►
From YouTube: IETF102-V6OPS-20180719-1330
Description
V6OPS meeting session at IETF102
2018/07/19 1330
https://datatracker.ietf.org/meeting/102/proceedings/
B
C
A
So
this
is
the
agenda
that
we
have
today
and
tomorrow
we
have
three
working
group
drafts.
We
have
George
Michael
son
with
basically
some
marketing
information
about
the
ipv6
market,
which
I
find
where
me
and
Russell.
We
have
frog
Templeton's
draft
tomorrow
morning.
If
we
have
time
I'll
move
that
forward
to
the
end
of
the
meeting
today
and
one
kind
of
will
be.
We
have
three
other
drafts
that
people
have
brought
to
me
within
the
last
week
or
two
weeks
and
asked
for
some
time.
A
A
E
So
initially
I
had
an
ambitious
friends
at
probably
Alisha
publish
it
as
the
cluster
is
a
big
variety
and
multihoming
drug
because
it
has
a
lot
of
justifications
while
slag,
why
not
DHCP
or
ICMP
errors
and
so
on,
but
looks
like
relative
draft,
has
kind
of
stuck
in
the
process,
so
I
think
actually
a
browser
graph
is
now
currently
an
informative
reference.
So
probably
it
doesn't
make
sense
to
wait
for
too
long.
So
I
think
Warren
can
correct
me,
but
it's
not
intelligent
right
for
conditional
arrays.
So
it's
waiting
for
final,
hey
news:
okay,.
A
A
Ones,
ipv6,
and
when
you
do
this
kind
of
a
graph
you're
saying
got,
the
packet
went
all
the
way
from
the
host
through
an
isp
through
another
isp
in
the
destination
network
actually
found
a
computer
in
there
about
the
v6
life
is
what
right
should
there
be
any
more
you
can
path?
It
doesn't
look
like
that.
A
A
G
A
figure
that
says
the
world
capability
is
somewhere
around
18
percent.
That
means
can
not
does
just
means.
Can
it's
based
on
a
sampling
technique?
It's
wet
measurements
using
recruited
ads
it's
about
a
15
million
a
day
sample
of
worldwide.
We
have
to
do
some
adjustments
because
of
the
right
placements
happen.
We
do
some
writing.
There
are
various
issues
with
over
and
under
sampling.
It
goes
to
the
error
rate.
Other
people
put
the
right
corner,
so
there's
a
community
that
are
saying
it's
north
of
20.
G
We
have
a
view
around
how
we
do
an
adjustment
reflecting
the
internet
population
based
on
stats
to
get
reported
back
to
people
like
the
ITU
and
the
UN.
So
we
do
an
adjustment
based
on
our
beliefs
about
users.
You
can
view
it
as
a
low
side
count.
You
can
view
others
as
a
high
side
and
we're
bracketing
things.
I.
Think
that's
not
about
view
anyway.
B
G
Have
a
look
at
the
ranking
of
the
economies
seen
worldwide
in
this
mechanism
above
and
below
18
percent,
so
the
tip
up
said:
if
you
look
above
there,
that's
a
very
interesting
mix
of
economies
that
are
consistently
visible
with
very
high
rates
of
penetration.
Look
at
India,
59%
ipv6
capability.
That's.
B
G
Let's
come
back
to
minute
so
how
much
of
the
world
v6
is
represented
by
each
of
those
economies?
What
percentage
of
population
of
people
who
are
in
the
game
is
in
the
game?
Okay,
so
let's
rescale
it
it's
no
longer
about
how
much
v6
in
an
economy.
So
it's
not
India
is
on
59%
it's
of
the
hundred
percent
of
the
world
of
v6.
What
percentage
of
that
is
coming
out
of
15,
because
after
that,
it
just
gets
a
bit
rubbery?
Okay,
so
who.
G
B
G
B
G
J
G
D
G
B
G
Japan
is
four
point:
seven
and
that's
kinda
cute,
because
there's
been
a
long
time
of
people
saying
Japan
is
not
going
as
fast
as
it
should
do,
but
their
net
contribution
probably
is
actually
quite
high.
Germany
is
four
point
two
and
coming
out
of
nowhere
in
China,
which
is
a
moment
we
might
also
talk
about
and
the
rest
of
it
is
kind
of
a
bit
small.
So
in.
G
B
G
B
B
G
J
J
G
J
L
G
L
B
L
G
G
G
B
G
Same
and
that's
the
principle,
there's
no
rational
story
that
says
or
authority
to
any
economy.
Your
v6
is
fantastic.
No,
sir!
No
such
story,
Oh
II
CD,
it's
similar,
there's
no
clear
signal
is
nothing.
We
came
because
of
membership
of
an
economic
grouping
and
what
I'm
trying
to
say
here
is
there
isn't
a
specific
economy
sector?
You
can
point
to
and
say:
v6
is
only
because
you
are
a
member
of
the
developed
Western
economies.
That's
not
true.
