution,
them
build
up
the
whole
document
set
or
all
consensus
of
the
this
working
group
or
the
others
working
group
and
also
the
IETF
and
CPP.
Then
this
is
just
a
beginning,
faith
to
publish
the
idea
and
then
discuss.
Ok,.
G
J
J
Rest
until
the
rest
to
not
doing
into
this
much
change
into
the
system,
basic
mode
fulfilled
that
to
the
requirement
to
keep
the
existing
system
as
much
as
possible.
So
then
I
will
get
more.
The
next
phase,
small
to
limited
moles.
It
easy
and
garage
areum
migrated
in
the
future
in
the
in
the
in
each
of
the
mobile
network.
J
A
Thank
you,
sir,
carries
on
so
I.
Think
what
you
know
your
thinking
is
to
just
you
know,
get
a
sense
of
you
know
what
the
working
group
you
know
feels
about
this
work
right
in
general,
I.
Think
it's
promising!
You
know,
I
think
we
are
oppositely
ietf
to
really
influence
3gpp.
This
is,
you
know,
I
think
looking
at
it,
you
know.
If
we
can
clearly
change
the
data
plan,
we
can
significantly
optimize
it.
So
there's
a
right
step
in
the
direction.
I
think
that's
what
you
know.
A
A
L
Sorry,
one
question
actually
I
showed
a
my
support
here,
but
I
always
that
I'm
confusing
dad
when
he
just
comes
to
her
some,
you
know
agreeing
to
go
with
a
certain
Idid
draft
for
our
working
group
document
that
maybe
I
remember
that.
Maybe
you
just
told
me
that
before,
when
I
just
try
to
make
that
some,
my
personal
document
for
our
working
group
document
or
adoption
that
you
told
me
that
there
is
no
charter
items.
But
how?
How
is
in
this
case
is
how
usually
working
here,
India
sure
I
think.
A
C
So
description,
and
so
as
ad,
so
the
thing
is,
the
Charter
does
cover
this
work,
but
there's
no
milestones
for
this.
Okay,
so
like
there's
like
a
slight
disconnection,
always
like
the
Charter
is
like
usually
scoped
wider
than
the
thing
the
working
group
is
currently
working
on.
So
if
this
doesn't
fit
in
the
Charter,
the
working
group
needs
to
be
charter.
Okay,.
E
C
If
it's
covered
by
the
Charter,
then
it's
a
question
of
adding
a
milestone.
So
what
happens?
Is
the
work
group
decides
to
adopt?
They
cannot
do
it
yet
right
like
so,
they
can
say
like
okay,
like
you
know,
it's
ready
for
adoption,
then
the
chairs
need
to
put
in,
like
a
milestone,
add
a
milestone
and
the
milestone
shows
up
to
me
and
if
I
approve
the
milestone,
I
look
at
it
and
I
say
like:
are
they
gonna
make
the
timelines?
Usually
the
other
work?
New
milestones
are
like
hopelessly
out
of
date.
Right
like
some.
C
Some
are
like
you
know
off
by
a
year,
some
are
off
by
8
years,
10
years,
whatever
right
but
like
at
some
point.
We
need
to
evaluate
whether
the
working
group
is
able
to
keep
the
milestones.
So
that's
what
I
do
so
I
look
at
it
and
I
say:
okay
can,
like
so
there's
gonna,
be
like
an
adoption
milestone
like
a
La
Scala,
milestone,
publication,
milestone,
okay,
so
I,
look
at
it
and
I
say
is
the
working
group
capable
of
doing
this?
C
L
C
Like
this,
so
the
way
it's
done
is
the
chatter
description
covers
what
what
are
the
things
the
working
who
wants
to
do
right,
but
it
doesn't
go
down
to
the
level
of
granularity
of
the
milestone.
Some
shadows
do
some
don't
the
DMM
Charter
does
not
right
and
like
this,
like
working
groups
where
I
have
charters
which
specify
no
milestones,
for
example
interior.
Okay,
it's
a
working
group.
We
don't
know
what
work
is
going
to
come
up
right,
but
we
know
the
scope
of
work
that
we
want
to
do
there.
C
Okay,
so
the
milestones
will
get
added
as
like
they
come
along.
It's
like
six-man
right.
We
don't
know
what
kind
of
so
DMM
is
like
a
no
matter
what
people
think
there's
a
maintenance
working
group?
Okay.
So
if
somebody
finds
an
issue
with
mobile
ipv6
tomorrow,
I
cannot
anticipate
that
in
the
Charter
right.
So
it's
gonna
be
something
that
comes
up
randomly
at
some
point
and
then
we
decide
to
do
the
work
or
not.
So
that's
like
yeah.
C
So
there's
like
newly
chartered
working
groups
for,
like
you
know
greenfield
things,
work
a
bit
differently
than
maintenance
working
groups
where
we
cannot
anticipate
up
front.
What
are
the
kind
of
things
we
need
to
do
in
the
working
group?
