►
From YouTube: IETF106-V6OPS-20191118-1550
Description
V6OPS meeting session at IETF106
2019/11/18 1550
https://datatracker.ietf.org/meeting/106/proceedings/
A
A
A
Okay,
a
couple
of
housekeeping
things:
my
co-chair
is
home
pneumonia.
So
if
you
see
people
that
wherever
it
is
the
screen
that
shows
people
that
are
trying
to
get
a
word
in
edgewise
from
remote
places
and
I,
don't
see
it
do
something.
Let
me
know
that
that
there's
somebody
who
wants
to
do
that
and
what
what
screen
is
that
on
that
should
be
on
one
of
them
hi
there
you're
looking
at
me
like
okay,
yesterday,
I
asked
for
a
jabber
scribe
and
a
note-taker
I
have
gotten
Herbert
you're
willing
to
do
notes.
A
A
A
C
C
A
D
B
F
B
B
It
gets
router,
IP,
address
and
MAC
address,
so
host
start
singing
a
lot
of
traffic
to
different
destinations
and
that
point
of
time,
router
successful,
you're
out
all
those
packets
outside,
but
the
router
still
does
not
have
any
neighbor
cache
entry
for
the
host
global
address,
because
the
router
doesn't
have
any
reason
to
then
traffic
return.
Traffic
starts
coming
back,
router
caches,
one
packet
drops
the
rest
of
the
packet
and
starts
neighbor
discovery
process.
B
The
resolution
process
for
the
host
global
address,
which
leads
to
some
packets
being
dropped
and
if
host
implement
some
kind
of
program,
companies
with
some
exponential
back-off
hosts,
might
actually
consider
the
network
to
be
unreachable.
For
example,
if
you
and
I
hope
you
I
actually
using
v6
on
the
network
here,
you
might
notice
that,
for
example,
Android
phones,
it
takes
five
to
seven
seconds
before
it
will
show
you
that
Wi-Fi
has
connectivity
actually
because
of
this,
because
under
o
it
sends
probes.
Probes
are
not
coming
back
until
neighbor.
Discovery
completes
the
next
part.
B
Next
probes
are
sends
five
seconds
later
and
so
on
so
yeah
in
ideal
world.
It
would
be
nice
if,
by
the
time
the
returning
traffic
is
coming
back.
Router
already
has
some
information
about
the
host
global
address
right.
So
we
avoid
this
by
design
packet
loss
so
and
actually
in
before
everything
works.
Fine
because
host
sends
our
advertising
say
before
addresses
routers
put
that
ipv4
address
and
they
are
cache
and
everyone
is
ready.
So
in
duels
that
network
nobody
noticing
a
problem
because
of
high
tables.
B
So
as
soon
as
you
turn
over
for
your
customers
and
your
managers
are
come
very
on
Hagen,
so
we
discussed
it
last
time
draft
document
was
adopted
and,
as
per
discussion,
we
had
I
had
split
the
draft
in
two,
because
this
is
about
a
correctional
problem.
About
the
solution
actually
in
the
six-month
area,
because
apparently
we
need
slightly
tweaked
neighbor
discovery,
so
the
current
version
of
the
draft
still
describes
what
the
problem
is
and
enumerates.
B
Various
approaches
to
solution,
and
the
proposed
solution
is
now
described
in
the
draft
which
I
posted
six-man
and
I
also
moved
security.
Consideration
section
to
the
six-month
graft,
because
I
think
security
consideration
talks
about
the
proposed
solution,
so
it
should
belong
to
that
draft.
So,
basically,
and
also
the
draft
suggests
that
host
advertise
in
addresses
in
Ansari,
Statten
ace,
routers
received
Ansari
sturdiness
and
create
incomplete
cache
entry.
I
think
eager
suggested
that
instead
of
learning
from
unsolicited
ace,
routers
could
also
learn
from
Dodds.
B
Just
looking
at
all
that
packets
received
it's
one
of
the
potential
ways
to
solve
the
problem,
however,
I
suspect
the
problem
is
that
now
routers
needs
to
receive
all
that
traffic
which
not
necessary
one
routers
to
receive,
and
also
there
might
be
kind
of
race
conditions
there,
because
host
sends
that
packet.
Router
receives
it
that
packet,
oh
I,
need
to
get
a
new
host.
B
What
routers
should
do
routers
and
I'm
not
sure
here
what
to
say
we
can
say
they
may
or
they
should
populate
their
caches
upon
the
receiving
unsolicited
A's
and
actually,
if
you
keep
it
as
male
and
it's
not
even
any
changes
to
neighbor
discovery,
48
61,
because
the
current
version
of
the
RFC
says
unsolicited
an
ace
should
not
create
any
new
cache
entries
and
I
read
through
not
as
actually
may
if
needed.
Oh,
probably,
it's
a
good
idea
that
routers
should
create
and
rest
upon
receiving
unsolicited
analysis
actually
subject
to
discussion.
B
All
we
can
have
this
discussion
in
6
min
on
Thursday,
where
I
have
15
minutes.
So,
basically,
not
so
many
changes
to
the
draft
just
splitting
into
and
discussing
gleaning
from
duplicate
address
detection
packets
instead
of
unsolicited
an
ace
so
yeah
and
plus
yes,
I
said
two
sections
move
to
six
month
after
security
consideration
and
how
we
can
ensure
that
no
disruption
opinion
in
case
of
duplicate
addresses.
So
it's
actually
it,
and
it
would
be
nice
to
hear
some
comments,
suggestions.
What
shall
we
do.
B
G
So
I
was
one
of
the
people
that
have
not
read
your
draft.
So
thank
you
for
the
quick
introduction
now
that
you
gave
that
was
useful.
You
mentioned
that
you
added
other
options
and
so
I
don't
know
what
things
you
didn't
talk
about,
that
you
might
include,
but
I
think
both
of
the
ones
that
you
presented
have
various
disadvantages
and
I
was
gonna.
G
You
also
have
issues
with
all
these
that
say
well.
Does
this
mean
that
if
you
we're
running
out
of
cash
entries,
you
might
kick
something
out
in
order
to
put
something
in
which
you
were
then
never
use
in
the
future?
That's
just
an
operational
issue.
You
should
discuss
that
will
happen
with
no
matter
which
approach
you
take
right,
yeah.
You
might
proactively
create
a
neighbor
cache
entry
that
you
never
need
for
anything
right
and
you
may
flush
your
cache
and
and
so
on.
