►
From YouTube: IETF103-ANIMA-20181105-1350
Description
ANIMA meeting session at IETF103
2018/11/05 1350
https://datatracker.ietf.org/meeting/103/proceedings/
B
A
B
Tenon
actually,
so
this
is
money
and
we
have
him
for
tours.
This
is
animal
working
group
are
trying
to
networking
integrated
model
and
approach.
So
if
you're
having
in
wrong
room
still
have
time
to
run
for
your
right
room,
let's
say
not:
well,
everybody
should
already
read
it.
If
you
don't
please
to
so,
cruciate
is
ski
around
you
to
make
sure
you're
signed
before
you
leave
the
room
and
with
it
minutes
taker
and
invite
here
for
me
to
take
her
once
big
thanks,
I
get.
B
This
is
support
full
remote
participation
and
also
lights
already
uploaded
on
the
website,
with
murderers
from
idea
and
meninist
as
euros.
It
should
be
the
first
working
place,
so
everybody
should
have.
The
you
know
invokes
it
is
cussing
and
here
and
take
participant
for
many
knees
to
discussing.
Normally
we
gives
the
Ferrari
for
those
tropicals
already
have
discussed
many
mists
during
the
ITF
meeting
cycles.
B
B
First,
the
working
group
document.
We
have
three
working
group
document
need
update
a
trauma,
control
plan,
bootstrap
remote
security
infrastructure
and
constraint,
vulture
artifacts
for
put
starting
protocols.
Each
of
them
have
ten
minutes,
but
since
the
hour
they
are
working
group
document
they
have
priority
over
the
other
presentations,
so
they
could
may
take
more
time
if
that's
needed,
then
we
have
the
our
potential
new
charter
inscribing.
B
So
it's
discussing
what
we
are
going
to
do
for
next
stage.
Some
proposed
text
will
be
explained
by
tariffs,
and
then
we
have
this
non
working
group
work
items
first,
information
distribution
in
a
container
to
networking,
then
that's
the
TSST
comfortable
service
discovery
grasp
accurate
that
can
buy
with
the
number
line
out
configuration
of
nork
service
is
AP
networks
wide
grasp.
So
thanks
terrace,
to
make
me
confuse
he
applied
for
two
slots,
but
he
decided
to
combine
them
together
and
you
know
the
one
slice
so
I
guess
you
have
ten
minutes
for
that.
B
B
No
full
documents
missing,
wife
they're.
Actually,
they
already
passed
the
publishing
procedure
for
a
long
time
last
year,
that's
across
and
also
the
profits
of
the
management
and
the
they
will
be
published
out
of
there,
as
well
as
their
dependents.
A
document
get
published
and
and
the
autonomic
control
plan
is
already
submitted
for
publication
and
it's
under
the
ISD
investigation.
Is
they
in
the
ad,
follow
cellular
aging
and
as
we
discussed
to
resolve
and
the
reference
model
has
been
submitted
to
a
city
in
this
way,
it's
the
reverse.
B
Risky
actually
submitted
to
ISD
for
vacation
in
August
and
still
waiting
for
a
deal
at
have
some
work
for
you
and
we
have
to
draft
in
the
working
group
stage
grasp
API
atop
the
December
last
year.
It's
in
version
two
at
this
moment
and
the
constrain
water
atop
team
a
at
will
make
updates
in
the
sweetie.
B
D
B
B
E
B
G
B
A
Okay,
so
I
wanted
to
give
updates
in
Scythia
102
that
was
16
now
we're
1718,
so
the
incorporated
feedback
is
from
Alicia
Elron
for
general
Frank,
chilling
from
sector
and
ethics
from
Pascal
and
I'm
going
to
talk
about
the
backlog
at
the
end
of
the
presentation,
so
the
main
changes
from
16
to
18
that
are,
you
know,
interesting
enough
to
be
brought
up
in
front
here,
mostly
technical
elements.
There
was
a
lot
of
textual
improvement,
of
course,
that
I
don't
want
to
go
into
detail,
use
RFC
DIF,
please.
A
So
we
had
to
introduce
an
option
in
the
certificate
to
indicate
that
the
address
of
the
device
is
not
assigned
through
the
certificate
itself,
because
typically,
the
acp
certificate
will
carry
the
ACP
address.
But
then
you
have
notes,
for
example,
in
the
NOC
that
are
attached
through
normal
ACP
connect
subnet
and
they
would
get
their
local
part
from
you
know
local
procedures,
so
they
wouldn't
be
fixed
by
the
certificate
and
we
hadn't
foreseen
that.
A
So
that's
now
done
by
indicating
in
the
certificate
that
try
to
figure
out
your
you
know:
ACP
local
address
part
differently
and
then,
of
course,
in
the
certificate
check.
We
wanted
to
make
sure
that
we
say:
okay,
this.
This
is
supported
if,
during
the
enrollment
of
the
device,
it's
made
clear
that
it
is
not
taking
its
ACP
address
from
the
registration
process
but
locally,
then
that
is
explicitly
shown
on
the
certificate
and
therefore
you
know
security-wise
permitted.
A
We
don't
currently
have
an
AI
mechanism
that
relies
on
taking
the
ACP
address
of
appear
from
the
certificate,
so
this
is
fine
and
won't
change
anything.
The
only
use
that
we
currently
have
is
that
the
note
itself
learns
its
own
address
from
the
ACP
address.
So
if
it
sees
a
zero
there,
then
it
would
know
I
have
to
figure
out
my
ACP
address
differently
and
we
do
know
how
to
do
that
for
NOC
device
that
are
connected
by
ASAP
connect.
So
I
was
kind
of
the
main
technical
point
that
came
out
of
the
discusses.
A
Then
clearing
up
any
remaining
confusion
about
what
is
normative
in
the
document
and
what
is
informative
and
so
I
think
one
of
the
main
pain
point
was
the
requirement
section
which
goes
back
to
the
fact
that
in
the
current
Charter
we
were
not
allowed
to
have
separate
requirement
documents.
So
then
we're
said,
put
the
requirements
into
the
actual
solution
document.
So
we
had
this
informational
requirements
section
in
there
just
to
stage
within
the
working
with
what
we
actually
are
working
against
right.