Okay,
so
at
this
fine,
like
I,
think
it's
called
by
the
Charter
right.
C
But
if
people
don't
believe
it,
then
we
can
probably
talk
about
like
you
know
how
we
go
about
it
like
you,
do
how
to
each
other
or
or
like
not
take
it
or
less
like
multiple
options,
then
there
right,
but
I,
do
think
the
Charter
covers
this
work
personally.
So
it's
just
a
question
of
adding
milestones.
C
So
if
there's
like
energy
in
the
working
group,
because
what
I'm
looking
for
is,
like
you
know,
people
willing
to
work
on
it,
people
willing
to
review
it,
because
that
is
something
that
has
completely
lacked
in
this
working
group,
a
set
of
views
and
that's
something
I'm
gonna
see
if
I
don't
see
like
you
know,
like
a
whole
bunch
of
people
willing
to
review
it
like
I'm,
not
gonna.
Let
you
take
this
up
right,
because
I
I
don't
want
poorly
reviewed
documents
coming
up
to
that
is
she
okay?
Thank.
L
J
E
E
I'm,
just
a
and
I
have
to
apologize
is
I've,
not
read
the
draft,
but
you
know
clearly:
it's
yeah
I
know
I'll
move
to
the
back
when
I
get
to
go
to
sit
down.
Just
of
course
there's
an
awful
lot
of
dependencies,
how
s
rv6,
ultimately
works,
and
if
this
is
is
going
to
end
up
diverging,
possibly
from
what
is
now
an
incomplete
work.
That
to
me
would
be
a
concern.
A
M
C
So
suresh
krisshnan
so
like
to
Dave's
point
right.
Stuff
always
diverges
as
long
as
they're.
Like
you
know,
you're
referring
to
drafts
normatively
like
we
have
to
keep
track
of
it.
I
think
probably
like
before
working
of
last
call
we
should
sink,
and
it's
not
just
spring
it,
because
it's
also
six-man
because
we're
using
SR
v6
and
you
know
what
is
happening
in
six-man
right
like
in
OS.
Our
v6
is
like
flipping
and
flopping,
like
you
know,
between
end
cap
and
and
an
insertion
right
like
so
that
that
thing
analysts
like
that
gets
resolved.
C
This
is
not
going
forward
like
at
some
point,
but
that
doesn't
mean
like
working
group.
Adoption
doesn't
mean
the
document
is
done.
It
means
it's
beginning
right.
Like
so
I,
don't
mind
you
like
starting
to
work
on
it
but
like
before
it
gets
published
it's
a
while
away,
because
all
these
other
things
that
they've
had
to
be
resolved
before
the
document
gets
published.
No.
A
A
A
M
M
A
M
A
coupon
to
you,
the
and
then
the
funds,
five
to
six
was
basically
addressing
the
comments
from
Carlos
and
and
he's
okay
with
them,
and
then.
A
M
Then,
from
Russian
for
the
five
who
were
the
comments
from
Terry
and
the
Jim
very
common,
had
animal
changes
and
on-
and
it's
also
okay
with
those
comments
and
they
just
come
in,
was
basically
on
the
document
being
too
long
to
read
and
he
was
he
was
not
not
concerned
about
the
content,
so
it
to
fix
it.
I
shortened
it
to
about
from
I
think
I'm
fifty
pages
to
forty
some
pages
yeah,
it's
not
much
but
but
to
take.
M
But
you
put
down
the
protected
additional
label
so
that
one
does
not
need
to
read
the
whole
document,
because
this
is
a
we're
trying
to
be
very
comprehensive
document.
All
the
functions
of
anchors
and
by
having
the
the
list
label
are
the
titles
to
to
show
that
if
you
are
not,
for
example,
if
not,
if
you
don't
care
about
Hannah,
you
can
skip
out
the
label
that
says
tandem
to
make
it
easy
to
rate
and
then.
M
They're
already
taking
your
feelings
fall
and
then,
and
also
before
that
there
were
a
lot
of
comments
from
uni
party
that
there
were
personal
comments
and
several
times
and
maybe
fixed
at
all
and
other
than
that
there
were
comments
from
co-authors,
okay.
So
those
are
they
videos
that
we
have
done:
okay,
yes,
okay,
so
everyone
that
is
more
content,
video,
so.
A
A
A
A
A
E
A
N
Okay
next
slide
just
reminder
some
history.
This
draft
well
went
through
workup
last
call
about
a
little
bit
more
than
a
year
ago
and
at
some
point
Suresh,
we
called
it
back
to
merge
with
another
draft
that
was
written
by
Ceylon.
We
did
that
and
since
then
we
got
some
good
comments
about
this
review
and,
as
a
result,
added
sample
code
and
change
the
socket
API
that
we
were
using
due
to
some
concern
about
the
blocking
nature
of
the
call.
We
did
all
that
in
the
last
face-to-face.