G
B
B
A
B
So
we
have
a
part
which
says:
ok,
here
is
a
problem
right.
So
then
there
is
a
part
which
says
there
are
possible
approaches
to
solution
and
I'm,
not
sure
where
this
belongs,
because
and
then
there
is
a
decision,
okay
out
of
all
those
things
we
decided
to
do
this.
This
is
this
which
belongs
to
chicks
man.
So
if
you,
since
there's
a
whole
possible
list
of
things,
we
might
want
to
do
belongs
to
chicks
man,
I
can
easily
a
move.
H
I
Hi
I'm
Bob,
hinden,
6-man,
co-chair
I,
mean
I
I'm,
not
sure
I
care
too
much,
where
the
different
possible
ways
of
doing
it
as
described
could
be
either
could
be
both
a
week.
If
there's
you
know
the
working
group
accepts
this,
then
we
want
to
have
a
document
that
describes
one
of
them.
They
care
more
about
that
part.
That's
helpful.
Okay,.
G
G
B
H
K
A
B
Okay,
so
I
guess
we
can
have
some
discussion
and
six
men
about
possible
solutions
and
yeah
I'm
still
looking
for
comments
on
this
one
as
humans.
That
solutions
list
might
move
to
six-man
area
right,
but
I
problem
problem
statement
isn't
the
draft
here
so
but
I,
it's
quite
short
one.
There
is
not
much
to
say
Sarah
if
I,
if
I
get
a
little
more
comment,
Seattle
to
post
0,
1
YES
on
looking
for
way
to
improve
the
document.
Well,.
A
L
M
N
N
N
There
again,
as
you
might
do,
if
you're
in
this
room,
you
run
an
nmap
against
the
global
public
IP
address
on
it,
and
you
see
quite
a
lot
of
stuff.
I
won't
go
through
it.
You
can
see
stuff
there,
there's
UDP
some
few
oddities,
which
I'll
talk
about
supportive
protocols
as
well.
Tcp
UDP
l
expected
there's
some
ones
that
I'm
not
that
happy
with
there
and
I
will
by
default
globally.
N
The
first
two
are
sort
of
necessary
they're
printing
protocols,
but
their
own
unauthenticated
open
to
the
internet.
So
people
can
print
to
my
printer
if
they
work
out
the
ipv6
address
I've,
no
idea
why
it's
got
a
to
FTP
server
running
by
default.
Snmp
is
there.
Yes,
it's
got
a
public
read/write,
a
read-only
community
readwrite
is
not
private,
but
it's
not
much
better.
I
have
no
real
idea.
Why
it's
got
this
Ventrilo
thing
on
it.
Apparently
it's
just
proprietary.
What
think,
maybe
there's
some
sort
of
printing
capability
in
it.
N
It's
also
got
mdns
listening,
which
is
somewhat
understandable
because
it
will
just
do
a
wild
card
socket
bind.
But
that
being
said,
it
could
possibly
be
exploited
or
exhausted
by
unicast
I
don't
need
a
global
ipv6
address
in
my
printer
or
my
prinnies
coming
from
the
link
local
lawyer.
Always
so
the
trouble
are
I've
got.
N
Is
that
I've
gotta,
just
a
single
SSID,
fairly
simple
setup,
probably
coming
from
most
consumers,
and
so
therefore
my
printer
configures
one
I
could,
of
course,
as
a
technical
person,
create
a
new
SSID
and
make
it
Yule
only
but
of
course
inconvenient
multicast
DNS
doesn't
work,
I
do
have
an
upstream
network
firewall
and
that's
probable,
but
devices
can
only
assure
what
they
can
control
and
if
I
pretend
to
be
a
non-technical
user,
which
I
think's
a
useful
thing
to
do.
Imagine
I,
don't
know
how
any
of
this
magic
works.
N
There
would
be
an
override
for
a
competent
system
in
Australia,
so
if
a
printer
like
mine,
was
deployed
in
an
office
or
something
like
that
or
so
on,
some
people-
weren't
I've,
proposed
this
I
said
yesterday.
You
know
you
should
use
mad
or
manufacturer
usage
description.
The
goal
seems
somewhat
similar,
although
the
network
is
sorted
to
enforce
the
security
policy
rather
than
the
device
itself,
so
talk
about
low
powered
IOT
devices
and
the
network
being
configured
with
access
control
lists,
and
so
on.
It's
much
more
discrete
as
well,
which
is
fair.
N
It
requires
a
supporting
mud,
server
infrastructure
and
it
assumes
a
local
network
administrator
to
manage
this
and
monitor
et
cetera.
Well,
the
typical
non-technical
home
user
is
not
going
to
be
a
capable
of
not
doing
that.
Effectively
so
in
comparison,
this
is
a
very
simple
mechanism.
It's
binary
it's
either
on
the
internet
or
not,
and
if
it's
not
on
the
internet,
whatever
services
it
running,
it's
running
out,
reachable
from
it
doesn't
require
the
technical
administrator,
and
it
could
also
be
seen
as
a
complement
to
mud
and
firewall.
N
Another
example
I've
had
some
experience
with
electricity,
smart
meters
now
they
must
really
be
using
you
allies,
because
you
don't
want
your
electricity
meter
infrastructure
sitting
on
the
internet
at
all
because
of
DDoS
and
electricity
networks
sabotage.
So
this
mechanism
could
be
seen
to
be
enforcing
that
requirement
of
safety
for
these
sorts
of
devices.
N
Another
comment
somebody
had
was
about
the
idea
of
using
having
these
devices
through
their
own
firmware
upgrade.
Well,
if
that's
the
model
of
software
maintenance
that
the
manufacturers
choosing
well,
then,
yes,
it's
clear
and
obvious
sister
vise
does
need
internet
access.
However,
an
alternative
idea
to
that
is
that
maybe
for
something
that
is
it's
a
it's
limited
just
to
firmware
upgrades,
then
the
device
actually
does
on
demand
global
addressing.
So
when
the
device
does
its
periodic
probe
to
see.
N
If
there's
a
new
version
of
firmware,
it
goes
and
acquires
rul
a
global
address
does
a
check.
There's
no
software
update
available
it.
It
deprecates
it
and
just
continues
to
be
available
via
relay
or
local
and
all
the
same
thing.
If
there's
an
upgrade
so
sort
of
that
middle
ground.
Any
thoughts
or
questions.
B
B
B
N
B
Like
again,
I,
don't
think
we
need
to
encode
into
the
device
and,
in
you
elation,
mind
kicks
what
what
kind
of
how
to
say
it.