A
What
is
it
that
we're
trying
to
achieve
with
the
ACP
in
a
high-level
fashion
and
that's
basically,
what
still
the
current
requirements
section
is,
so
that
is
informative,
like
typically
any
requirements
document
that
other
working
groups
are
producing
is
true
and
then
the
solutions
normative
and
yeah
the
requirements
section
was
using.
Words
must
and
should,
and
so
those
obviously
are
not
2119
normative
requirements
word
so
I
changed
them
to
underscore.
If
that's,
hopefully,
a
good
way
to
make
clear
that
they're,
just
you
know,
requirement
solution
requirements,
but
not
a
normative
must
insured.
A
So
yeah,
then
also
there
were
I,
think
several
reviewers.
Finally,
with
the
Alicia
saying
that
you
know,
we
should
make
sure
that
we
don't
have
too
much
futures
in
the
normative
section,
so
I
try
to
remove
all
that
is
unnecessary
and
then
I
created
appendix
a10
for
those
pieces
that
really
are
important
to
keep
documented
about
future
so
that
any
future
work
that
we're
doing
actually
knows
what
the
ACP
needs
in
every.
A
In
that
case
and
I
think
I'll
show
one
or
two
things
in
the
coming
slides,
so
really
try
to
straighten
up
the
normative
part
to
really
only
be
what
we
know
must
be
done.
That
included,
for
example,
on
the
technical
side,
primarily
this
our
sub
element.
That
also
comes
up
in
the
security
review
right
now.
So
that
is
really
an
important
thing
right
now,
and
not
only
for
the
future.
So
one
of
the
things
is
that
it
helps
a
lot
to
avoid
any
undesired
ula
hatches
when
we're
assigning
addresses.
H
A
A
Anything
formal
would
be
nice,
but
if
we
can't
find
something
good
formal,
then
let's
go
with
that,
so
yeah
so
intend.
We
have
the
intent
to
define
intent.
That
was
our
intent
given
to
us
by
NMR
G,
so
I
have
a
lot
of
issues
with
intent,
but
I
think
at
this
point
in
time.
Our
overall
strategy
and
you'll
see
that
also
in
what
we've
written
into
the
candidate
Charter
is
to
bug
NMR
G
with
that,
and
we
had
at
102
and
off-site
meeting
in
NMR
G,
where
I
brought
up
the
point
about.
A
A
So
if
you're
using
the
ACP
to
distribute
information
that
the
ACP
needs
to
be
billed,
you
may
be
shooting
yourself
in
the
feet,
creating
security
issues,
and
so
that's
definitely
something
that
we
want
to
be
very
careful
when
we're
defining
through
potential
future
work
like
the
information
distribution
is
a
scheme
that
does
that,
and
so
that's
what
that
Appendix
A
8
now
is
focused
on,
and
it
gives
one
example,
because
there
is
the
desire
to
allow
a
CPS
to
be
interconnected.
But
what
are
the
policies
for
interconnection?
A
They
could
be
distributed
through
the
network
itself,
so
some
part
of
the
information
distribution
could
really
impact
on
saying.
Okay,
this
information
distribution
is
always
done.
Then,
on
the
other
side
of
the
network,
you
are
seeing.
If
you
can
accept
that
information.
If
so,
then
you're
acting
on
the
information,
then
you
may
bill
PACP
through.
So
this
is
just
hinting
at
it,
but
primarily
it's
a
problem
statement
for
future
work,
around
information
distribution
that
leverages
the
ACP.
A
A
We
don't
want
to
give
them
any
opportunity
to
break
the
ACP,
and
there
is
pretty
much
only
one
key
point
left,
which
is
that
all
the
ACP
panels
are
specified
in
the
standard
to
be
on
top
of
normal
data.
A
plain
ipv6
link,
local
and
that
can
basically
and
does
in
you
know,
implementations
that
I've
been
involved
with
break
the
ACP.
A
If
the
data
pane
ipv6
link
local
is
misconfigured
and
the
fact
that
this
is
possible,
or
still
part
of
the
conscious
choice
of
trying
to
find
the
lowest
common
denominator
for
how
to
carry
ACP
channels
that
we
know
will
work
across
all
possible
platforms
and
router
types
right
and
that
basically,
any
improvements
that
can
be
done
in
the
future
would
rather
be
done
as
simple.
You
know,
add-on
documents
that
would
just
specify
additional
link
local
procedures
and,
as
I
was
outlining
in
this
section,
also
some
of
them
there.
Even
you
know
implementation.
A
F
F
F
We
can
sign
it
with
CMS,
just
like
the
regular
brewski
vouchers.
You
can
sign
it
with
Cosi
at
first
I.
Couldn't
understand
why
you
would
want
to
a
constraint
voucher
that
you
then
signed
with
a
very
big
CMS
object
and
turns
out
that
there's
some
good
reasons
to
do
this,
and
one
of
them
is
just
because
they
want
uni
P
rather
than
TCP
and.
F
F
But
is
that
is
there's
a
desire
from
a
lot
of
people,
then
say:
okay,
but
now
I
really
want
to
use
co-op
to
do
brewski
operations,
and
so
the
question
is:
does
that
text
belong
in
this
document
right
now?
The
text
lives
in
a
six
dish
document,
because
that's
the
working
group
that
initially
originally
four
years
ago
wanted
to
actually
do
this,
and
so
doc
text
has
been
leaving
that
document
over
the
years
and
going
into
other
documents.
F
So
I
really
want
some
feedback
from
this
and
I
just
little
think
box
we're
actually
having
a
site
meeting
on
on
enrollment
period
at
6:00
p.m.
tomorrow.
So
come
if
you
care
about
that
or
if
you
have
an
opinion
about
this.
So
next
slide
I
think
that's
question
yeah.
We
still
have
some
issues
with
SID
allocation.
F
These
are
political
layer,
layer,
8,
layer,
9
issues,
they're,
not
happy
they're,
not
what
are
the
SID
values
so
much
as
what
is
the
process
at
the
IETF
by
which
we're
going
to
manage
these
things,
and
we
have
a
couple
of
different
proposals
that
we
need
to
talk
to
the
core
working
group,
which
is
by
the
way
court
code.
Also
in
this
time,
slot
I
think
it
is,
and
so
we
got
the
early
allocation
for
the
content
format
numbers.