N
We
basically
had
one
comment
about
asking
to
modify
the
word
code
to
pseudocode,
because
the
comment
was
that
the
code
won't
compile,
and
there
was
a
question
about
or
request
two
more
for
more
clarifications
about
the
different
type
of
services,
so
I've
changed
the
text
to
pseudocode
and
while
reviewing
the
document,
I
added
some
text
that
describes
the
merging
of
of
this
functionality
with
the
functionality
that
is
mentioned
in
our
c
50:14.
I.
N
Than
that
I
put
on
the
list,
I
actually
I
copied
the
description
of
the
different
service
types
from
the
draft
and
I
added
some
examples
or
use
cases
as
to,
in
which
case
different
service
types
should
be
used
and
I
think
the
document
is
ready.
I
didn't
receive
additional
comments
after
posting,
the
updates
no.
A
A
N
A
N
A
L
L
L
Actually,
after
last
IDF
meeting,
we
try
to
address
and
we
try
to
get
the
comment
and
feedback
in
the
list
and
also
we
addressed
their
comments.
We
believe
so,
and
so
we
got
useful
feedback
and
comment
in
the
list
and
sense
to
someone
and
the
truck
and
actually
but
it
was
about
the
mostly
the
editorial
point
sent
in
it
and
we
didn't
get
any
some
and
his
prospects.
L
You
know
some
against
under
main
the
main
you
know
pilla
and
so
some
typos
detected
we'd
resolved
and
the
reference
requested,
especially
that
in
case
of
NS
h,
is
nation.
So
actually
that
is
about
that.
What
is
about
NS
h
classifier,
so
we
put
that
some
RFC
76
65,
also
extension
of
acronyms,
that
mobility,
controller
and
routing
controller.
Also
we
put
that
this
and
actually
the
before
that
definitions
has
not
been
given
there,
but
so
we
put
that
this
definition
in
the
technology
section
so
and
two
days
before
we
submitted
a
jerry
version.
L
So
now
you
can
see
that
revised
version
and
also
will
be
very
helpful
that
if
you
can
give
some
more
comments
and
feedback
and
and
it'll
be
more
highlighting
and
so
I
think
that
we
suggest
moving
this
idea
into
learn.
It
leave
you
and,
as
we
suggested
before
in
the
pro
meeting,
but
as
the
chair
and
the
mention
in
the
first
time
that
we
are
also
carefully
see
that
also
pcs
progresses
and
the
order
and
draft
perspective
and
aspect
that,
in
which
point
that
we
need
to
address
in
the
diplomatic
model
document.
L
D
Okay,
so
in
kind
of
another,
ten
percent
in
this
on
behalf
of
my
co-author
Antonio
fav,
you
and
Juan
Carlos,
so
well.
This
has
been
already
discussed
a
few
times
in
in
this
working
group.
This
is
a
basically
a
solution
for
DMM
for
proxy
mobile
ipv6,
so
I
will
try
just
to
summarize
the
way
it
operates
and
then
some
some
also
some
words
about
some
demonstrations
that
we
have
done
and
open
source
code
that
is
available,
implementing
the
solutions
and
then
some
hopefully,
some
questions
to
the
working
group
and
potential
next
discussion
on
potential.
D
Oops
sorry,
so
this
rough,
actually,
that
rather
impress
untinted
a
replaces
a
couple
of
drafts
that
we
had.
We
try
to
as
a
result
of
the
discussion
that
we
have
in
the
last
IDF
meeting,
tried
to
combine
and
provide
a
unified
one
single
single
draft,
with
all
the
solutions
that
we
had
in
the
past
for
proximal
ipv6
as
I
mention
is
basically
an
extension
to
pin
it
in
which
we
basically
move
the
mobility
to
the
2vh,
and
we
do
that
by
basically
having
an
architecture
that
have
a
distributed
data
plane
with
partially
centralized
control
plane.
D
So
we
have
a
as
we
will
see
later,
the
the
routers
at
the
edge
basically
assume
the
functionality
of
LMS
and
Mac's,
depending
on
on
the
on
the
prefix
and
the
node.
And
then
we
have
a
kind
of
control,
only
LM
a
which
is
basically
keeping
track
of
all
the
prefixes,
whether
they
are
anchor
on
and
ensuring
that
the
signaling
is
is
done
between
the
different
nodes
to
maintain
the
the
connectivity
of
the
different
prefixes.
D
So,
in
terms
of
sorry
terminology,
we
have
these
two
nodes.
One
is
the
mobility
and
column
access
router,
which
is
you
can
see
that
as
an
exit
router
that
is
behaving
as
LM
a
and
a
Mac
for
the
different
prefixes.
Then
what
additional
terminology
when
this
Mar
I
mean
for
the
nodes
that
are
attached
or
handled
by
this
this
mark?