Yeah
so
like.
If
device
has
a
globe
has
v6
connectivity,
it
should
have
be
six
connectivity
and
so
otherwise
yeah.
We
really
would
need
to
have
a
sticker
on
the
printer
box
right.
You
can
connect
two
printers,
this
internet
or
that
printer
could
not
be
connected
to
internet
right
because
I,
don't
think.
B
N
P
N
P
P
N
You've
written
a
draft
on
this
and
I
missed
the
meet
I've.
Had
this
idea
of
myself
as
well
the
questioner,
a
software,
you
know,
developers
gonna
do
that
for
each
and
every
service,
listening
on
my
printer
or
so
on
and
so
forth,
it
sort
of
this
is
sort
of
where,
as
much
as
discrete
application
level,
security
stuff
appeals
to
my
into
end
philosophy,
I
sort
of
think
there's
something
much
more
fundamental
at
sort
of
the
the
the
IP
layer.
It's
a
it's
a
blunt
object.
N
P
Is
it
like
to
be
on
the
internet
or
Dena
fee
has
to
be
on
the
Internet?
What
I
mean
is
that,
for
example,
I
I
could
ambition
that
an
application
could,
for
example,
somehow
signal
to
the
operating
system.
Okay,
this
service
is
supposed
to
be
local.
This
service
is
supposed
to
be
global
and,
based
on
that,
you'd
bind
that
port
to,
for
example,
some
subset
of
the
addresses
that
you
have.
Unfortunately,
we
don't
have
such
an
API.
N
Clear
and
obvious
to
the
vendor
that
it's
gonna
be
on
the
internet
or
not.
You
know
if
it's
ambiguous,
this
wouldn't
apply,
but
if
there's
certainly
you
know
device
cases,
I
look
at
light
bulbs,
for
example
the
the
ipv6
enabled
light
bulbs.
You
might
have
a
central
controller
in
your
house
that
might
be
on
the
internet
that
might
download
the
firmware
that
provides
your
remote
access
to
your
lights,
light
bulbs,
but
the
light
bulb
itself
might
not
possibly
shouldn't
be
on
the
internet,
so
it
would.
N
P
But,
for
example,
if
you
have
a
device
like
you
were
mentioning
that
has
automatic
updates,
for
example,
yeah,
at
least
during
that
window
of
time,
when
the
device
is
checking
with
some
server
over
there.
Whether
there
is
an
update
or
not,
there
will
be
a
window
where
the
device
can
be
attacked.
Okay,
so
that's
that's
an
indication
that
also
the
application
should
somehow
you
know
signal
to
the
operating
system,
whether
it's
support
the
the
application
is
supposed
to
be.
You
know.
P
R
N
R
N
It's
not
perfect,
I'm,
not
selling
it
as
perfect,
but
it's
just
you
know
my
printers
already
like
I
had
to
do
a
firmware
upgrade
on
it
because
there
was
a
remote
exploit.
The
the
password
on
telnet
was
default
to
admin
or
something
like
this.
This
is
2019
the
the
approach
of
trusting
people
to
build
internet
proof.
Software
is
not
working
I'm,
not
saying,
there's
a
solution,
but
it's
an
incremental
defense-in-depth
and
that
that's.
G
It
sorry
I
was
gonna,
say
that
too,
but
I
figured
that's,
probably
what
you
meant.
That's
what
has
a
danger
of
operational
issues,
because
the
case
that
I'm
worried
about
is
where
the
printer
only
has
is
following
this
rule
and
the
other
device
that
one
that's
trying
to
print
to
it
is
on
link
and
only
has
a
GUI
yeah,
but
it's
on
link,
and
so
you
can,
where
you
could
open
a
discovery
for
it.
You
know
it
was
on
link.
Okay,
and
this
could
happen.
G
If
say
one
of
those
a
slack
only
and
then
the
one
is
dhcpv6
only
or
there
was
some
time
out
when
you
were
trying
to
do
the
request
or
whatever
so
there's
operational
issues,
and
so,
if
you
were
to
extend
it
to
say,
or
you
have
reason
to
believe
that
it's
on
link,
because
your
arrays
or
whatever
on
says
it's
not
link
prefix
you'd
have
less
operational
problems.
Yeah.
N
S
Yeah
yeah,
okay,
Eric.
Thank
you,
I
see
a
major
flow
somewhere
in
your
approach.
You
assume
that
ula
or
even
link-local
are
safe
can
be
trusted,
and
it's
not
the
case.
There
is
a
printer
on
this
audience
s
ID
here
by
the
IETF,
which
can
be
acted
by
1000
people.
So
it
is
the
printer
safer
because
it's
using
you
la
vie.
S
Typically,
you
will
get
a
firewall
so
but
whatever
is,
and
then
the
other
point
is
well
you
la
you
see,
maybe
a
slash
64
or
the
slash
48
of
your
ula
I
would
actually
just
all
seven
it
and
then
the
last
point
right.
Maybe
you
know
there
is
an
attack
ipv4
to
be
sure,
but
typically
where
you
were
downloading
a
malware
on
your
PC
and
it
was
scanning
inside
so
basically
to
attack
behind
the
net.
So
the
same
thing
could
apply
your
new
el
ABC.
This.
N
It
it's
trying
to
be
incrementally
better
than
what's
out
of
the
box.
You
know
printer
like
this.
For
example,
you
can,
we
can
lobby
to
have
you
know.
Fifteen
different
software
developers
develop
internet
proof
software.
If
we
could
get
that
to
happen,
we
wonderful,
we
wouldn't
have
any
of
these
issues,
but
they
just
don't
seem
to
be
going
well
and
just.
S
P
Peace
I
think
that
trying
to
make
you
lace
up
a
level
in
every
network
that
can
have
this
kind
of
devices
is
a
big
operational
complexity
and
in
general
I,
don't
recommend
it
to
my
customers.
I
I
ask
them
to
go
away
as
much
as
possible
of
you
allies,
because
the
reality
from
the
operational
perspective
is
that
it
adds
a
lot
of
complexity.
T
Hi
Andrew
mark
thanks
very
much
I
must
say:
I
was
listening
to
this
I'm
kind
of
on
the
fence.
Here,
on
the
one
hand,
I
see
what
you're
trying
to
achieve.
On
the
other
hand,
I
look
at
this
and
I
go.
We've
lived
for
the
last
25
years
with
the
lie.
That
net
is
a
security
mechanism
and
there
is
security
through
office
keishon,
and
this
to
me
would
seem
to
propagate
that
lie
which
worries
me.