F
There
is
in
fact
a
fair
hair,
Alliance
Interop
plan
for
the
end
of
November,
which
will
need
to
use
these
numbers
and
I
guess:
I'll
use
private
allocations
if
we
have
to-
and
so
that's
pretty
pretty
good,
because
we
basically
have
four
potentially
five
companies.
Implementations
that
are
are
going
to
be
interoperating.
All
of
this
you
know,
within
weeks
and
I
also
have
some
things.
I
can
say
that
I
just
wrote
about
for
coop
es
T's,
there's
some
other
people
that
are
just
doing
est
for
that,
and
we've
been
trunk
weekly
meetings
next
slide.
F
Please
that's
it.
That's
last
slide
yeah
questions,
so
in
particular
the
question
of
whether
or
not
the
operational
Bruschi
part
belongs
this
constraint
voucher
document.
If
so,
then
the
document
becomes
potentially
much
larger.
If
not,
then
we
either
need
another
document
somewhere
else
or
some
other
decision
and
I'm
I'm
I
personally
would
rather
just
have
small
documents
that
sail
through
quickly,
although
that
never
seems
to
really
be
the
truth.
B
B
A
We
started
to
write
up
a
proposed
Charter
and
started
to
circulate
that
on
the
working
group
list,
and
so
with
the
more
active
discussion
like,
for
example,
here
right
now
to
progress
that,
and
hopefully
conclude
so
we
had
one
round
of
review
and
feedback
with
our
area
director
Ignace
who
people
who
don't
know
him
so
and
I
think
he
just
opened
the
you
know
the
official
thing
and
they'd
had
Becker
about.
We
chartering
I'm,
not
actually
sure
what
the
detailed
process
there
is,
but
I
think
unis
could
explain
that
to
us.
A
Also,
the
relevance
of
the
NOC
kind
of
knock
is
really
just
the
most
commonly
understood.
Example.
For
some,
you
know
decentralized
or
centralized
external
entity
that
executes
on
which
a
lot
of
control
is
being
exercised,
whether
that
is
you
know,
involving
a
lot
of
human
action
or
future
automation,
so
hopefully
were
capturing
that
correctly,
there
was
the
confusion
about
REM
what
OM
means
traditionally
for
20
years
in
the
network
management
area,
there's
one
understanding,
and
then
people
in
the
routing
working
group
started
to
introduce
terms
like
IOM.
A
There
are
all
these
wonderful
things
we
could
do
right
now,
but
sorry,
these
three
things
are
coming
from
the
same
authors.
Maybe
we
should
rather
serialize
them
so
really
doing
time
management
and
then
also
seeing
that
by
default,
we're
looking
into
things
that
could
be
finished
faster,
so
I
think
we
all
went
through
learning
exercise
with
chat
around
one
and
the
documents
that
we
started
out
doing
where,
ultimately,
you
know
very
big
grasp
brueski
and
acp
specifically.
A
A
Next
slide,
so
we
do
have
you
know
a
long
list
of
drops
that
have
been
presented
over
the
time.
On
the
right
hand,
side
you
can
see
and
I'm
not
going
to
go
through
them,
just
to
see
that
we
have
a
lot
of
interesting
potential
work
pending
and
that
isn't
even
including
you
know
all
the
things
that
I,
for
example,
would
like
to
work
on
and
just
finishing
up
around
one,
or
maybe
others
as
well.
That
haven't
been
posted,
and
so
that's
a
exactly.
A
If
somebody
else
can
do
it
too,
why
should
we
and
what
can
effectively
be
done
in
terms
of
where
we,
you
know,
as
I
said
from
the
time
management
the
resources
that
we
have?
Can
do
it
and
then
also
you
know
that
relate
to
the
Charter
around
one,
because
what
we
have
done
with
the
Ani
is
a
framework
that
obviously
we
want
to
leverage
in
the
future
work
and
we
have
undefined
components
like
the
a
si
the
autonomic
functions.
A
So
those
are
obviously
key
pieces
that
we
would
like
to
have
follow-up
work
done
because
they're
so
far
from
charter,
one
bard
next
slide.
Okay,
so
here's
the
next
slides,
really
the
you
know,
read
down
of
the
text
of
the
Charter
and
it
would
really
be
great
if
you
know
we
can
have
comments
on
these
any
you
know,
suggestions
for
improvements
or
so
so
that
we're
making
sure
that
we're
really
taking
the
feedback
from
the
working
group
into
account
and
I'll.
You
know
try
to
ask
that
at
the
end
of
each
slide.
A
So
let
me
go
through
this,
so
the
autonomic
networking
integrated
modeling
approach
working
group
is
developing
specifications
and
supporting
documentation
for
integral
products,
implementations,
operational
procedures
for
automated
network
management
and
control
mechanisms
for
networks
that
are
developed,
built
and
operated
by
professional
personal.
So
actually
that
should
have
been
read.
So
you
know
these
professional
networks,
that's
what
I've
replaced
there
with
you
know
it
developed,
built
and
operated
by
professional,
personal
and
I.
Think,
especially
when
we
get
to
the
ASAS
the
autonomic
functions.
A
I
think
what
we're
trying
to
achieve
is,
how
can
you
know,
network
software,
be
build
in
a
more
modular,
fast
and
easier
fashion
through
the
concept
of
you
know,
modular
ASAS
that
are
combined
to
autonomic
functions
as
opposed
to
what
we
have
right
now,
which
is
you
know
in
most
of
network
equipment
today,
humongous
tens
of
millions
of
lines
of
code,
monolithic
code,
which
totally
is
the
opposite
to
what
we
know
to
be
successful
in
software
and
data
centers
for
example.
So
that's
why
you
know
to
me
I.
A
Think,
specifically,
the
aspect
of
you
know:
what
can
we
do
for
the
building
of
a
better
network
automation?
Software
should
also
be
a
key
part
and
not
only
the
operational
part.
You
know
once
the
software
is
there,
how
can
it
easily
be
used?
Hopefully
with
the
least
amount
of
work?
I
think
that's
a
lot
clearer,
a
lot
better
understood
and
much
less
controversial.
A
So
please,
if
you
think
we
shouldn't
support
the
developers
to
sign
up
and
chime
in
here,
but
I
think
that
would
really
be
a
great
outcome
of
things
we
would
do
with
autonomic
function
in
autonomic
service
agents.