D
We
call
that
serving
mark,
because
our
is
that
they
know
that
is
serving
the
the
mobile
node
and
then
that
node
may
be
also
anchoring
some
prefixes
for
attached
notes
or
for
other
nodes
that
are
attached
to
a
different
Mar,
and
for
that
we
use
the
terminology
and
core
Mar,
because
it's
encoding
the
projects.
So
basically
those
entities
are
the
ones
that
are
involved
in
the
data
plane
in
the
forwarding
of
the
packets
and
then,
additionally,
we
have
this
centralized
mobility
database,
DCM
D,
which
is
basically
the
LMA
for
the
control
plane.
D
In
terms
of
I'm,
sorry
take
some
time.
Okay,
in
terms
of
so
this
in
this
diagram,
I
tried
to
summarize
the
the
operation
of
the
solution.
We
have
here
the
control,
the
CMD,
the
control
plane
entity.
We
have
a
domain,
we
have
different
anchors.
He
has
a
couple
of
them
for
the
sake
of
example,
and
we
have
a
mobile
node
so
for
the
initial
registration
when
the
mobile
node
that
touches
to
one
of
these
anchors,
we
have
here
the
diorama
for
the
frou-frou
sequence.
D
We
have
the
attachment,
they
marked
it
at
that
it
detects
the
attachment
and
creates
a
prefix
for
the
node
and
basically
sends
a
PBU
to
the
CMD
to
the
LMA
for
the
control
plane.
So
then,
as
similary
us
with
pimp
v6,
we
have
see
indeed,
Ellen
may
assign
in
a
the
prefix
and
basically
send
in
the
PVA
the
proxy
binding
acknowledgement,
and
at
this
point
the
access
router
can
just
send
the
router
advertisement
with
the
prep
picks
and
the
Monod
can
configure
that
prefix
and
start
using
that.
D
D
Okay,
so
we
have
mobility,
so
we
have
a
mobile
moving
to
another
router.
Basically,
we
have
here
that
were
to
happen.
What
will
happen
will
be
that
the
mobile
node
attaches
to
the
action
router
to
the
extra
tooth.
It
detects
that
connection
it
will
do
follow,
follow
the
same
procedures
that
before
we
send
a
PV
you
to
the
CMD,
the
CMD
will
check
that
the
guy
is
already
connected
into
the
in
the
domain,
so
basically
we'll
send.
D
D
Router
word
the
guy
has
attached
in
order
to
ensure
that
this
prefix,
one
that
was
anchored
a
design
quality
access,
router
one
will
keep
with
the
connectivity
or
that
reach
ability
will
be
maintained
and
sending
the
traffic
through
this
new
tunnel
created
here
and
additionally,
we
will
just
say:
okay,
it
will
also
send
a
PPA
to
the
exit
router
to
saying
okay.
This
is
the
prefix
that
you
have
to
use
for
the
locally
prefix
locally
and
called
prefix
at
the
exit
router.
D
D
So
the
more
I
note
now
has
two
prefixes
one
anchor
here
and
the
second
one
anchor
here
and
you
can
see
here
the
different
state
in
the
in
the
different
routers
and
for
this
the
monitor
one,
a
mobile
node
one
for
the
traffic,
using
that
the
previous
prefix
traffic
will
go
through
access
rotor
through
the
tunnel
and
then
being
forwarded
man
access,
rotor
one
and
for
the
new
traffic
new
sessions.
It
will
use
the
new
prefix
local
anchor
here,
and
we
just
follow
that
traditional
path.
We
know
tunneling.
D
A
Have
a
question
yeah,
quick
question
now,
given
that
in
the
cups
kind
of
more
up,
you
know,
architecture
with
the
user
and
control
plane,
separation
right,
use
the
plane
and
control
plane
separation?
How
does
this
approach
you
know?
Can
you
repeat
the
question
now,
with
the
separation
of
control
and
user
plane,
yeah.
D
D
We
have
the
the
control.
Plane
is
just
plain:
this
isn't
the
only
note
that
is
isn't
rising
the
control
plane,
and
then
we
have
the
data
plane
managed
by
the
ankles,
the
the
DeMars,
the
rotor.
So
all
the
control
plane
is
control
here
is
this:
is
the
guy
keeping
the
cash
the
prefixes
were
they
they
are
anchor,
and
this
is
the
guy
sending
the
previews
and
PPAs
to
this
this
guy's.
So
basically,
this
data
plane
control
plane,
so.
D
N
So
the
older,
the
open
session,
needs
to
continue
to
use
a
non
optimal
route
using
the
the
first
prefix
and
new
sessions
can
use
the
new
prefix
and,
as
the
mobile
node
continues
to
be
mobile,
you
might
have
more
and
more
prefixes.
What
I
wanted
to
point
out
is
that
if
you
use
the
different
types
of
services
that
a
mobile
node
might
want
to
request,
there
is
also
a
mode
where
a
mobile
node
wants
a
session
lasting
or
what
we
called
graceful
replacement.
Meaning,
yes,
don't
break.