Physics
is
happening
because
v6
is
happening.
The
logic
here
is
around
other
things.
It's
about
your
investment
strategies.
G
G
For
me,
at
least
the
idea
that
there's
a
two-track
internet
there's
people
who
don't
seem
to
be
moving
and
there
are
people
who
have
moved
and
thinking
about
internet
for
a
long
time.
This
is
no
secret.
We've
known
this.
We
talk
about
this
all
the
time,
but
the
thing
is,
it
looks
like
it
really
is
a
thing
and
that's
kind
of
a
Rory,
so
I
kind
of
like
six
I
lied
about
innovation
and
I.
Think
incumbency
I've
got
my
assets.
I've
got
my
v4
I.
Don't
need
to
worry!
Nothing
can
happen
to
me.
G
Well,
you
know
stuff
can
happen
to
you
and
that's
what
it's
coming
up
next,
okay!
So,
let's
drill
down
into
these
economies
a
bit,
let's
look
at
them
inside.
Let's
look
at
various
statistics,
which
is
an
approximation
for
eyeballed
share
based
on
our
random
distribution
model,
and
let's
see
what
things
look
like,
so
there's
a
good
chance,
I'm
going
to
put
it
up
side
by
side.
This
is
mecco
chart
and
the
vertical
size
is
how
much
we
think
we
saw
of
you
and
the
horizontal
dark
is
how
much
of
you
we
saw,
six
capable.
G
B
G
Minimum
data
planner
and
if
you
look
at
the
orange
line
that
is
reliance
and
that
is
their
scaling
or
v6
capability
and
they
basically
hit
80
to
90
percent
penetration,
pretty
much
inside
the
first
year
of
deployment
and
they
single-handedly
spoke
in
India
into
world
rankings.
But
then,
if
you
look
across
the
line,
there's
this
weird
moment
where
Bharti
and
Hutchinson,
which
is
actually
two
ASN,
it's
ICL
and
it's
Hutchinson
Telecom-
actually
reacted
in
time.
G
So
if
you
look
across
at
this
there's
a
time
delay
and
then
some
other
people
say
I
think
we
might
compete
in
this
market.
Space
I
think
we
might
do
something
here
and
they
actually
started
deploying
v6
now
I'm
not
going
to
say
it's
an
economic
reaction,
I'm
not
going
to
say
its
competition
in
they're
competing
in
v6,
but
there
are
certainly
signs
that,
after
after.
G
G
G
It
was
not
a
moment
because
v6
does
not
figure
in
selling
telephony
to
the
Indian
masses.
What
figures
that
the
Indian
masses
is
I
can
bring
my
mother
for
free
and
I
get
a
data
plan
so
to
that
question.
But
what
meaningful
percentage
of
these
users
are
doing?
V6
will
have
to
tell
you
if
its
reliance,
all.
D
G
Them
all
of
them,
because
it
is
a
pure
state,
v6
solution.
If
your
one
line
in
India
and
you're
using
reliance
and
PS
50%
50%
of
the
market,
looks
to
be
there
you're
using
v6
I
would
say
it's
pretty
solid.
Okay,
this
is
America.
Now
this
is
a
weird
story.
I
mean
look
at
this
line.
You
have
some
only
movement
from
some
is.
Is
that
basically
hit
90%
de
Mond
and
stay
there
and
you
have
Comcast
doing
a
kind
of
disordered
rise
up
towards
a
full
state?
G
Although
there
are
some
questions
around,
why
is
it
not?
A
hundred
percent,
because
we
know
comcast
is
meant
to
be
fully
deployed
thing
is
that
they
have
some
legacy
gears
they're
never
going
to
get
there,
and
they
also
have
some
issues
around
commercial
service.
That's
hidden
behind
the
AAS,
that's
also
providing
domestic,
and
we
all
know
that
the
predominant
usage
going
on
here
is
essentially
domestic
use.
So
there
are
kind
of
stories
there
Jerry
ed
hi,.
I
I
G
What
I
wanted
to
go
with
this
diagram
is
that
there
is
actually
quite
a
lot
of
competitive
tension
in
who's
got
eyeballs
in
America.
It's
not
a
one
AAS
story
right.
This
is
not
Reliance.
This
is
a
different
economy
with
a
different
pattern
of
competitive
tension.
In
the
left
hand,
box
mecco
chart
there's
different
levels
of
deployment,
but
there
is
six
one
across
the
story
for
the
USA
and
at
different
times,
different
players
have
decided
to
deploy.
It's
happened
to
an
interesting
place.
Okay,
this
is
the
Brazil
story.
G
What
I've
been
told
by
Carlos,
my
friend
from
the
lac
community,
is
that
there's
actually
a
national
agenda
and
what
you're
seeing
here
is
a
concerted,
no
in
a
federated
structure
for
one
of
the
players
to
move
the
market,
broadly
speaking
in
concert.