The
vision
of
enema
is
the
fully
self
job
network
configure
optimize
protect
the
strategy
is
the
incremental
introduction
of
components
to
make
it
easier
to
evolve
existing
and
next
generation
networks
into
that
direction,
including
the
evolution
of
DevOps
for
networks
through
the
support
of
more
agile
and
modular,
developed
and
deployed
networking
automation.
A
So
that's
basically
the
restatement
of
exactly
that.
You
know
development
supported
view
of
the
architecture.
The
basis
for
ongoing
work
and
enema
are
the
framework
and
components
developed
by
enema
so
far
documented
in
draft
ITF
enema
reference
model
standalone
work
not
relating
to
any
component
of
this
framework,
is
welcome
for
review,
but
working
group
adoption
of
such
work
will
be
done
through
explicit
recharging.
So
this
is
a
theory.
This
is
the
most
narrowing
narrowing
part
of
the
Charter.
A
So
please,
you
know
if
you
are
very
interested
in
work
where
you
don't
think
you
can
relate
it
to
the
reference
model
that
we've
written.
Please
stand
up
and
chime
in,
because
I
think
that
that
would
be
very
crucial
to
understand.
The
hope
was
that,
obviously,
the
things
that
we
have
in
the
reference
model
do
very
well
lay
out
the
scope
of
work
items.
The
components
developed
so
far
in
this
enema
framework
constitute
the
autonomic.
F
J
F
Problems
with
that
I,
don't
know
what
it
means
to
our
Charter.
I
know
why
DevOps
isn't
that?
Were
there
I
think
it's
wrong,
but
I
don't
know
what
I
don't
even
know.
What,
even
though
I
just
crossed
that
out.
I
still
don't
know
what
the
sentence
actually
means
from
a
working
group
point
of
view
right,
I.
A
A
Think
primarily,
for
example,
on
the
aasa'
to
think
about
what
is
necessary
to
ensure
that
these
ASAS
can
ultimately
be
deployed
incrementally
into
a
running
system.
Instead
of
just
the
monolithic
update
of
complete.
You
know,
network
operating
software
that
it
comes
with
a
hundred,
a
SAS
or
so,
but
that
you
can't
modularly
upgrade
right
so
that
you
know
it.
F
A
Let's
say
this
so
there's
an
interesting
evolution
in
the
ITF
and
and
it's
one
of
my
rants
write
that
in
the
past,
anything
that
happens
within
the
box
stays
within
the
box
and
the
ITF
doesn't
care
and
any
API
is
between
four
components
in
a
box
that
should
possibly
come
from
different
vendors
and
interoperate
was
still
considered
to
be
just
informational
because
it
was
in
the
box.
The
best
examples
were
any
specifications
of
api's
like
in
the
transport
layer,
but
recently
you
know
in
taps
the
API
specifications
have
been
put
onto
the
standards
track.
A
So
hopefully
this
is
a
good
evolution
and
this
is
effectively
what
I
was
thinking
about
right.
If
we're
building
you
know
ASAS,
can
we
get
standards
for
the
api's
between
software
components
that
we
would
like
to
come
from
different
vendors,
and
there
are
being
you
know,
by
the
operator
put
together
into
the
same
devices.
K
Annexed
my
way,
yeah
one
comment:
yeah
all
solders
I
find
is
also
a
little
bit
confusing
and
actually,
maybe
also
in
similar,
come
to
me.
It
seems
to
be
a
little
bit
the
conviction
drums
first
busy
anime
talks
about
the
self
shop
network,
a
self-configuring,
self-healing,
octopus
self,
optimizing
self
protection
Betty.
If
that
vision
is
fulfilled,
the
question
is:
what
role
actually
is
there
for
DevOps
to
play
so.
K
K
A
A
Devops,
the
DevOps
that
I
said
is
really
about
you,
know,
building
autonomic
functions
and
allowing
them
to
be
really
modularly
introduced
into
the
network,
and
maybe
that's
that's
a
better.
You
know
explanation
than
saying
DevOps
right.
So
maybe
that's
that's
not
yet
a
well
enough
understood,
passwords!
A
K
I
think
there's
understood
as
a
is
word
but
I
think
what
I'm,
having
a
little
bit
troubled
with
is
that
there
are
things
we
have
two
layers
of
abstraction.
One
thing
is
the
function
and
the
other
way,
then
the
other
aspects
of
development
of
those
functions.
So
you
wanna
shoot
both
here.
I,
don't
know
if
I
didn't
help
you
to
propose
better
texts.
That
would
be
great
I.
A
E
E
Those
are
normal
things
so
trying
to
trying
to
describe
how
things
should
be
used
versus
how
things
should
be
developed.
Those
are
two
different
things
and
I
think
the
focus
of
the
chatter
should
be
on
how
to
develop
the
things
or,
basically,
the
specification,
not
necessarily
how
they
are
used
in
all
possible
cases.
Right.
A
Now
I've
taken
the
point
that
instead
of
writing,
just
the
keyword,
def
herbs,
just
the
fact
of
being
able
to
support
you
know,
modularly,
developed
and
deployed
software
should
be
within
the
scope
right.
So
maybe
that's
that's
around
thee,
at
least
that's
what
I
try
to
achieve
with
this,
and
you
know
I,
wasn't
sure
how
well
DevOps
was
understood
to
enable
that,
but
it
seems
definitely
not
a
will.
L
From
highway
so
I
find
that
the
scope
here,
the
configuration
house,
optimization
protection-
is
actually
the
value
very
wide
ambitious
version
and
to
me
that
the
health
and
optimized,
maybe
the
is
to
term,
have
some
overlap.
I'm,
not
sure
whether,
because,
if
you
see
house
the
network
is
health
and
or
the
network
is
protected
or
it's
optimized.
Sometimes
this
a
similar
concept
actually
I'm,
not
sure.
How
would
you
define
clearly
this
job?
What
is
the
concept
of
the
clear
scope
of
this
job.
A
M
M
L
C
M
L
Would
I
could
agree
with
that?
If
you
say
that
the
house
is
something
because
when
a
network
could
have
some
issues
mean,
it
mainly
comes
to
reasons?
Why
is
that
some
software
or
Hardware
mistake
and
the
other
one
is
that
we
got
attacked
by
some
outside
of
I
mean
attractions
so
that
if
you
define
that
too
clearly
I
think
that
that
that's
fine
with
my
site?
Okay,
next.