N
N
A
M
A
question
just:
would
it
help
to
try
to
relate
the
new
terms
that
you
have
later
a
relate
to
the
terms
that
are
defined
in
the
DMM,
such
as
it
depends
and
contra
pending
and
and
those
terms
I
either
use
those
them
or
explain
at
that
explanation?
How
they're
related
yeah.
H
I'm
Charlie
Perkins,
I'm
I,
remember
a
draft
called
fast
mobile,
IP,
F,
MIT
and
I'm
sure
you
read
and
understand
that
completely
and
there's
a
seem
like
there's
some
overlap
between
that
and
here
in
F.
Mit
of
access
routers
have
a
make
perhaps
a
bit
more
control
over
establishing
the
tunnel
and
and
carrying
context
back
and
forth.
H
But
on
the
other
hand,
the
way
it
interacts
with
the
you
know,
the
sort
of
augmentation
to
neighbor
advertisement
and
solicitation
message
seems
like
it
would
work
pretty
darn
fast,
so
just
wondering
if
your
would
be
interested
in
trying
to
identify
the
differences
between
F
MIT
approach,
and
this
I
mean
if
I
guess.
The
most
obvious
thing
is
that
the
CMD
seems
to
be
in
charge
of
all
the
mobility
here.
We're
in
the
FF
case.
There's
a
lot
more
local
control
over
there
yeah.
O
D
Can
I
can
work
on
that,
so
just
to
I
mean
I
will
skip
this.
The
leaf
part
I
mean
this
is
just
to
to
mention
that
there
is
another
stuff
on
the
document
for
how
to
ensure
that
the
the
mobile
node
gets
always
the
same
accident.
Logical
access,
router,
point
of
view
when
it's
moving,
so
basically
it's
like
every
time
it
connects
and
get
a
local
yank
or
traffic's.
You
will
see
we
will
interact
with
a
kind
of
local,
actually
router
interface
that
will
move,
will
follow.
D
The
more
I
know
that
it
moves
and
we
have
stuff
on
the
draft,
but
I
think
that's
not
the
main
point
for
today.
There
are
how
this
is
playing
here,
and
these
allow
some
nice
features
like,
for
example,
ensuring
that
the
traffic
for
a
local
network
can
be
managed
by
the
router
that
is
providing
access
to
that
network.
D
While
the
Monod
moves,
but
that's
again
not
really
the
point
for
today
very
very
fast
I
really
mentioned
in
the
past.
We
have
running
code
with
demonstrated
ease
into
ITF
meetings
and
there
is
open
source
implementation.
Is
you
wanna?
Try
the
implementation
of
this
and
then
the
last
one
at
firm
is
the
most
important
one
is
we
are
in
a
DMM
working
group.
We
have
been
working
on
EMS
staff,
very
good
stuff
on
in
architecture
and
deployment
models
considerations.
The
point
is:
do
we
want
or
that's
the
question
I
would
like
the
chairs.
D
Allow
me
to
ask
is
to
the
working
of
this:
do
we
want
to
work
on
a
standardized
in
a
premie
p6
based
solution,
and
if
the
question
is
yes,
I,
don't
know
how
many
people
have
read
this
draft
for
the
previous
graph
and
considered
maybe
good
good,
a
starting
point
for
for
this.
So
I
don't
know
if
you
wanna
answer
questions
I.
A
D
A
B
Okay,
I'm
going
to
present
for
Wi-Fi
work
EMM
for
Wi-Fi
next
yeah,
so
this
is
in
kind
of
use
case
to.
B
To
investigate
how
we
can
deploy
technologies
to
in.
B
The
part
of
it
should
say
big
part
of
it
was
done
in
the
university
and
afterwards
we
followed
up
to
finish
it
and
write
this
draft
and
the
university
part
was
written
only
in
your
side.
We
have
several
papers.
Even
a
book
chapter
was
published
part
of
this
part
of
this
project,
so
don't
take
it
so
lightly,
so,
basically
yeah
the
scope
is
and
to
basically
call
investigated
the
u23
mobile
team
people
say
well.
What
are
you
doing
about
the
year?
B
Well,
because
in
the
previous
Network
slightly,
that
is
this
provider
called
part-time?
We
wanted
to
handle
the
mobility
there
also
and
then
stn
okay,
so
that
another
part
of
is
this
project
is
a
big
scope.
Mother
part
of
it
was
demon
the
sorry
Sdn
control.
So
we
wanted
to
do
the
layer,
2
question
control
and
the
earth
Christian
control
the
you
to
astern
control
part.
B
We
did
this
the
you
to
mobility
there
and
in
the
SDN
controller,
and
then
they
will
tree
the
routing
control,
so
route
establishment,
drought,
management
and
then
using
more
modern
technologies
like
yang
next,
there
yeah
there
to
mobility.
Basically,
our
contribution
there
is
the
mobility
table,
keeping
for
each
mobile
and
near
23
information
together.