So
this
is
a
number
of
large
cable
deploys
in
large
capital
population,
Sao
Paulo
Rio,
the
state's
making
their
move.
The
interesting
moment
there
is.
It
includes
telefónica-
and
here
you
have
telefónica.
G
G
G
F
F
Id
needs,
and
so
on.
So
if
I
try
to
look
at
the
differences
between
RFC
7080
for
and
this
document,
there
are
some
new
general
requirements.
One
of
them
is
supporting
RFC
76
0a,
the
top
one,
of
course,
is
newer
than
70
84.
So
that's
that's
the
reason
for
that
in
the
LAN
we
are
now
requesting
as
a
mast,
the
DNS
proxy.
This.
B
F
One
of
them
is
that
the
routers
must
support
RFC,
80
26,
which
is
a
prioritization
mechanism
for
selecting
which
transition
mechanism
rater
want
to
enable-
and
this
is
basically
done
in
order
to
allow
them
to
move
from
dual
stack,
for
example,
to
an
ipv6,
sorry
ipv4
as
a
service
transition
mechanism
and
so
on.
So
this
has
been
discussed
in
the
mailing
list
and
we
find
a
way
to
without
any
new
DHCP
options,
just
using
existing
options.
To
do
that,
we
also
asked
the
CP
to
support
graphic
user
interface
and
or
CLI.
F
M
F
It
goes
back
or
not
that
one
seven
six
away.
Yes,
we
we
are
just
saying
that
the
length
of
the
of
the
prefix
is
a
recommendation
of.
If
the
router
want
to
use
something
different
in
the
lungs,
it's
it's
possible
to
that's
what
they
document
I,
think
it
says:
I,
don't
recall
right
now.
It
specific
details,
but
I
think
was
that
so
it's
a
parameter.
Basically
the
perfect
length
right.
F
M
M
F
M
F
M
Human
right,
sorry,
you
continue
in
routing,
you
want
to
be
able,
but
it
must
never,
because
we,
because
we
have
requirements
at
least
an
element
that
says
that
we
should
at
least
get
a
64
to
them
to
go
to
the
home,
possibly
more
yeah
and
our
or
guidelines
around,
like
delegate
at
least
a
block
of
many
64.
And
so,
if
we
do
this
and
that's
like
entirely
contradictory,
okay
I
think
it.
M
F
E
Yeah
I
just
turned:
if
you
think
we
should
have
it,
then
it
would
be
nice
to
document
the
use
case
because
I'm
thinking
of
CPA
about
side
as
device
which
has
a
pink
in
the
future
SP
in
some
local
networks.
It's
not
I
can't
understand
why
the
device
must
have
routes
in
the
routing
table
which
will
be
like
a
funny
prefix
legs.
It
most
likely
wouldn't
be
the
case
right,
so
that.
E
M
A
M
A
M
M
M
M
Had
a
question
about
this
document
that
I
remember
somebody
actually
I
think
it
was.
It
was
Barbara
observing
its
bring
a
pressure.
This
document
says
you
should
implement
at
least
five
different
transition
mechanisms
for
V,
436
and
I
recall
her
position
on.
That
was
that
that
would
increase
costs
for
CP
manufacturers
and
which
are
already
particularly
sort
of
low
margin
operations
and
would
either
result
in
nothing
or
very
buggy.
M
F
Think,
well,
we
have
two
of
the
routers
of
the
draft,
our
CP
vendors,
and
they
decided
that
this
is
not
actually
that
much
extra
cost
and-
and
they
have
already
implemented
it,
and
the
main
problem
is
that
if
you
are
a
big,
a
big
ISP,
a
recuperator,
you
go
to
the
CP
vendor
and
they'll.
Tell
them
I
need
this,
but
the
small
ones
cannot,
and
the
other
thing
that
is
already
in
one
of
their
nexus
of
the
document
is
that
the
data
playing
is
the
same
for
all
of
them.
F
M
M
F
N
My
view
on
how
this
gets
used,
you
know
from
the
pragmatic
perspective
once
it's
published,
is
you've
got
basically
two
uses
for
it.
One
is,
you
know,
I'm
perfectly
happy
to
like
pick
and
choose
you
know
if
there's
something
I
want
from
my
vendors
to
say:
do
this?
Do
this
and
just
kind
of
you
know
I'm
not
expecting
you
to
be
compliant
with
everything.
This
is
all
I'm
interested
in.
That's
fine!
Now
then,
there's
another
set
which
is
kind
of
your
run-of-the-mill
CA
vendors.
N
Now
they
can
read
this
and
say
you
know,
I'm,
really
not
willing
to
do
all
that
and
that's
fine
all
it
means
is
they
cannot
come.
They
cannot
claim
compliance
to
the
specification
when
there's
the
ones
who
say
are
ever
going
to
do
all
of
this,
in
which
case
they
can,
they
can
claim
compliance
to
the
entire
specification,
so
we're
extra-special,
good
and
I
think
that's
a
really
good
thing
to
be
able
to
say,
and
so
to
me
it's
kind
of
I'm
perfectly
happy
with
it.