T
It
also
concerns
me
that
if
there
is
an
attacker
and
you
implement
something
like
this,
although
the
lies
going
to
do
is
look
for
something
with
a
global
address
on
the
same
network
and
he's
going
to
come
in
that
way
and
I'm,
not
sure
that
you
really
gain
anything.
Considering
the
operational
complexity
that
you
introduce
I
do
think
that
we
could
look
at
a
better
security
posture
on
the
CPS
and
the
edge
rather
than
going
through.
T
N
So
just
the
thought,
if
you
got
your
Apple
iPhone
or
your
Android
phone
plugged
into
the
network
here,
you're
trusting,
Apple
and
Google
to
be
doing
that
already,
because
I
haven't
found
a
firewall
interface
in
my
phone
yet
and
I'm
sure.
No
one
else
here
has
either
so
this
is
actually
sort
of
happening
already.
With
the
controversies
about
centralized,
we
got
centralized
device
security
for
a
lot
of
our
devices
as
it
is
look
I'm
very
much.
N
N
U
David
summer
terrace
erupting
it's
kind
of
echoing
some
of
the
other.
What
other
people
said
but
like,
like
you
said
you
know
putting
on
your
non-technical
users
head
on
and
trying
to
think
from
their
perspective
is
interesting,
different
examples
you
gave
like
printer
in
the
home
or
smart
meters
and
I.
U
As
a
non-technical
person,
I
want
to
enable
this
feature
and
if
I
get
delivered
device
which
vendor
says
well,
most
of
our
user
base
is
in
point
to
the
internet
stuff.
So
we'll
do
you
have
a
default?
I
might
go
to
my
router
and
then
like
open
up
the
firewall
so
that
it
can
talk
to
the
Internet
and
then
expect
it
to
work,
but
then
find
out.
I
have
to
go,
read
a
book
on
how
ula
works
in
order
to
get
this
to
work.
U
N
J
P
A
P
P
We're
under
santino
network
alessandro,
which
is
one
of
the
co-authors
of
the
document,
was
saying:
hey
I
succeeded
to
implement
ipv6
I
want
to
use
for
six
Forex
lat,
but
I
have
the
problem
that
that
will
mean
most
of
the
traffic
going
to
the
to
the
Syrians
will
become
now
ipv4
instead
of
staying
in
ipv6
because
which
for
six
Forex
lat,
it
could
be
converted
to
two
ipv4
right.
So
why
that
happens?
If
I
can
move
this
correctly.
A
P
What
happens
is
that
if
we
I
guess
you
will
need
to
go
move
this
screen
every
slide
when
you
have
an
ipv6
capable
device,
the
traffic
from
the
LAN
network
in
the
user
will
go
this
way
so
clearly,
because
the
CDN
is
dual
stack,
we
will
get
dual
stack
and
sorry
ipv6
and
2n.
Okay.
So
that's
that's
the
perfect
case.
That's
the
ideal
situation
now
what
happens
if
we
have
an
ipv4
only
device,
and
a
very
clear
example-
is
Smart
TVs
or
set-top
boxes.
P
Is
that
even
if
the
CDN
is
dual
stack,
god,
the
set-top
boxes?
Looking
is
for
an
ipv4
address
and,
of
course
the
dns64
and
the
next
64
will
give
him
and
I
pin
for
dressed.
So
we
will
be
using
all
the
way
through
for
that
flow,
then
at
64,
even
if
we
have
already
converted
that
to
a
TV
sixth
as
the
green
part
of
the
of
the
arrow,
but
we
will
convert
it
again
to
ipv4.
Okay,
it
don't
make
any
sense.
P
P
P
This
is
the
situation
I
I
already
mentioned
for
set
of
boxes,
Smart
TVs,
which
very
few
of
them
will
go
like
this.
Like
all
ipv6
n2n,
there
will
be
okay,
there
will
be
others
which
are
not
using
ipv6,
and
there
is
still
a
lot
of
Smart
TVs
I
check
it
with
several
Smart
TV
vendors
and
they
are
not
going
to
date
them.
So
unless
you
buy
a
new
Smart
even
two
years
from
now,
they
all
wants.
Most
of
the
time
will
not
get
updated
to
support
ipv6.
P
They
will
go
this
this
way.
Okay,
so
they
will
do
a
translation,
a
stateless
translation
in
the
salaat
ipv4
to
ipv6
and
then
again,
a
stateful
translation
in
the
indian
at
64,
from
ipv6
to
AP
before
again
no
sense
because
we
have
a
dualistic
CDN.
So
what
we
want
to
do
is
to
conserve
somehow
that
translation
that
has
been
done
already
by
the
CP
and
and
keep
going
as
ipv6
until
the
Syria
work
it
out
three
possible
approaches
for
this.
P
One
of
them
is
a
DNS
routing
based
solution,
which
means
basically
that
the
CDN
should
provide
addresses
from
the
well-known
prefix
or
the
network
specific
traffic's,
depending
on
which
one
is
using
their
ISP.
So
that
will
just
work.
Okay,
that's
one
approach,
but
it
has
some
some
issues
that
we
have
described
it
in
the
document
and
it
seems
it's
not
the
best
way
to
do
that.
In
fact,
when
we
have
been
talking
with
different
CDN
providers,
some
of
them
maybe
will
do
that,
but
some
others
said
no.
We
are
not
going
to
do
that.
P
Okay,
it
adds
a
lot
of
complexity
to
our
operations
and-
and
we
don't
want
to
do
that
way.
So
we
have
a
second
approach
which
is
using
explicit,
address
mapping
tables.
Okay,
this
is
already
define
it.
I,
don't
remember
the
receipt
number,
but
it's
something
that
is
being
used
already
for
some
time
ago
and
the
idea
is
that
not
64
the
still
happened
and
under
C
they
are
already
doing
all
the
DNS
thing
so
because
the
C
is
also
the
proxy,
we
can
have
an
internal
mapping
automatically
created
okay.
P
So
the
idea
is
that
by
looking
into
the
traffic
flow,
the
CP
can
guess
is
not
for
sure,
but
it
can
guess
that
a
device
is
an
IP
before
only
device,
and
if
it
can
guess
that,
then
he
can
create
an
entry
to
make
this
magic.
Okay,
if
the
guess
is
not
done
correctly,
we
are
not
breaking
anything.
The
traffic
will
be
converted
twice,
but
we
are
not
breaking
anything.