A
One:
okay:
there
are
known
gap,
so
this
was
basically
the
overview
and
then,
basically
all
these
following
slides
are
about
different
sub
arias
of
what
was
covered
in
the
previous
slide.
So
there
are
known
gaps
in
the
framework,
including
defining
the
domain
boundaries,
membership
of
the
domain
structure,
life
cycle
roles,
authorization,
coordinating
of
autonomic
service
agents,
integration
with
net
operation,
centers
and
reporting
mechanisms,
information,
distribution
within
an
autonomic
Network
interaction
with
yang
based
management
systems.
A
Additional
generic
use
cases
such
as
resource
management
or
SLA
assurance
animal
will
work
on
these
gaps
and
other
aspects
of
the
existing
frameworks,
more
detailed
below
but
exception.
There
are
long
term
issues
that
are
not
yet
well
understood
to
consider
specific
solutions
for
intend
and
tie
into
machine
learning
and
other
AI
techniques.
I,
don't
know,
will
not
work
on
these
issues
without
explicitly
chartering
right.
A
There's
always
a
lot
of
you
know,
research,
interest
and
so
on,
but
will
really
have
a
hard
time
and
we've
seen
that
in
the
past
two
years
to
bring
down
these
pieces
to
something
that
somebody
could
actually
implement
and/or
operate
next,
like
animal
will
continue
to
work
on
fixes
extensions
variations
in
operational
or
implementation
detailing
of
the
overlay
and
I
and
its
component
examples
of
such
work.
It's
not
limited
to,
but
includes
a
and
I
ôm
p
interfaces.
This
is
this
term
operations.
A
Administration
management
provisioning,
for
example,
yang
models
for
the
a
and
I
so
that
obviously,
is
one
of
those.
You
know
most
short-term.
You
know
missing
pieces
of
what
we've
done
in
chart
around
one
and
I've
got
something
you
know
have
worked
lying
up
for
for
a
long
time,
structuring
Ani
virtualization
compounding
right.
There
were
always
these
questions.
Okay,
I
got
one
a
and
I.
How
do
I
connect
another
one?
Variations
of
voucher
formats
that
work
is
already
going
on,
as
we
saw
on
presentations,
brueski
bootstrap
protocol
aspects.
A
Different
proxies
extensions
for
wireless
network
also
is
seeing
a
lot
of
interest
and
then
common
grasp
extensions
for
multiple
use
cases.
All
transfer,
DNS
SD
compatible
grass
objective.
So
these
this
slide
is
really
about.
You
know
the
most
immediate
things
to
leverage
and
extent
missing
pieces
in
what
we've
done
chat
around
one
any.
You
know
questions
suggestions
on
this
one
here.
A
D
A
But
you
know,
and
that's
exactly
good
to
know-
the
different
use
case
is
to
collect
them,
because
I
think
there
definitely
is
a
difference
between
you
know
a
management
plane
that
you
need
to
have
for
something
virtual
versus
the
management
plane
you
need
to
have
for
physical
equipment,
and
so
there
was
just
basically
the
starting
point
to
think
about
it
in
so
that
multi-tenancy
of
actual
hardware
is
definitely
part
of
that.
Next,
like
so
now,
autonomic
functions
just
we're
explaining
for
the
charge
of
the
term
right.
A
It's
the
enema
framework
term
for
distributed
functions,
leveraging
the
Ani,
preferably
automatically
managed.
The
enema
working
group
will
consider
work
items
to
specify
individual
autonomic
functions,
including,
but
not
limited
to
standardized
autonomic
function,
use
cases
such
as
auto
discovery
of
decentralized
services
such
as
those
in
the
network,
operation,
centers,
autonomic
slice
manage
made
and
autonomic
SLA
management
work
on
a
and
I
use
case,
and
autonomic
functions
must
be
detailed
and
complete
enough
in
this.
A
By
the
way
is
a
summary
that
is
meant
to
apply
to
all
the
previous
points
made
not
only
to
the
ones
here
on
the
on
this
slide
must
be
detailed
and
completely
enough
to
support
implementation
of
solutions
that
can
be
deployed
and
operator
standards
track
is
preferred,
but
more
exploratory,
experimental
or
information
informational
work
proposals
will
be
accepted.
Based
on
the
expected
operational
benefits.
E
Ignores
book
donors
to
comments
on
this
slicing
ztf
openly
does
not
work
on
slicing
as
such.
Work
on
slice
management
seems
a
bit
strange
in
in
the
chapter.
So
on
another
hand,
this
is
seems
to
be
a
really
very
specific
work
item
compared
to
the
rest
of
a
chatter
which
is
generic
enough
to
accommodate,
say,
broader
scope
of
the
work
so
from
without
an
ad
hat,
as
as
working
group
interested
working
group,
member
I
would
vote
for
removing
slice
and
completely
from
the
chatter
it
just
does
not
fit
here
and.
A
You
know
I,
don't
even
know
what
any
official
understanding
in
the
idea
of
slicing
is,
except
for
that
we
haven't
been
able
to
figure
out
what
it
means.
But
you
know
to
me
it
could
exactly
be
something
similar
to
the
multi-tenancy
of
hard
work
right.
So
maybe
other
terminology
like
multi-tenancy
would
be
better
terms
to
express
something
but
I
mean
I.
E
A
E
Instance
and
logical
network
entity
models
which
fit
into
that
and
they're
precisely
done
for
this
use
case
and
I
recommend
on
and
basically
what
is
a
criteria
for
the
exit
from
X
from
experiment
if
we
are
talking
about
the
experimental
track
documents.
So
what
is
success?
Great
success
and
exit
criteria
for
the
experiments
this?
This
should
not
be
listed
in
the
chat
or
such,
but
this
is
one
area
that
needs
to
be
thought
of.
So,
if
we're
doing
experiment
when
that
experiment
is
over
and
what
does
it
mean,
succeeded,
failed
or
what
so.
E
That's
something
about
the
criteria
of
experimentation:
this
is
an
IETF
working
group.
This
is
not
a
RTF
which
does
say
experiments
or
research,
so
something
tangible
needs
to
come
out
of
this,
which
results
in
the
reigning
code
and
actual
deployment
experiences.
If
that
cannot
result
in
running
code,
I
would
probably
have
a
question
whether
this
working
group
should
work
on
that.
No.