Thanks.
M
M
F
History
of
the
document
I
started
this
as
an
outcome
from
a
discussion
in
in
the
IDF
mailing
list,
because
there
was
a
comment
that
we
should
have
only
a
BB,
six
and
not
64,
and
then
I
suggested
for
six
Forex
lot
in
that
case,
because
there
are
some
applications
still
that
will
need
that
in
the
discussion.
Then
comes
the
point
about
dns64
that
may
break
the
NSA
and
I
think
it
was
Christine.
F
You'd
Emma
said
we
should
never
had
dns64
and
we
can
perfectly
use
forces
for
exact
without
the
NS
64
and
in
fact
it's
something
that
I
am
trying
in
some
big
trials,
which
some
operators
so
I
documented
that
in
in
document
considering
at
that
time,
only
for
six
projects
lat
and
it
was
presented
in
ITF
hundred
and
then
the
working
group
suggested.
Let's
go
for
an
overall,
not
64
deployment
guidelines
document.
F
So
that's
that's
what
I
did,
unfortunately,
in
the
last
IDF
in
London,
the
presentation
was
very,
very
quick
within
her
time
and
I
didn't
got
almost
comments,
but
anyway,
in
the
last
months
we
had
some
clarifications
in
the
list
and
some
points
that
I
want
to
note.
This
is
not
the
discussion,
because
some
comment
said
we
say
that
I
think
in
the
list
or
in
private.
This
is
not
a
discussion
about
not
64
versus
other
transition
mechanism.
F
This
is
not
a
discussion
about
the
intersect,
but
it's
important,
and
there
is
a
lot
of
text
in
the
document
to
reflect
that
and
what
this
document
is
trying
to
do
is
if
you
already
decided
to
go
for
something
based
on
not
64,
then
here
are
your
choices
in
section
3,
which
I
call
the
scenarios
and
then
what
you
should
take
care
of
to
deploy
those
scenarios,
which
is
section
4.
So
basically,
the
document
has
an
introduction
which
talking
about
three
main
issues,
which
is
the
intersect,
may
break
DNS.
F
Sorry
dns64
may
break
DNS,
SEC
nat64
dns64,
don't
work
with
which
literals
and
alder
IP
is
not
sista
for
alone
does
not
work
for
I
big
for
only
hostel.
Replications
will
not
all
knew
that,
but
the
people
that
is
going
to
read
the
document,
maybe
not
so
we
need
to
make
sure
they
understand
that
situation.
So
looking
at
the
scenarios
but
I
have
done,
is
looking
at
what
new
scenarios
have
been
coming
to
the
mark
and
since
dns64,
whose
publish
I
mean
that
document
they
took
it
basically
about
only
three
scenarios.
F
Now
we
have
new
ones,
including
all
those
based
on
for
six
Forex
lot
to
comment
and
I
am
looking
at
the
document
from
the
perspective.
If
I
am
an
operator
or
an
enterprise
network
and
I
want
to
deploy,
nat64
I
really
need
to
make
sure
that
it
works
for
my
user
cases
right,
so
I
really
need
to
go
away
if
something
is
going
to
be
broken.
F
So
what
is
not
the
work?
We
have
a
service
provider
in
a
64
that
has
also
dns64
and,
of
course
those
two
functions
can
be
in
their
network
outsource
outsourced.
Then
we
have
the
possibility
of
464xlat
which
the
under
64
and
forces
for
each
lat
without
the
NS
64.
So
those
are
the
three
main
scenarios
and
then
we
have
other
three
scenarios
that
we
work
only
under
certain
conditions.
I
think
they
are
still
valid
because
in
some
cases
they
they
may
be
useful.
Okay,
so
that's
the
reason
I
put
them
in
the
document.
F
So
there
is
a
service
provider
that
provides
not
64
without
the
NF
64,
a
service
provider
that
has
not
64
and
is
assuming
that
the
host
had
the
dns64
functionality
and
then
a
service
provider
which
DNS
service
not
64
and
the
dns64
is
in
the
remote
network.
Okay,
I'd
be
the
completion
of
dohsa
scenarios
by.
M
The
way,
let
me
back
up
one,
if
you
would
let
me
report
on
a
conversation,
I
had
a
dinner
last
night.
I
think
we
came
up
with
one
more
scenario,
and
that
is
that
you
have
a
user
that
is
configured
their
host
to
use
for
their
DNS
service,
some
public
service,
Open
DNS.
We
will
take
your
back
one
of
them,
and
so
when
they
go
looking
for
the
ipv4,
only
thing
they're
gonna
go
to
that.
The
service
provider
has
her
own
DNS
and.
M
M
M
F
This
is
one
of
the
scenarios
that
I
have
consider
it.