So
that's
that's
the
interesting
point
here.
So
this
is
an
example.
So
we
have
a
website
that
has
an
a
tracker
and
a
quad.
P
P
We
have
went
already
into
several
versions
of
this
document.
I
think
this
is
version
number
four,
because
we
discovered
several
things
that
may
happen.
For
example,
one
of
the
things
that
was
discovered
is
maybe
there
is
not
a
single
ipv6
address,
matching
a
single
ipv4,
one,
so
Indo.
In
that
case
we
are
just
discarding
the
translation.
We
are
not
trying
to
optimize
the
traffic
in
order
to
avoid
that.
Okay.
P
Well,
this
is
a
little
bit
of
of
the
explanation
of
all
the
details
that
we
have
work
it
out,
and
this
is
the
change
that
we
have
done
in
the
last
review.
So
this
is
the
important
view.
You
already
have
read
this
draft.
Before
this
version
there
was
a
very
extensive
email
from
Eric
nearing
and
we
have
some
some
private
email
exchange.
He
was
saying
basically
that
if
we
do
this
way,
this
is
not
sufficient
in
order
to
ensure
that
individual
applications
are
also
able
to
keep
the
existing
connections.
P
So
we
may
be
breaking
the
operations
of
the
CDN
okay,
so
he
was
suggesting
basically
what
with
it
that,
instead
of
using
the
explicit
address
mapping
for
just
creating
the
DNS
connection
from
the
a
record
to
the
quad,
a
record,
we
just
make
possible
to
create
a
stateful
NAT
entry
among
those
two
addresses
okay.
So
that
means
that,
even
if
the
DNS
change,
depending
on
TTLs
or
whatever
happens,
the
existing
connections
will
still
survive.
He
was
mentioning
the
example
of
that.
P
If
you
keep
changing
that,
obviously
the
video
will
meet
to
restart
recover
and
so
on,
and
you
will
be
breaking
it
old
all
the
way
so
that
that's
basically
the
thing
that
we
change
it
I,
don't
think.
Ok,
we
have
one
additional
approach
which
is
instead
of
the
CPE.
Creating
these
explicit
address,
mapping
tables.
P
We
can
do
something
and
we
have
not
done
too
much
work
on
that,
because
we
don't
believe
this
is
the
best
approach
we
can
do
something
so
the
ISP
sent
to
the
CPS
or
the
C
piece
query
from
the
ISP,
the
explicit
mappings.
The
problem
with
that
is
how
we
make
sure
that
those
mappings
are
getting
updated
fast
enough
to
accommodate
to
CDN
changes.
P
So
that's
the
reason
we
don't
believe
this
is
a
possible
available
approach
and
it
has
also
the
complexity
on
looking
for
which
protocol
we
use
for
doing
those
signaling
between
the
CP
with
the
isp
and
so
on
and
I.
Think
that's
it
well.
There
is
one
additional
detail:
I
almost
forgot
today
you
can
move
it
a
little
bit
high
fret
you
can
move
it
a
little
bit
high.
P
Now,
if
we
go
for
this
optimization
and
we
create,
let
me
go
up
if
we
create
a
fake,
a
record.
No,
no,
the
previous
is
light,
but
the
bottom
part
yeah
that
one.
If
we
create
a
fake,
a
record
in
an
ipv6
only
service,
we
are
not
breaking
anything
because,
anyway,
a
IP
before
devices
are
not
able
to
use
that
service.
But
if
the
CP
has
this
optimization,
it
will
work.
Ok,
so
we
believe
we
don't
break
anything,
but
instead
we
are
providing
a
solution
for
those
ipv4
only
devices
and
that's
it.
T
P
P
V
Hi
third
notch
Akamai
I'm
real
quick
about
this-
is
I'm
wondering
if
we
should
also
be
consulting
with
the
folks
who
are
looking
at
shifting
DNS
away
from
the
the
local
ISPs
and
consulting
with
them
about
the
activities
around
that
that
a
number
of
the
application
and/or
browser
vendors
are
currently
investigating,
because
that
has
a
potential
to
significantly
impact
these
technologies.
We.
O
O
O
P
A
Jordi,
let
me
take
you
back
to
Jarrod's
comment,
bypassing
the
ISP.
You
basically
argued
that
the
optimisation
wouldn't
break
anything
in
the
same
case,
but
in
an
enterprise.
Environment.
Enterprise
often
have
very
specific
rules
about
DNS
and
are
actually
pretty
concerned
about,
do
H
and
do
T
because
they
bypass
the
corporate
in
information
security
policy,
and
so
there
has
been
some
work
done
in
chrome
and
in
in
Firefox.
In
order
to
not.
J
P
I
can
check
I
can
check
again,
but
my
observation
is
that
if
you
are
not
using
the
DNS
from
the
ISP,
the
optimization
will
not
work
ISP
in
this
case
being
the
enterprise
the
enterprise.
Exactly
so,
that's
that's
why
I
say
if
the
CPE
is
not
able
to
do
optimization,
because
there
is
not
a
possibility
to
use
the
DNS
from
the
from
the
ISP.
You
are
not
breaking
it.
You
are
not
changing
the
existing
situation
right.
You
don't
apply
the
optimization.
That
might
be
a
question
of
viewpoint.
P
A
P
Let's
see
much
better,
okay,
so
I
present
it
also.
This
document
in
in
Montreal
and
I
updated
it
because
I
got
comments
from
Eric,
I,
think
and
Marc,
and
someone
else
I,
don't
remember.
I.
Think
I
got
comments
for
him
from
at
least
two
or
three
people.
P
This
document
I
started
initially
in
2006
at
that
time,
not
not
too
many
people
was
paying
attention
to
that.
But
I
realize
it
after
a
few
years
that
a
lot
of
people
is
using
what
I
was
trying
to
say
in
the
document
from
2006
and
in
fact
there
is
explicit
dhcpv6
prefix
delegation
option
to
exclude
that
prefix.
So
that
goes
a
small
part
of
my
original
document
from
2006,
which
is
okay,
let's
get
to
customers
last
48
and
let's
use
the
first
as
last
64
for
the
for
the
point-to-point
link
for
the
Wan
link.
Okay.
P
So
there
is
already
a
document
that
that
is
using
that,
but
is
not
describe
it
in
any
other
place.
So
that
document
is
just
saying:
okay,
you
can
exclude
the
prefix
for
what
it's
not
explained
for
what
okay,
so
what
I
am
trying
to
do
in
this
document
is
to
explain
that
and
take
a
novel
review
about
how
we
can
do
point-to-point
links.