E
Yes,
I'm
not
disagreeing
with
that,
and
my
comment
is
about
the
words
experimental
informational,
because
those
match
the
document
tracks
and
experimental
document
is
basically
the
question
of
experimental
document.
Is
it
defines
an
experiment,
it's
kind
of
a
continuing
process.
What's
a
criteria
out
of
that,
the
transient
short
I
do
not
have
a
text
to
propose
right
now.
I
will
think
about
this.
Well.
A
I
mean
I'm
just
looking
at
the
experimental
I've
seen
in
the
recent
past
and
I
thought
in
the
IDF.
We
do
fairly
well
understand
that
there
is
risk
more
risk
involved
and
we
don't
recommend
it
for
any
anybody
to
adopt.
But
you
know,
for
you
know,
people
who
want
to
innovate,
get
involved
in
the
experiment
and
obviously
by
default.
The
the
goal
of
you
know,
experimental
stuff
in
the
ideas
is
that
we
figure
out
all
the
things
missing
to
make
it
standards,
so
I
thought
that
was
implied
but
I'm,
not
sure.
F
Michael
Michael
Richardson,
so
so,
if
I
understand
this
slide
correctly,
what
you're
kind
of
premise
your
premise
here
is
that
we
have
built
a
good
foundation.
F
We
know
how
to
pour
concrete,
and
your
proposal
is
now
to
open
it
up
to,
in
my
metaphor,
to
a
variety
of
different
house
shapes
on
top
of
the
concrete
foundation
that
we
have,
and
some
of
them
will
be
crazy,
experimental
things
and
some
of
them
will
be
less
and
but
that
that's
the
idea
that
we
now
have
a
common
base
in
our
in
our
process,
and
now
we
can
go
upwards
in
the
kind
of
concept.
That's
that's
what
you're
thinking.
A
And
you
know,
especially
when
we
come
to
autonomic
functions,
do
things
automatically
right
with
distributed
agents.
I
can
see
a
lot
more.
You
know
resistance
from
people
with
established
a
management
practice,
so
that
I
think
it
would
be
a
lot
harder
for
any
such
you
know
automated
management
to
get
enough.
You
know
support
from
operators
to
go
to
standards.
A
Trek
they'd
rather
want
to
see
you
know
it
being
given
the
label
of
experimental
so
that
people
who
are
you,
know
Courage's
enough
to
put
these
things
into
the
running,
wild,
collect
experience
and
then
come
back
and
revisit
the
you
know,
result
of
that
experience
and
I
think
that
would
be
similar
to
what
I
think
I
received
to
rock
I.
Think
that
I
think.
F
That
this
is
great.
This
is
really
important
stuff
to
do,
and-
and
maybe
it's
okay-
that
it's
in
our
charter,
but
I'm,
I
I,
would
be
surprised
if
we
have
any
the
concern
I
have
is
we
puts
a
bunch
of
stuff
like
this
in
our
charter,
you
have
very
few
people
show
up
to
do
any
work
and,
and
then
we
think
the
working
group
is
more
abundant
dead
and
should
be
closed
right
when
what's
really
happening.
F
Is
that
our
belief
that
the
foundation
is
well
built
was
not
yet
correct
and
that
anyone
in
an
implementation
point
of
view,
maybe
the
protocols
are
perfect
or
maybe
they're
not,
but
but
that
actually
the
people
that
were
out
trying
to
do
innovative
house
construction
and
design
discovered
that
their
foundation
was
miss,
miss
poured
and
they
have
to
rebuild
that
part.
And
so
there's
going
to
be
actually
a
fair
pair
bit
of
delay.
F
I
would
say
until
people
actually
feel
confident
that
the
stuff
underneath
them
is
good
and
so
I
I,
don't
know
how
to
write
that
into
a
charter,
because
it's
more
of
a
industry-wide
kind
of
problem
but
I
just
want
to
say,
I,
think
that
something
we
should
just
be
aware
of
that,
and-
and
maybe
we
don't
want
to
an
option-
is
to
just
defer
this
text
to
another.
Retarder
is
what
I'm
trying.
A
F
A
Oh
no
I
presented
enema.
Also
the
history
with
you
know:
NMR,
G
and
Anna,
but
I
think
you
know
the
the
collaboration
between
di
energy
and
animal
would
be.
You
know
slightly
different,
but
hopefully
you
know
there
could
be
success
story
made
out
of
that
as
well,
but
I
rah
I
have
a
hard
time
seeing
that
anything
from
the
research
side
coming
up
being
implemented
deployed
would
be
more
than
experimental
first,
so
that
that
was
basically
and
that
I
think
is
actually
the
opportunity
to
really
get
interesting.
N
N
Clarification
question:
I'm
Sanjay
from
my
tree
regarding
on
the
autonomic
slice
management,
to
mean
comic
slice
management
as
our
management
of
network
slices
of
many
networks.
Or
do
you
mean
that
the
virtualization
of
a
and
I
the
enema
network
infrastructure
and
then
managing
the
fashion
of
managing
of
that
ni?
Multiple
slice
instances
is
autonomic.
N
If
it
is
a
the
letter,
I
mean
if
it
is
former
I,
think
the
person
who
mentioned
that
I
think
it's
I
know
that
there
is
some
efforts,
a
previous
effort
to
bring
the
slice
management
in
general
into
the
ITF,
but
I
had
a
I
heard
that
there
are
some
issues
with
that.
But
if
it
is
a
letter,
I
think
that's
the
still
within
the
scope
of
anymore
working
group,
because
you
see
so,
as
you
said,
making
multi-tenancy
aspect
of
any
my
infrastructure
as
much
supersizes,
so
yeah
justice.
E
E
The
differences
are
on
a
platform,
specific
aspects,
implementation
aspects
and
so
on,
so
they
are
not
strictly
exactly
the
same
elements
but
from
a
manageability
point
of
view,
and
if
we
are
talking
about
the
service
management
is
the
same
so
trying
to
distinguish
one
particular
flavor
of
that
works,
so
to
say,
does
not
seem
to
be
right.
Well,.
A
I
would
say
in
general,
I
see
one
big
difference,
which
is
that,
if
I'm
building
any
form
of
virtualized
slice,
however,
you
call
it
network
on
top
of
an
underlying
network
that
is
already
running
and
working
and
very
often
going
to
do
the
management
of
that
virtualized
network
through
the
underlying
network.