The
point
that
I
was
going
to
come
in
right
now
is
that
even
if
I
have
only
six
main
scenarios,
each
of
the
scenario
has
different
possibilities
or
super
scenarios,
so
in
total
I
come
out
to
twelve
possible
scenarios,
and
some
of
them
include
I
believe
if
I
understood
you
correctly,
what
you
are
commenting.
D
F
F
F
F
So
if,
for
example,
if
I
change
the
DNS
I
call
that
following
DNS
in
so
messin
areas
doesn't
matter,
for
example,
if
you
are
prepared,
you
have
prepared
your
network
to
not
use
the
n64,
but
if
the
network
use
DNS
64
and
you
choose
what
one,
what
a
white,
nine
or
whatever
of
the
popular
being
a
service
will
not
work.
Okay,
so
I
am
I,
am
looking
into
that.
Then
I
look
into
the
other
possible
issues.
One
of
them
and
a
big
section,
of
course,
is
DNS
SEC
considerations
and
possible
approaches.
F
So
I
am
looking
at
not
using
DNS
64
DNS
SEC
validator,
aware
of
dns64
estoppel
I
later
select
with
DNS
proxy
ACL
of
clients.
Mapping
out
in
for
clients,
I
look
also
into
dns64
reverse
mapping.
Considerations
I
am
looking
also
into
using
4
6
4
X
lat,
which
and
without
the
inner
system,
for
what
are
the
implications
of
each
case.
I
am
looking
also
of
what
I
just
mentioned.
If
the
users
are
changing,
the
DNS
good
happens,
questions
about
DNS
privacy
like,
for
example,
DTLS,
quick
and
so
on.
F
I
look
goes
into
well-known,
graphics,
Bursar's,
Network,
specific
graphics,
usage,
Iping
for
literals
and
IP
is
IP
for
only
host
or
applications.
Silla,
translation
considerations,
then
I
have
a
summary
of
the
deployment
recommendations.
I
have
also
a
section,
a
small
section
that
tells
basically
the
scenarios
reported
here
are
generally
the
same.
If
you
are
an
ISP
or
you
are
a
big
enterprise,
Nathan
a
small
network
mean
basically
the
the
consequences
of
how
you're
doing
is
are
the
same
and
then
I
had
three
annex.
F
One
of
them
is
an
example,
a
specific
example
of
deploying
for
six
Forex
lat,
which
and
without
dns64
in
broadband.
That
was
an
input
from
fret.
He
want
to
know
some
details
about
how
can
be
implemented
silat,
and
then
somebody
suggest
that
why
not
having
some
details
about
benchmarking
I
am
not
really
sure
yet
about
that
section,
because
we
may
not
have
enough
data
to
do
it
correctly
and
I
think
that's
that's
it
basically.
M
M
Don't
see
Fred
temple
he's
scheduled
for
tomorrow,
but
I
said
last
time
and
this
time
that
if
we
wound
up
with
time,
we'd
bring
him
forward.
Given
what
he's
not
here,
Lorenzo
you
have.
You
actually
asked
for
this
slap.
Can
I
put
you
to
work.
A
M
E
E
E
E
Actually,
I
dunno.
If
there
is
a
not
six
for
translator
in
writing.
I
know
the
pile
is
around
the
flavored
somewhere
when
it
could
be
translated
or
not
so
network
is
actually
the
proper
source
of
getting
this
information
and
to
run
any
other
services,
because
I'm
doing
that
six
phone
I
do
not
want
to
run
any
additional
services
which
I
need
to
support
just
to
native
right
host
about
not
six
four
prefix
and
he
has
amazing,
isn't
really
leveling
a
trick
of
asking
for
before
only
name
getting
quad
a
bear
can.
E
So
using
DNS
first
and
throughout
the
response
to
used
in
a
secretary,
it's
kind
of
some
string
to
me,
the
next
likely
so
your
option
very
simple
type
lands
right
time.
Sixteen
beads,
which
similar
has
the
same
value
of
default,
router
or
lifetime
I,
mean
have
the
same
range
and
ninety-six
bit
of
previous.
So
next
slide,
please
RFC
for
translation,
actually
says
that
prefixes
might
be
of
different
lengths,
and
it
specifies
six
values
right,
but
it
looks
like
use
case.
E
E
E
E
A
M
M
M
E
E
Don't
see
you
seven
to
the
five
described
in
scenario
when
you
might
want
to
map
different
before
address
space
to
different
prefixes
and
signal
it
to
host
again,
it's
significantly
increased
complexity,
so
I
I
know
well
network,
which
did
that
and
I
talked
to
them
yesterday
and
they're
like
yeah.
Alright,
we
actually
can
do
it
without
you
actually
utilize
routing
to
do
this
so
again.
E
M
Answer
the
96
question
from
David's
Kazi
Apple.
From
our
perspective,
our
stock
supports
legs
different
from
96,
but
we've
never
really
tested
them
and
as
far
as
I
know,
when
I
confront
some
people,
no
one's
actually
deployed
that
so
I
think
fixing
the
length
296
for
this
use
case
sounds
like
there
is
a
both
going
to
me
personally.