So
anything
anyone
has
can
change.
So
you
you
have
the
mobility
one.
So
this
way
the.
B
E
B
Dmm
technologies
basically
P
meet
after
me,
and
so
on
so
hand
over
timidly,
signaling
and
then
route
establishment
after
handle
were
using
Australian.
So
basically
a
man
is
given
a
fixed,
perfect
and
then
it
keeps
this
prefix
as
it
moves.
So
that's
why
it's
kind
of
like
assaulting
and
then
so
basically
for
the
new
New
Age
ADP
EndNote
from
the
deployment
drugs
then
had
the
hospital
for
the
man
and
the
propagates
upstream
and
that
it
previous
ones
little
next
night.
B
B
E
B
How
to
establish
the
this
part
of
this
work
is
a
feces
not,
for
example,
Marco
was
talking
about
establishing
the
relationship
of
our
RPC
work
and
then
the
FPC
seems
to
be
the
only
young,
the
RPC
type
of
work
net.
Cough
methodology
work
in
the.
So
what
is
there
any
relationship?
One
thing
here:
we
noticed
that
there's
no
rot
management
there,
so,
but
if
there
is
any
other
relationship
we
have
to
know
so,
especially
on
this
aspect,
we
want
comments
from
the
work.
B
A
Comments
are
welcome,
very
good
I,
think
Oh
looks
like
you're
doing
some
experimental
work
and
all
of
that
I
think
you're
continue
at
some
point
when
you're
ready,
okay,
thank
you.
So
last
presentation,
or
maybe
I,
think
without
in
your
presentation
on
your
pH
CV
is
here
or
not
anybody
from
John
Hurt's.
You
know
so
team,
but
that's
okay
would
take
that
off.
So
then
this
is
the
last
presentation.
F
N
L
N
Regarding
the
types
of
services
you
evil
desires,
another
way
of
informing
the
network
is
via
DHCP,
and
this
draft
was
presented
here
a
couple
of
times
in
the
last
or
last
face-to-face
I
mean
the
one
before
I
was
asked
to
present,
that
to
DHD,
and
so
I
did
that
in
the
last
face-to-face
in
general,
the
feedback
was
positive,
I
got
some
comments
and
then
I
updated.
The
draft
based
on
that.
These
comments
mostly
were
editorial,
but
the
worst
some
comments
regarding
some
option:
format
which
I
updated
yeah
so.
A
One
shirring
and
early
so
a
couple
of
questions
I,
you
know
I
I,
unfortunately,
I
didn't
review
the
doctrine
in
the
past,
but
today
I
just
browsed
it.
So
one
question
is
like
you
know,
when
you
specify
a
request
saying
that
you
know
I
want
this
particular
type
of
address.
Right,
yeah,
I,
just
I
forgot
the
food.
The
different
types
of
addresses
that
you.
E
A
But
session
continuity,
for
example-
let's
take
all
right
so
this
does
you
do
you
think
DHCP
server
should
be
exposed
to
this
liver
of
aspects,
because
there's
so
much
of
a
boot
session
management
right
when
you
say
DHCP
server,
give
me
an
address
via
a
new
session
continue.
What
does
it
mean
from
the
server
point
of
view?
Okay,
okay,
so.
N
Basically,
the
UE
requires
a
certain
type
of
service,
not
just
an
IP
prefix
I.
Don't
expect
the
DHCP
server
to
be
able
to
allocate
a
certain
type
of
service,
but
I
expected
the
DHCP
server
to
communicate
with
some
control
entity
in
the
network
and
receive
back
actually
what
the
DHCP
server
should
receive
back
is
a
prefix
and
an
acknowledgment
regarding
whether
this
service
type
was
provided
or
not,
and
as
a
and
as
a
consequence
of
that
acknowledgement.
A
See
that
is
a
big
assumption.
The
issue
is,
like
you
know,
I
think.
What
is
that
interface
right?
How
does
the
beach
just
should
the
DHCP
server
talk
to
like
LM
me
or
the
P
gateway
or
exactly
yeah
I?
Think
that
is
a
big
interface
I
think
that's
the
concern.
I
have
right.
That
is
one
point
right.
Second
point
is
like:
if
you
look
at
you
know,
I
think
even
serration
was
we
did
some
work
in
the
past.
A
To
you
know,
has
this
kind
of
request
in
generalized
form,
like
you
know,
ask
a
client
can
potentially
ask
I
want
this
particular
type
of
address
right,
like
class,
that
does
type
is
a
very
generally.
We
try
to
model
it
along
the
class-based
or
whatever
different
types.
There
was
a
lot
of
discussions
and
all
of
that
that
work
never
went
forward
that
you
know
we
tried.
So
you
know
ten
years
all
right
few
years
we
tried
and
didn't
go
in
here
right,
I,
think
that
is
I.
Think
you
may
want
to
look
at
that.