So
when
we
started
out
with
enema
or
we
basically
with
primarily
focusing
physical
networks
where
there
is
nothing
underlying
right,
so
there
was
a
he
remember-
kind
of
the
out-of-band
physical
management
that
we
wanted
to
try
to
get
rid
of.
A
E
E
Yes,
there
are
subtle
differences
they
need
to
be
accounted
for,
but,
overall,
if
we
are
talking,
if
you
take
any
practical
router
today,
it's
a
virtual
router,
even
if
it's
books
as
a
physical
box,
the
way
how
that
runs
internally
is
very
different
than
it
was
just
a
normal
hard
work
life
and
that's
just
a
reality.
Right.
A
I
can
manage
my
virtual
network
on
top
of
it
in
the
same
way,
which
may
actually
not
be
true,
because
the
ownership
may
be
different
right,
so
I
think,
but
simply
looking
into
that
problem,
space
and
figuring
out
what
even
to
do
right
and,
as
you
said,
maybe
it's
nothing
more
than
informational.
There
is
nothing
new.
It
could
be
perfectly
great
outcome,
but
I
don't
think
we
have.
You
know
done
that
analysis
and
come
to
conclusions.
Yes,.
E
A
No
I
mean
that's
fine,
I
think
the
DevOps
and
slice
seem
to
be.
You
know
too
much
marketing
with
too
few
people
agreeing
on
what
exactly
they
mean.
So
that's
that's
for
sure
that
those
guys
yeah
next
time,
all
right.
Autonomics
software
agents
are
the
enemy
framework
component,
representing
software
modules
that
implement
autonomic
functions.
The
animal
working
group
will
consider
work
items
relating
to
a
si.
Another
aspect
of
autonomic
functions
that
are
applicable,
independent
of
specific
autonomic
functions,
and
it's
a
si.
A
These
work
items
include,
but
are
not
limited
to
design
implementer
guidelines
for
a
si.
A
is
a
lifecycle
management
that
is
also
where
kind
of
this
modular
deployable
would
come
in
is
a
coordination
dependency
result,
and
that's
basically,
you
know
what
as
documents
over
an
acceptance
of
work
items
will
be
based
on
the
perceived
value
to
implementers
and
operators.
Status
of
work
can
be
standards
track
if
the
work
includes
normative
statements
about
note
external
behavior,
such
as
requirements
for
yang
interfaces
on
a
si
exploratory
work
in
this
area
can
be
experimental,
work
describing
note.
A
Internal
only
behavior
will
be
information.
This
actually
is
what
I
wrote
before
I
figured
out
that
recently
I
think
as
as
late
as
one
or
two
months
ago,
they
started
to
put
taps.
You
know,
api's
on
to
standard
strike,
so
I'm
still
trying
to
figure
out
what
the
official
evolution
in
the
ITF
policy
for
standardizing
internal
interfaces
is
that
you
have
any
ATS
X
IDs
in
the
room,
any
idea
what
what
kind
of
the
the
policies
now
are?
A
I
was
always
working
from
the
principle
which
I
think
has
been
there
for
four
decades
that
internal
interfaces
aren't
standardized.
They
must
be
informational.
That's
basically
why
I
wrote
this
text,
but
you
know
it
now
was
very
positively
surprised
that
that
seems
not
to
be
the
case
anymore,
with
at
least
my
interpretation
of
what
happens
in
tabs.
J
Came
from
to
the
Korea
in
the
academia
in
a
point
of
the
Khedive
point
of
view,
is
we
have
our
colony?
We
have
a
sink
about
to
the
add
visual
interest.
Technology
is
applied
to
us
whatever
in
the
environment
and
the
register.
Is
there
any
possibility
to
insert
artificial
intelligence
some
machine
learning
algorithm,
bringing
to
the
some
help
to
the
automatic
function
so.
A
This
is
a
big
reach
are
Turing,
but
it
doesn't
mean
that
if
we
figure
out
that
there
is
something
about
AI
that
you
know
is
really
grounded
enough,
that
it
can
be
implemented
and
deployed
that
that
can't
be
added
to
the
two
through
the
Charter,
through
very
simple
recharging
steps,
as
we
seem
that
difference
between
big
recharges
and
small
individual
item
recharges
that
that
exists,
but
at
this
point
in
time,
I
think
the
first
step
about
those
things
is
individual
submissions
and
then
verifying.
How
much
is
this?
A
G
So
let
me
share
my
feedback
on
the
Charter.
So
far.
Maybe
it
won't
be
easily
actionable,
but
we
are
on
the
fifth
or
six
slides
on
the
Charter
already
and
I've
seen
every
single
buzzword.
That
makes
that
we'll
make
sure
people
will
come
to
you
to
this
working
group
to
do
something.
There
is
AI.
There
is
machine
learning
there
is
virtualization,
there
is
self
whatever
you
know
there
is
lifecycle
management.
G
So,
basically
you
know
we've
got
two
type
of
charters,
the
one
that
are
inclusive,
where
you
say
this
is
what
you're
working
on
and
we
could
be
accepting
more
work
or
the
one
that
are
very,
like
precise
and
right
now,
we've
got
the
mix
of
the
two
and
I'm
wondering
if
this
is
really
white
with
the
right
way.
It's
perfectly
fine
to
have
like
the
vision
of
a
new
map
is
there
is
one
line,
it's
the
vision,
you
know
texting
everything,
that's
where
you
want
to
go
and
there
there
is.
A
Yes,
it
is
some
replication
or
what
we've
done
in
the
reference
model,
but
you
know
it
should
hopefully
help
a
lot
more
to
you
know
the
open-ended
invitation
that
we
have
for
people
to
come
present
work,
but
hopefully
in
the
future,
making
that,
because
you
know
in
the
first
charter,
we
didn't
have
this
more
detailed
specification,
all
right.
So
this
this
is.
This
is
new.
There
are
a
lot
of
people
that
I
think
it
could
bring
in
interesting
work,
but
very
often
it's,
you
know
too
high
level
too.
A
Abstract
not
grounded
enough
not
relating
to
what
we've
done
in
our
mind
before
so
yeah.
So
there's
a
little
bit,
you
know
an
attempt
to
set
the
scope
for
people
who
are
new
to
the
work
and
can
actually
much
more
leverage,
the
Charter
2-0
to
figure
out
what
they
would
need
to
do
to.