E
F
Jordi
ballot
I
read
the
document
right
before
the
session
I
liked
it
I
regard
in
the
discussion
of
those
last
96
the
analysis,
the
different
practices
than
96.
Maybe
it's
it's
also
a
possibility
to
to
date.
The
the
document,
the
DNS
I,
don't
remember-
is
this
1750
document
the
1750
document
about
that.
E
F
E
People
also
want
to
get
this
option
in
re,
because
if
it's,
for
example,
a
datacenter
when
I
can
imagine,
they
might
want
to
use
this
theoretically,
they
might
not
want
to.
Is
this
re
option
because
every
single
big
provision
statically
so
that's
a
worry
about?
Would
it
would
vary?
The
prefix
can
be
useful
in
scenario
when
you
want
an
array
option:
okay,.
F
The
other
common,
a
heart
I
have
been
reading,
but
during
the
session
was
difficult
for
me
to
do
both
things.
At
the
same
time,
I
was
reading
1751,
which
is
the
one
that
that
analysts
do
the
analysis
of
the
different
options
for
the
discovery
and
I
think
in
that
document
they
were
comparing
the
array
option
with
the
dhcpv6
option,
so
I
am
going
to
probably
later
on
tonight.
Send
some
comments
about
if
I
find
something
that
it's
really
important
to
comment
or
to
put
in
that
document.
F
E
M
M
M
In
the
day-
and
they
say
you
don't
wanna-
you
want
to
ignore
those
Corti
records
and
actually
do
synthesis
to
the
day
for
ignore
particular,
very
full
record
before
addresses,
and
not
the
synthesis
and
those
things
that
sort
of
information
also
needs
to
get
out
there.
In
addition
to
the
actual
prefix.
M
In
eighteen,
but
if
you're
running
a
different
before
Network,
you
don't
want
to
translate
your
onion
kernel
right.
It
dresses
you
you're,
using
inside
the
network
to
quote
I
and
stuff
like
that
so
I'm.
What
I'm
saying
is
there
are
other
addresses
where
you
don't
want
to
do
things.
You
don't
want
apply
translation
to.
E
Mind
is
to
increase.
There
are
some
rules
for
doing
a
six
for
how
a
synchronous
should
be
done
right
and
it's
written
in
the
RFC,
so
I
would
expect
that
local
resolver,
which
follows
the
same
rules.
So
there
is
no
difference
if
that
information
is
in
the
logic
of
Google,
Public,
DNS
6-4
or
your
local
resolver
and
again,
for
example,
directly
this
when
I
do
want
to
have
my
private
address
space
to
be
translated
because
I'm
deploying
v6
on
your
local
network.
So.
M
Never
offended
can
I
borrow
marks,
comment.
David's
cannot
see
Apple
speaking
as
after
of
our
CD
three
or
five,
which
is
happy
eyeballs,
be
to
how
to
handle
this
on
house.
I
suspect
that
when
you're
describing
or
some
networks
that
have
these
things
today,
those
networks
do
not
have
a
way
to
communicate
that
to
hosts
and
whatever
Android
and
iOS
does
is
broken
on
all
those
networks.
Today,
I,
don't
think
that's
something
we
need
to
fix
right
now,
I,
don't
know
if
those
networks
exist.
M
If
you
want
to
fix
that
problem,
I
think
that
would
be
should
be
in
its
own
draft,
because
today,
when
Android
will
do
is,
if
he
wants
to
talk
to
that
before
address
it'll
synthesize
it.
The
network
has
no
way
of
telling
that
I
droid
device,
not
4
6
4,
X
Y
or
the
siraat,
but
only
it
only
applies
to
the
regarding
the
porn
that
this
is
trying
to
serve,
which
is
the
common
deployment
of
nat64
and
cellular
networks.
This
really
solves
a
problem.
B
M
Me
also
suggest
a
solution
for
this
I
believe
the
solution
would
be
if
your
network
has
v6
only
then
you
delegate
all
of
this
to
the
network.
If
you
have
no
before
locally
know,
you
were
a
resolver
you're
in
the
middle.
You
don't
know
what
you
have
you
don't
know
what
the
client
has
but
I'm
at
the
edge.
If
I
don't
have
before,
then
everything
has
to
be
translated.
End
of
story.
So
basically
we
could
say
this
is
only
gonna
work
ever
if
you
have
a
v4
address,
disregard
this,
no,
no
jewel
jewel,
step.
M
N
M
K
M
Think,
as
at
me
as
an
author,
I
guess,
I
would
say
we're
gonna
go
ahead
and
basically
say
you
can
only
use
this
if
you
have
no
before
on
that
network
and
if
you
grow
a
before
address
you
turn
it
off.