P
So
summary,
we
start
the
document
saying
that
yes,
RFC's
seven,
six
one,
six
four
describes
that
one
of
the
possibilities
is
one
two
seven,
and
that
document,
which
is
being
misread
by
a
lot
of
people
is
not
saying.
That's
the
only
way.
That's
the
mandatory
way.
That
document
is
that
must
be
supported
by
routers
okay.
So
there
is
not
a
very
clear
idea,
because
the
people
read
that
document
like
hey
I,
need
to
use
one
two:
seven,
not
that's
not
what
the
document
is
saying.
So
I
clarify
that
and
again.
P
That
document,
of
course,
is
not
excluding
other
options
and
in
fact,
I
I
keep
doing
a
survey
among
ISPs
that
are
deploying
ipv6
and
about
62%
of
them
are
using
slash
64
instead
of
a
slash
1
to
7.
62
percent
is
a
lot.
It's
not
just
peanuts.
Ok,
a
new
section
in
the
document
that
somehow
was
suggested
by
Eric
and
also
mark
provided
some
some
inputs
on
that
is
reviewing
again
the
ping
pong
problem:
ok,
because
that
was
the
main
reason
for
the
previous
documents
to
go
into
into
that.
P
Then
we
describe
the
different,
perfect
choices
like
as
last
64
I
added
some
text
with
reference
to
mitigations
for
that
added
some
text
for
the
slash
127
to
ensure
that,
if
that's
being
you
said,
you
need
to
make
sure
that
both
endpoints
are
using
the
1
2
7.
Otherwise
you
still
have
the
same
ping
pong
problem,
I
added
a
reference
to
1
to
6
I.
Think
Fred
asked
it
for
that,
because
he
was
surprised
that
people
was
using
1
to
6
or
somebody
asking
for
that
and
additional
small
changes
in
that
section.
P
P
P
This
is
a
picture
about
some
of
the
data
that
I
before,
like
the
the
number
of
people
that
is
using
at
the
62%
of
people
that
is
using
already
is
last
64
instead
of
slash
1
to
7,
and
that's
it
basically.
So
so
not
not
measure
changes
in
the
document.
Just
updating
some
sections
clarifying
some
text
and
that's
it.
A
P
W
P
To
you,
I'm
Kryon
I
will
be
presenting
a
document
on
the
reaction
of
slack
to
flash,
remembering
events.
You
might
remember
that
at
the
prior
IDF
meeting,
both
in
six-man
and
basic
swabs,
we
had
been
discussing
this
topic
at
the
time
we
had
a
single
draft
that
essentially
had
the
program
statement,
discussed
operational
mitigations
and
also
discussed
like
possible
protocol
improvements
based
on
the
feedback
that
we
received.
Since
then,
what
we
did
is
split
the
original
document
into
three
different
documents,
so
this
is
the
first
piece
of
it
which
essentially
just
contains
the
problem
statement.
P
So
next
one
so
essentially
there
are
a
number
of
scenarios
in
which
you
might
face
a
flash
remembering
event,
which
essentially
means
that
all
out
of
the
sudden,
a
prefix
becomes
invalid
and
a
new
Ballad
prefix
is
introduced.
Of
course,
that
means
that
the
host
will
end
up
with
stale
network
configuration.
Information
and
ipv6
connectivity
will
normally
fail.
Obviously,
in
cases
where
happy
eyeballs
is
employed,
that
will
mean
that
ipv4
will
be
employed
instead
of
ipv6
next
slide.
P
So
in
most
of
the
discussion
that
we
had
on
the
on
the
mailing
list,
many
people
were
focusing
on
a
single
scenario:
you,
where,
essentially,
you
have
a
nice
P
that
is
providing
dynamic,
prefixes
to
users
and
SE,
p,
router
crashes
and
reboots,
and
that's
when
the
flash
remembering
event
is
phased,
but
actually
in
the
document.
We
were
covering
like
several
examples.
P
For
example,
there's
a
possible
scenario
where
is
switchboard
is
moved
to
a
different
villain
in
that
case
that
never
will
face
a
flash
remembering
event.
It
could
also
happen
that,
for
example,
the
net
worries
partition
so
when
the
RAS
are
sent
by
the
routers
to
essentially
invalidate
or
replicate
the
the
prefixes,
those
are
lost.
So
essentially,
when
the
productivity
is
resumed,
you
will
face
the
flash
remembering
event.
P
There
are
scenarios
where
you
might
be
manually,
configuring,
your
network
and
you
just
change
network
configuration.
You
just
kill
the
demon
and
start
it
again
and
and
restarted,
and
that
would
also
mean
a
flash
remembering
event
and
then
last,
but
not
but
not
least,
there's
the
scenario
that
health
has
been
mostly
discussed,
discussed
on
the
mailing
list,
which
is
the
case
where
CPU
router,
for
example,
crashes
and
reboots,
and
you
know,
since
the
ISP
is
doing
dynamic,
prefixes
the
CPE
gets
a
different
prefix
and
again
the
flash
remembering
event.
Of
course,.
P
Just
as
a
note
short
adidas
RV
on
on
ISPs,
and
he
included
a
question
on
whether
ISPs
were
doing
dynamic
or
stable
prefixes
and
37%
of
them
were
doing
dynamic,
prefixes,
okay
next
slide.
So
what
does
this
document
do?
Essentially,
as
I
mentioned
before,
we
originally
had
a
single
document
that
tried
to
do
everything
analyze.
The
problem
also
discuss
possible
operational
mediations
and
also
discuss
possible
protocol
updates
to
improve
the
reaction
of
slag
to
renumbering
events
and
based
on
the
feedback
that
we
received.
We
ended
up
splitting
the
document
into
three
pieces.
P
Now
this
specific
piece
that
we
are
discussing
here
essentially
analyzes.
The
this
problem
describes
some
operational
mitigations,
for
example,
in
the
specific
case
were
that
I
was
referring
to
regarding
the
CPU
router,
noting
that
if
you
were
doing
stable,
prefixes,
of
course,
the
Prophet,
the
problem
would
be
mitigated
and
also
that
you
might
operationally
tweak
the
lifestyles
that
you
use
for
the
pios
options
that
you
ever
tries
in
the
erase.
So
obviously
that
wouldnt
avoid
problem,
but
would
limit
the
amount
of
time
during
which
this
problem
persists.
P
Unless
this
document
essentially
points
to
the
other
two
documents,
noting
that
okay,
this
is
the
problem.