A
N
N
N
P
N
They're
not
using
ITF
protocols-
and
this
sounds
like
a
kludge
to
me
because
they
will
be
the
UE-
will
request
service
via
one
protocol
and
receive
the
reply
via
a
different
protocol.
It
does
it's
it's
kludgy,
but
in
order
to
solve
that,
we
will
also
need
to
modify
router
solicitation
and
I
heard.
This
is
practically
impossible.
So
that's
why
I
was
trying
to
push
DCP
as
an
alternative
solution.
Okay,.
C
P
C
Not
a
query
response
protocol.
Okay,
so
even
if
you
modify
an
RS
and
RS
is
optional,
so
iris
does
not
need
to
be
sent,
so
you
still
need
to
figure
out.
How
are
you
II
get
Sabri
fix.
So
if
you
don't
have
an
RSS,
a
signaling
mechanism,
how
do
you
give
them
the
right
terrific?
So
that
means
to
work
anyway.
So
if
any
protocol
that
depends
on
an
RS
being
sent,
it's
gonna
fail.
C
Okay,
because
first
thing
RS
don't
have
to
be
sent
based
on
link
layer.
You
can
decide
not
to
send
RS.
Second
thing:
RSS
can
be
lost
their
sent
on
them
to
a
multicasts,
all
routers
address.
Okay,
they
can
be
lost
like
on
Wi-Fi
on
an
IC
length
like
you
could
probably
lose
RSS
okay.
So
if
you
depend
on
RSS
to
work,
that
thing
is
finished.
Okay,
second
thing,
as
Sri
said
before,
so
this
like
not
with
AD
hat
on,
we
have
tried
this
many
times.
C
It's
not
like
one
graph
that
went
over
ten
years
and
died.
So
this
is
like
multiple
drafts
with
multiple
stuff.
So
there's
like
something
that
I
did
with
like
the
Nokia
people
like
the
Raj
and
like
da
Malloy,
and
everything
Fairview
specifically
had
only
like
very,
very
specific
things,
which
was
like
home
address
kind
of
thing,
which
gives
you,
mobility,
support
and
care
of
address.
C
Where
you
don't
care,
okay,
that's
all
it
had
and
then
there's
like
the
stuff,
like
you
know,
Sri
did
like
and
we
work
together
in
myth
as
well,
which
is
like
more
gender
akan,
labeling
prefixes
with
characteristics,
okay,
if
it's
just
not
like
home
or
whatever
you
could
like
put
whatever
label
on
them
as
long
as
the
mobile
or
the
node
asking
for
it
understands
what
that
thing,
it
wants
it
and
then
we
can
get
it.
Then
things
would
work
and
this
didn't
work
either
right.
C
C
Okay,
so
so
now,
that's
done
so
like
as
long
as
did
so.
Just
to
answer
Charice
point
is
that
the
CP
server
does
not
understand
what
is
handing
out
same
way.
The
DHCP
server
does
not
understand
what
NTP
does
when
it
hands
out
NTP
server
information.
It
does
not
understand
what
some
know
that
gives
you
session.
Continuity
does
but
there's
like
I
assume,
there's
some
kind
of
interface
that
populates
the
DHCP
server,
with
a
set
of
prefixes
that
give
you
mobility,
properties
and
set
of
things
which
don't
so
and
we're
not
doing
that
interface
here.
C
C
Architecture,
as
somebody
said
like
we
do
protocols
so,
like
maybe
that's
3gpp,
like
you
know,
maybe
it's
like
some
interface
between
the
HS
S&P
gateway
that
you
have
to
do
or
you
can
kludge
it
on
something.
I
don't
know
right
like,
but
that's
something
that
needs
to
get
done,
because
the
DHCP
server
needs
to
get
this
set
of
prefixes
like
to
hand
out,
because,
without
that
it's
just
like
all
the
DHCP
server
is
gonna.
Do
is
it's
gonna
get
an
option
right?
C
It's
gonna,
look
in
a
leased
file
or
some
database
that
it's
gonna
hand
out
and
then
hand
out
the
thing
you
asked
for
that's
it.
It
does
not
understand
the
semantics
of
what
it's
getting
used
for.
So
I.
Don't
see
that
changing
in
DHC,
like
nobody's
gonna
change,
DHCP,
to
make
this
understanding
and
I
don't
see
that
worked
relevant
in
DMM
here,
because
it's
not
about
doing
this
stuff.
Yeah.
C
Take
a
look
at
it,
but
I
think
it's.
It's
got
a
very,
very
different
way
of
solving
things
right,
because
all
you
get
is
like
a
pointer
like
a
very
can
go
query
for
the
things.
So
it's
like
more
lightweight
in
that
way,
because
our
it
just
tells
you
where
to
ask
for
more
info.
Sorry
itself
does
not
contain
the
info,
so
that's
a
different
mechanism
to
do
it.
So
if
you
want
to
pursue
that,
that's
okay,
too,
but.