I
mean
that's
what
we
do
today,
if
we,
if
we
basically
grow
before
addressing,
if
we,
if
there's
a
network
that
somehow
provides
both
that
6,
4
and
v4,
if
v6
only
wins
the
race,
we
turn
our
Pratt
and
then
v4
address
appears.
We
turn
it
off.
So.
A
M
E
Only
use
case
I
know
when
people
using
multiple
prefixes
and
using
I'm
Martin
this
before
the
respect,
resist
v6,
prefix
and
another.
If
part
of
my
before
3
space
and
another
prefix.
It's
because
my
islands
do
not
have
before
connectivity,
but
there
could
be
six
connectivity
and
I
want
to
make
sure
my
traffic
goes
to
the
right
Island,
but
it
basically
means
you
can
have
more
specific
than
/
96
in
your
v6.
Now
eating
terrible
and
route
and
Union
still
the
same
six
prefix,
but
her
most
bits.
If
it's
an
erotic,
dream
yeah.
H
The
internet
servers
on
on
your
gentle
as,
for
instance,
and
then
you
are
configuring,
two
distinct
964
prefixes
for
those
and
then
when
you
are
doing
the
other
cities,
you
need
to
to
be
sure,
though,
in
the
packet
are
seen
by
your
first
are
ending
on
the
appropriate
instances,
are
not
the
wrong
one.
So
that's
why,
in
the
7225
there's
August
discussion
about
this
nation
today
and
she's,
also
discussing
in
the
RFC
1951,
which
this
recommend
again
the
option,
also
the
eh-eh
Europe's.
No,
she
was
also
proposed
by
Jordan
and
so
on.
So
I
think.
H
That's
the
problem.
The
little
bit
context.
It
is
not
just
this
position
is
not
only
there
for
fan
it's
there
for
to
some
of
the
fraud
networks
and
after
three
during
the
specification,
the
negative,
for
instance,
if
you
wanted
to
the
domestic
c4
with
something
which
is
and
chicks'll
neutral,
and
that's
when
you
need
to
have
something
which
is
the
static
spurt
in
order
to
notify
what
you
are,
you
need
the
perfect
size.
H
So
that's
why,
in
the
RFC
7225,
we
have
this
kind
of
specification
there,
so
that
if
someone
in
the
world
I
don't
know
who
really
want
to
have
in
Essex
in
a
sixty-four
function
which
ensure
and
preserve
the
chicks'll
neutrality,
it
can
do
that
with
the
specification
so
with
a
depth
only
for
I
would
say
for
found
it's
something
which
is
motivated
there
and
there
are
multiple
Arabs
who
analyze
all
of
this
stuff,
so
I
think
that's.
We
have
to
be
really
careful
about
revising
the
consensus
that
we
have
achieved
in
the
ATF.
H
M
That
how
that
differs
from
routing
more
v6
specifics
on
your
network,
because
in
Europe,
so
so
you
basically
the
v6
there.
That's
it
for
prefix
gets
stripped
when
you
hit
the
not
explore
right
yeah
right.
So
the
problem
is
that
you
want
to
reach
for
different
before
destinations
different.
That's
it's
worse!
Right!.
B
M
M
Contiguous
blocks
right,
so
those
continuous
blocks
take
each
one
of
them.
There
also
considers
blocks
in
v6
space
because
the
things
at
the
end
right
so
they're,
contiguous
blocks
and
v6
trees.
Therefore,
if
you
only
support
the
theory
for
address
at
the
end,
that's
like
all
your
contiguous
before
blocks
are
contiguous
blocks
in
v6
space
and
therefore
you
can
just
move
those
blocks
in
v6
space
wherever
there's
36
packets
to
the
right
nat64.
M
A
H
You
receive
a
sip
communication
in
your
sip
communication.
You
have
this
DP
bar
in
this
part.
You
have
some
the
connection
line
in
which
you
it
there's
an
IP
address
there,
and
it
happened
that
the
ipv4
address,
which
is
in
this
DP
far,
is
your
internal
proxy
server,
and
you
want
to
signify
this
ipv4
address,
to
transform
it
into
an
ipv6
one
so
which
ipv6
prefix,
that
you
knew
to
do
that.
H
B
H
H
To
transform
it
with
a
bit
of
radish
into
a
net
physics
one
and
the
hacker
is
that
I
should
not
send
that
ipv6
packet
into
my
gentle
face,
but
sent
it
to
an
internal
IP
v4
proxy,
which
is
in
my
in
my
behind
another
nat64
instance.
But
I
don't
underst
what
yes
I
get
the
definition
of
about
the
destination
gains,
not
64.
So
that's
good
to
make
sure
that
we
are
talking
about
the
same
problem.
But
the
network
can
take
care
of
that.
If
the
prefix.
M
Ok,
so
thank
you
again
for
Templin.
You
walked
into
the
room,
I
asked
for
you
earlier
and
you
are
out
of
the
room
so
question.
We
can
do
your
talk
tomorrow
morning
and
you
can
have
longer
time
what
we've
got
left
in
this
slot
right
now
is
15
minutes.