There
are
some
things
that
you
know
you
might
do
operationally,
but
probably
there
is
work
to
do
in
other
areas
next
slide,
so
you
know
so
far.
I
think
that
you
know
when
it
comes
to
the
problem
statement.
The
document
is
pretty
stable.
So
our
question,
as
authors,
is
whether
this
is
something
that
you
know
false
feel
that
we
should
be
working
on.
A
M
B
A
P
P
So
what
happens
in
many
scenarios?
Is
that
when,
for
example,
with
a
CPU,
Robert
crashes
and
reboots
and
the
seat
and
the
CPU
rudder,
does
the
ACP
prefix
delegation
again?
It
gets
a
different
prefix.
But
the
thing
is
that
if
the
CPU
router
doesn't
record
on
stable
storage,
the
prefixes
that
it
has
been
delegated
before
it
doesn't
have
the
knowledge
to
actually
these
try
to
disable
or
duplicate
the
prefix
on
the
local
network.
P
Another
thing
is
a
requirement
that
the
CPU
rotor
essentially
implements
the
the
proper
prefix
delegation
and
slack
interface,
essentially
meaning
that
the
life
times
that
are
used
by
slack
never
span
past
the
DHCP
prefix
lifetime.
That's
a
requirement
in
the
DHCP
version,
six
specification
already.
P
The
final
one
that
we
have
in
this
light
is
that
it
has
been
reported
to
us
I
think
it
was
Ted
lemon
that
reported
this,
that
some
boxes,
when
you,
when
you
reboot
them,
they
send
the
HTTP
version.
6
release,
meaning
that
even
if
open
a
reboot
event,
the
CPU
router
would
get
the
same
prefix
since
the
city
router
did
their
release,
then
in
that
case
it
will
probably
get
a
different
perfect.
This
time,
if
I
remember
correctly,
Ted
lemon
had
posted.
P
P
A
P
Fernando
I
I
think
you
are
too
optimistic
quiz.
The
recommendation
that
CP
bender
should
save
the
information
in
stable
storage.
I
wish
it
happens,
but
it
will
not.
It
will
not
because
doing
that
increase
a
lot.
The
price
of
the
of
the
CP
and
this
kind
of
cities
typically
are
meant
to
be
very,
very
cheap.
There
are
many
kinds
of
flash
memories,
and
some
of
them
are
meant
to
basically
be
able
to
reflash
it
a
few
times
and
after
that
they
will
fail.
P
They
will
not
work
anymore,
so
asking
the
Bender's
to
to
be
able
to
save
a
prefix
once
per
week
or
once
per
month,
I.
Don't
really
think
it
will
succeed,
so
I
rather
think
we
should
go
for
a
different
solution
and
in
addition
to
that,
the
problem
is
always
that
even
if
we
finally
decide
to
go
in
that
direction,
I
am
NOT
optimistic
that
the
people
will
update
the
CPS.
P
In
fact,
one
of
the
problems
that
we
are
having
with
AP
v6
deployment
is
that
most
of
the
vendors
don't
provide
new
versions
of
of
filler
for
the
CPS
and
that's
the
same
for
the
transition
mechanism.
So
I
rather
think
that
we
should
look
much
better
for
a
solution
from
the
DHCP
server
or
something
like
that.
P
I
would
say
that
I
probably
agree
a
lot
with
what
you're
saying
but
well,
we
were
you
know
discussing
the
topic
and
possible
solutions.
There
were
Forks
arguing,
you
know
it
would
be
nice
if
the
CPS
we're
doing
this,
and
so
what
we
came
up
with
is
a
set
of
let's
say,
recommendations
such
such
that
if
you
are
a
vendor,
a
CPU,
router
vendor,
and
you
want
to
behave
more
nicely
to
these
events.
T
Fernando,
firstly,
thank
you
very
much
for
me.
I
like
what's
in
here
and
I
would
definitely
support
it.
Proceeding
I
do
agree
with
a
lot
of
what
Rudy
said
about.
Well,
people
may
ignore
this
etc.
I,
don't
think,
that's
a
reason
not
to
do
it,
particularly
because
once
you
do
this
and
you've
got
that
recommendation
in
standards.
It
creates
a
situation
where
guys
like
myself
when
we
go
out
of
RFPs
in
the
market
can
say
you
must
comply
to
this,
and
we
you
don't,
have
it
in
a
document.
X
Winner
is
un
hol,
so
you
know
I
had
a
question
about
in
your
draft.
You
say
you
know
when
we've
done
testing
for
flash
free
numbering
on
CPS.
What
we
see
for
the
slack
side
is
often
they'll
deprecated
the
address
they
used
to
have,
but
by
always
putting
it
in
a
Pio
once
right,
and
so
then
the
hosts,
if
they
miss
that
packet
on
the
lossy
network,
would
never
get
it
I
see
in
your.
As
you
say,
you
know
any
prefix
that
was
previously
transmitted.
It
should
continue
to
send
with
zero.
X
P
And
that's
also
probably
an
indication
that
that
you
know
folks
might
also
be
willing
to
use
more.
You
know
a
prepay
lifetimes,
because
if
we,
if
they
keep
using
the
lifetimes
that
we
currently
do
like
you
know,
prefer
for
a
week
valid
for
a
month.
That
means
that
in
theory,
you
should
be
trying
to
deprecate
those
prefixes,
for
you
know
that's
much
time,
which
obviously
doesn't
make
any
sense
at
all.
B
Java
so
I
think
it
does
make
sense
to
specify
recommendations
for
outer
vendors,
because
obviously
some
of
them
will
ignore
recommendations,
but
still
some
implemented
might
look
at
them.
Otherwise
it
doesn't
make
sense
to
publish
any
router
requirements
at
all
and
also
it
helps
to,
as
Anna
mentioned,
I
defined
requirements
and
tasting
criteria.
So
we
basically
say
what
we
shall
expect
from
CPS
and
that
requirement
should
be
there.
Yeah.
P
Exactly
just
just
to
clarify
my
point:
I
am
NOT,
saying
that
we
should
not
do
that.
I
am
saying
that,
in
addition
to
that,
we
should
also
approach
from
the
other
side,
because
I
think
part
of
the
problem
is
the
concerns
that
we
have
that
we
should
use
non
persistent
prefixes,
that's
creating
a
lot
of
problems
to
us.
I
think
we
need
to
show
there
is
peace,
that
the
model
that
we
are
using
for
ipv4
is
broken
and
that's
going
to
create
probably
other
problems
that
we
didn't
see
right
now.