N
N
We
were
asked
to
support
that,
and-
and
it's
very
very
difficult
I
mean
it's
not
a
very
complex
feature
and
still
we
are
struggling
how
to
support
that
and
I
don't
see
any
alternative
other
than
either
RA
or
DHCP
or
both,
and
it
looks
like
we
are
hitting
brick
walls
in
both
alternatives.
So
what
will
be
I
mean
and.
N
C
A
A
N
C
There
is
some
work
to
be
done
in
3gpp
that
you
need
to
do
like
on
how
this
thing
gets
populated,
I.
Think
that's
how
I
see
it
right
like
so
this
work
is
not
gonna,
be
entirely
done
in
ITF,
so
there's
gonna
be
idea,
protocols
which
you
need
to
put
into
something
the
same
way
right
if
I
need
to
get
an
RA
from
a
3gpp,
Network.
C
A
N
I,
don't
think
we
will
be
able
to
advocate
for
more
work
in
3gpp
regarding
the
HCP,
because
they
are
relying
on
our
a
and
and
and
I
believe
they
will
only
consider
I
mean
using
DHCP,
it's
a
big
change
for
them
and
they
will
they
I,
don't
see
them
considering
that,
but
if
they
will
consider,
that
is
only
if
RA
is
blocked.
So.
C
Again
so
description,
so
going
back
to
like
that
right
so
like
there's,
no
dhcpv6
support
in
in
mobiles,
okay
right,
which
is
clear.
Okay,
so
back
when
we
start
off,
there's
no
DHCP
PD
support
and
3gb
reader.
We
just
had
to
work
like
very
hard
to
get
it
included.
We
had
to
do
work
in
the
ITF,
which
ended
up
being
like
6603
right,
which
was
the
whole,
because
the
IP
concession
could
only
have
one
prefix,
so
we
had
to
go
and
do
work
in
IETF
and
3gpp
to
get
that
done.
C
So
it's
not
easy,
but
it
needs
to
get
done.
So,
if
that's
what
it
it's
required
to
make
it
work,
that's
what
you
need
to
do
because,
like
you,
need
to
go
write
in
some
spec
in
3gpp.
I
said
no
wait.
Wait
like
like
you
need
to
write
in
some
TDP
expect
the
Mobile's
need
to
support
dhcpv6
and
it's
up
to
you
to
pass
it
like.
It's
ITF
cannot
like
go
and
tell
3gpp
that
dhcpv6
is
required.
We
can
probably
put
out
a
recommendation.
Rfc
we've
done
them
before.
C
So
there's
like
I,
think
seventy
six
to
six
that
we
worked
on
a
couple
of
years
ago,
which
talks
about
you
know
like
what
are
the
things
the
3gpp
host
needs
to
do
for
v6
so
like.
If
you
can,
like
you
know
thumb
of
it.
Another
RFC
go
to
v6
offs
or
somewhere
and
say
like
okay.
This
is
like
a
new
set
of
things
to
do,
and
and
3gpp
takes
it
awesome
right,
but
that
works
needs
to
get
done.
It's
not
going
to
be
done
from
DHCP.
A
A
P
Actually,
we
have
developed
relative
mobility
support
protocol
like
mobile,
IP,
próximo
variety,
and
we
have
a
lot
OB
extension
of
them
and
we
are
now
developing
TMM
protocol
as
well.
So,
but
we
don't
have
a
protocol
selection,
which
means
that
when
dumbo
below
the
one
that
oxide
had
your
video,
several
protocol
can
be
supported,
but
we
don't
have
any
negotiation
was
election
protocol
for
mobility.
So
this
is
the
motivation
of
buddhist
activity.
P
P
Here
you
category,
actually,
we
categorized
to
a
network
of
a
stand
host
based
the
protocol
and
at
the
NATO
place
at
the
protocol
side,
we
have
a
TTP
and
the
froakie
froakie
mobility,
which
is
the
DMM
and
damián
protocol
and
test
hand
over
pole,
approachable,
VAR,
t
v6
and
the
hosts
aside.
We
have
Bob
our
IP
and
mobile
IP
extensions
like
path
to
handle,
bow
and
Heidi
car
and
M
weather.
P
The
first
priority
we
just
follow:
Network
ability
and
the
second
priority,
the
fellow
host
two
or
three
prongs
and
sort
and
post
we
have
the
priority
and
in
the
port
or
network
based,
the
mobile
support
protocol
will
be
used
deep
religious
emoted
and
at
the
repertory
are
illustrated.
Overpass
uppercase
like
like
this.
P
A
A
C
Yeah
so
I
saw
like
Charlie's
new
draft,
so
there's
like
I,
think
still
one
item
left
in
there,
which
is
the
saying
that
this
is
only
in
signaling
messages
but
not
Indian,
applying
messages.
I
think
we
just
need
to
add
it.
So
that's
not
in
three
stacks.
So
there's
like
two
items
like
three
had
to
send
you
some
and
you
had
to
add
it.