Can
you
do
your
talk
in
15
minutes
and
do
it
now
or
do
you
want
to
give
it
tomorrow
morning.
M
M
So
there's
many
different
British
delegation,
alternatives.
Dhcpv6
is
the
primary
example,
but
there
are
other
examples,
including
the
PIO
exclusive,
provide
through
IP
MS
network
management,
the
static
configuration
we
want
to
look
at
the
classic
routing
model
when
the
node
that
provisions
the
delegate
predicts
the
downstream
attached
networks.
For
example,
it's
heathered
Internet
of
Things
or
internal
network
of
virtual
machines,
etc.
M
Multi
addressing
host
models
when
the
node
uses
the
delegated
prefix
for
its
own
internal
multi,
addressing
purposes
so
in
case
one,
the
classic
Ronnie
model.
Here
we
have
the
requesting
router
receiving
a
delegated
prefix
that
distributes
to
its
downstream
attached
networks.
Again,
it's
useful
for
Internet
of
Things
example
would
be
a
cell
phone
with
feathered
external
Network,
for
example,
Bluetooth
devices,
a
laptop
with
an
internal
network
of
virtual
machines
or
a
home
network
router
with
home
devices
in
the
home
network
case
to
multi,
addressing
on
virtual
interfaces.
M
In
this
case,
the
requesting
node
assigns
addresses
from
delegated
prefixes
to
internal
virtual
interfaces
like
a
loopback
without
invoking
multicast.
They
listen
or
discover
your
dad
on
that
stream
interface.
This
would
allow
the
known
to
have
unlimited
numbers
of
addresses
available
an
example
that
that
can
do.
This
is
any
host
with
an
internal
virtual
interface
on
which
addresses
can
be
assigned
examples.
M
A
loopback
again,
our
case
3
multi,
addressing
on
the
upstream
interface,
so
the
requesting
node
assigns
addresses
from
the
delegated
prefix
to
an
upstream
interface
of
what
should
receive
that
perfect
delegation
again
without
invoking
monthly.
That's
live,
will
sort
of
discovery
or
dad
unlimited
numbers
of
addresses
get
are
available
and,
for
example,
any
host
that
cannot
assign
addresses
to
any
other
interfaces.
Besides,
the
upstream
interface
so
remember,
addresses
are
assigned
the
interfaces.
M
If
the
only
interface
you
are
able
to
assign
the
dresses
on
is
the
upstream
interface,
then
that's
what
you
have
to
do
and
finally
case
for
an
application
addressing
here.
The
requesting
node
assigns
addresses
from
delegated
prefixes
to
its
local
applications,
where
each
application
acts
like
a
virtual
interface
for
address
assignment,
as
required
by
RFC
4291
again,
a
limited,
multiple
numbers
of
addresses
are
available
and
it
provides
a
new
model
with
the
distinct
benefit
that
a
unique
address
per
application
may
obviate
the
need
for
port
numbers
in
the
future.
That's
a
very
important
distinction.
M
Chain
since
ITF
101,
we
clarified
that
the
classic
routing
model
applies
to
both
external
physical
networks
and
internal
virtual
networks.
We
drop
the
discussion
of
weak
and
strong
in
system
because
several
comments
pointed
out
that
it
may
not
be
applicable
in
this.
In
this
context,
we
added
discussion
on
address
per
application,
a
discussion
on
in
relation
to
RC,
79,
34
and
82-73,
and
we
received
most
comments
on
six
fifteen
and
twenty
eighteen
and
posted
proposed
resolutions.
M
A
couple
days
later,
we'll
be
folding
those
resolutions
into
the
next
draft
version
draft
history,
the
first
version
this
draft
was
posted
in
November
2015
announced
the
v6
ops.
Several
rounds
of
comments
came
from
a
review
team
that
looked
at
that
document.
We
presented
15
at
ITF
101
again.
We
had
working
with
comments
and
lists
comments
on
that
draft.
19
was
presented
in
IETF,
whirl
1
or
in
that
presentation,
ITF
101.
It
was
a
virtual
presentation
with
no
time
for
questions
afterwards.
M
M
Gregg
Fred:
what
is
the
real
purpose
of
this
graph
is
documenting
use
case?
Are
you
proposing
use
it's?
It's
documented.
They
the
the
cases
of
what
can
be
done
with
the
delegated
prefix,
the
classic
routing
model,
the
downstream
interface
assignment,
the
upstream
interface
assignment
and
the
upper
address
assignment.
M
M
Of
those
people
who've
read
the
draft,
could
anybody
maybe
have
a
home
and
the
two
options
will
be?
If
you
think
it's
useful
and
we
should
progress,
it
maybe
think
about
adopting
it
home
and
then
the
second
homeowner
ask
is,
if
you
think
it's
dangerous
and
we
should
not,
then
oh.
So
this
is
the
first
question
home.
If
you
think
it
is
useful,
we
should
consider
progressing
it.