P
A
To
that
point,
inserting
myself
into
queue
seems
like
the
simplest
approach
is
not
to
use
stable
storage,
which
dies,
provably
dies,
but
to
put
something
in
an
RA
that
says:
I
have
a
new
prefix,
whatever
I
may
have
told
you
before.
Please
consider
invalid
and
I
thought
about
that
for
at
least
a
minute.
You
know
I,
don't
make
any
great
claims,
but
that
that
might
be
an
approach
to
consider
in
the
case.
A
Okay,
so
thank
you
for
Nando,
where
I'm
going
to
go
a
last
couple
of
years.
We've
found
a
youth
well
ron
and
I
have
found
it
useful
to
pull
the
working
group
after
a
meeting
and
say
what
do
you
think
about
some
draft
that
was
on
the
on
the
agenda,
and
people
have
often
made
some
pretty
good
comments
in
email
when
when
they
can
get
back
to
the
ranch
and
think
about
things,
I
detect
a
comment
about
to
come.
I
just.
P
You
can
prefer
that
you
cannot
invalidate
because
there's
text
in
are
seeing
3861
that
says
that
if
you
receive
a
valid
I'm
smaller
than
two
hours,
you
should
ignore
it.
R
Also
Fred
you,
this
is
Bernie
volts.
Also,
you
wouldn't
want
to
use
just
a
single
bit
because
it's
back
to
the
old
problem
of
how
many
times
you
send
that
are
a
right.
So
it's
probably
a
router
routing,
advertisement,
sequence
number
or
something
that
you
want.
So
if
that
changes,
these
you've
got
a
different
configuration,
and
you
know
the
old
stuff
should
be
thrown
away.
Okay,.
N
You
can't
you,
you
can't
advertise
a
valid
lifetime
lower
than
two
hours,
so
there's
a
any
do
s
that
denial
service
protection
in
there,
but
the
other
thing
I've
realized
is
that
I
think
the
most
effective
solutions
are
the
ones
that
are
applied
where
the
problem
exists
and
the
further
with
the
way
I
see
this
partly,
is
this
factor
of
trouble
getting
CPE
vendors
to
follow
it.
Updating
all
the
hosts
is
perhaps
even
a
bigger
problem.
N
I've
also
learned
that
sometimes
education
is
cheaper
than
just
more
technology
and
so
I
wonder
a
lot
of
these
I
Spees
I
suspect
they're,
doing
it
just
because
they're,
following
on
from
the
way
they've
done
ipv4
a
broadband
which
is
the
way
they've
done
dial-up
so
I'm,
not
fundamental,
I'm,
sort
of
not
against
making
things
more
robust
in
CP
or
hosts,
but
I'm
just
a
bit
skeptical
of
its
effectiveness,
because
of
partly
what
I've
seen
with
CPE
I've
got
an
old
presentation
from
2011
smog
about
how
good
or
bad
ipv6
CP
is
I
recently
heard
that
there's
still
memory
leaks
in
CP
a
10
years
later.
N
So
there's,
if
you
want
to
play
with
some
of
these
options,
play
with
red
beady
because
they're
already
there
it's
deprecated,
prefix
and
decrement
lifetimes
to
address
some
of
these
things.
But
I've,
certainly,
you
know,
be
advocating
encouraging
eye
space
to
adopt
more
stable
models,
just
because
that's
sort
of
where
the
problem
exists.
Fundamentally,.
P
N
A
K
Hello,
Metro
City
from
a
safari
lt's
I
found
your
proposal
about
seven
feet
of
configuration
is
very
interesting
because
have
you
tried
to
consider
other
devices
except
of
for
a
CPS
in
this
case
other
types
of
devices?
It
could
save
this
configuration,
for
example,
the
IOT
rotor,
for
example,
out
you
wrote
it's
not
the
device
it,
it
could
be
rebooted
in
much
frequent
times
and
CPS
or,
for
example,
if
you
have
some
IOT
systems,
it
has.
K
P
K
C
Yeah
Marconi
yeah
I
think.
As
so
people
said
there,
there
are
more
long-term
solutions
thinkable,
but
if
their
solutions
include
like
yet
the
world
should
go
to
static,
addresses,
I'm
afraid
the
market
is
against
us.
People
stick
to
what
they
know.
There's
so
many
dynamic
address
you
won't
convince
all.
Is
fish
I
like
oh
yeah?
Let's
just
go
static.
C
I
do
see
certain
solutions
in
terms
also
on
the
DHCP
server
side,
but
as
a
quick
fix
and
I
think
I
support
Andrew
in
the
statement
is
I
think
we
should
move
this
forward,
at
least
as
an
intermediate
until
we
solve,
for
instance,
like
you,
said
DRI
protocol,
because
that's
gonna
take
a
long
time.
This
appears
to
be
a
short
term
for
what
is
a
real
problem
in
the
market.
So,
let's,
let's
move
it
forward
than
she
was.
We
can
do
in
this
place.
A
Okay,
thank
you.
So,
at
any
rate,
what
I
was
getting
at
was
what
what
is
going
to
be
our
work
plan.
Basically,
between
now
and
between
now
and
IETF
107
we've
got
some
interesting
events
that
will
consume
our
time
in
the
u.s.
we
have
Thanksgiving
next
week
in
most
parts
of
the
world
people
celebrate
Christmas
and
that
two
week,
or
maybe
two
weeks
in
in
the
other,
the
rest
of
the
world,
they
celebrate
Chinese
New
Year's
and
that
gets
interesting.
A
So
what
I
think
I
want
to
do
is
basically
use
the
first
three
weeks
in
December
in
the
first
three
weeks
in
January
and
send
a
note
referencing
one
of
these
documents
that
we've
got
and
asking
for
people's
comments
and
part
of
that
will
be
in
Fernando
I,
know
you're.
Waiting
for
this
question.
Do
you
want
to
adopt
this
as
a
working
group?
Draft
Fernando
would
really
like
to
see
these
his
two
documents
adopted.
A
So
that's
what
I
expect
to
do.
It
might
be
Ron
sending
the
email,
but
that
that's
what
I
expect
to
do.
As
far
as
collecting
comments
and
working
group
status
on
the
drafts,
you
can
be
looking
forward
to
that
and,
please
feel
free
to
say
something
when
we
send
those
notes.
So
with
that
do
we
have
any
other
business?
Are
there
things
that
we
should
be
talking
about
at
this
point.