You
know
well
work
with
an
animal.
B
Actually
I
mean
kinda
to
agree
with
Ben.
Why?
You
know
this
new
charter
text
looks
so
long
to
including
everything
also,
they
did
here
explanations.
Maybe
what
we
could
do
is
you
know,
try
to
short
the
Charter
tags,
but
put
those
people
in
a
explanation
into
a
you
know,
working
group
document
which
is
not
going
to
be
published
as
RFC,
but
as
a
in
a
document
you
know,
working
document
for
the
working
group
to
say
was,
you
know,
could
be
in
charter.
What
was
motive
here
in
that
document?
How
that
well.
G
G
Is
part
of
the
Charter,
or
not,
maybe
so
the
right
way,
but
having
something
concise
you've
got
a
couple
of
sentences
there
that
could
be
discussed
like
DevOps.
What
do
you
mean?
So
you
must
be.
You
must
have
requirements.
We
matchable
okay,
end
of
story
describing
things
about
experimental
that
might
be
so
not
track
that
potentially
frontal.
Typically,
what
it
in
the
past
is
that
if
you
work
on
this
work
item
well
in
the
Charter,
it
says
not
the
Charter,
the
milestone,
this
one
is
San
attract.
E
Hearing
all
of
this
weakness,
McDonough's
hearing
all
of
this
F
kind
of
worrying
feeling
that
the
scope
might
be
really
too
broad.
Yes,
new
work
should
be
coming
in
into
the
idea.
Does
that
mean
that
it
needs
to
come
to
anima
versus,
say
potentially
new
working
group,
which
focuses
on
a
fragment
of
what
is
being
discussed
here?
I,
don't
have
an
answer
right
now,
but
I
don't
believe
that
animal
should
be
basically
all
gathering
place
for
all
new
work
which
is
somehow
related
to
this
chart.
E
So,
yes,
I
would
agree
that
this
needs
to
be
more
precise
and
probably
what
really
needs
to
be
is
the
deliverables.
You
probably
will
talk
about
that
in
the
coming
slides
and
about
the
timelines
and
the
deliverables.
So
as
the
scope
seems
to
be
broad,
the
set
of
deliverables
needs
to
be
very
concrete
and
specific,
but
remember
the
the.
A
First,
two
slides
said:
I
think
a
rather
strict
scope
insofar
as
that
we're
looking
for
our
pieces
that
are
relating
to
the
reference
model
that
we've
built
and
that
reference
model
I,
don't
think
that
any
pieces
you
know
that
I
can
imagine
within
that
reference
model
would
be
too
broad
right,
so
I
mean
just
because
we
list
these
different
components
on
each
slide.
More
explanatory
doesn't
make
it
actually
wider.
It
just
provides
more
details
right.
So
our
our
very
much
question,
you
know:
what
are
your
criteria
for
broad
I?
A
E
Not
a
number
of
slides,
but
at
least
the
list
of
documents
that
you
have
there
is
already
quite
large.
So
if
we
are
talking
about
for
meeting
cycle
intervals,
the
question
is:
can
those
documents
actually
be
progressed
to
a
tangible
result
during
the
for
meeting
cycles
and
I'm
not
expecting
an
answer
for
you,
I'm
just
commenting
and.
A
O
Jefferson
I
agree,
shink
justification
when
you
mentioned
another
document
to
describe
the
work
done
in
the
the
working
group.
You
know
you
know
I
got
confused
because
at
the
end
of
the
day,
what
what
the
fine
is,
what
we'll
be
presenting
in
the
the
meeting
is
the
Charter
so
because,
in
the
you
know,
like
I,
don't
know
what
this
can.
You
know
can
help
us.
A
Autonomic
functions
not
well
enough
understood
to
allow
adoption
by
anima
right
so
working
group
items
accepted
under
this
Charter
will
be
tracked
through
milestones
and
have
by
default
to
be
brought
into
working
group
last
call
and
not
more
than
four
ITF
cycles.
Work.
I
didn't
expect
it
to
take
longer
time
are
subject
to
ADA
approval
right,
so
you
know
we
can.
We
can
cut
out
a
lot.
A
You
know
explanations
from
the
middle
of
the
document
just
listing
you
know
without
explaining
and
without
giving
examples
and
then,
for
example,
having
as
Shane
said,
you
know
and
in
you
know,
just
informational
document
explaining
a
little
bit
better
to
newcomers.
What
that
could
mean
in
detail
and
then
we
have.
A
First
Charter
but
yeah
this.
This
is
new
to
us.
So
please
help
in
getting
the
Charter
outline
that
that
you
prefer
I
mean
then,
while
you've
you've
seen
a
lot
more
than
Ignace
I
think
over
the
years,
what
what
work
was
being
brought
into
animal
working
group
and
that's
I-
think
what
we're
trying
to
manage.
G
So
since
you
mention
my
name
binoculars,
let
me
go
back
to
Mike,
so
maybe
what
I
would
do
right,
but,
okay,
it's
just
an
advice,
because
I've
not
been
following
an
imagined
little
discussion,
but
typically
there
is
a
mix
of
many
thing.
The
Charter
there
is
a
mix
of
the
process
right.
The
last
paragraph
is
about
tracking
through
milestone
and
for
ITF
cycle
and
then
is
edie
approval.
But
you
know
it's
too
late,
for
example,
because
they're
going
to
say:
okay,
we
accept
this
document
and
another
for
I
have
meetings,
we're
going
to
say.
G
Well
now
it's
too
long
you
go
back
to
idiot,
prove
all
doesn't
work.
Then
you
explain
also
for
process
point
of
view
that
it
could.
It
should
be
an
attractive,
could
actually
Montel
it's
a
process,
stuff
right,
so
this
in
my
man
should
be
removed.
Then
there
is
like
all
different
use
cases.
Then
there
is
the
vision.
Then
there
is
all
the
keywords.
I
would
simplify
this
a
lot.
G
There
is
I
was
reading.
The
Charter
and
I
was
thinking
that
I
could
present
like
10
new
drafts
here
and
pretends
that
they
fit
into
the
Charter
in
one
of
the
previous
slides
I
was
reading
about
operator
interests,
though,
if
you
could
go
back
yeah
acceptance
of
work
items
will
be
based
on
the
perceived
value
to
implementers
and
operators.
Sure
it's
like
normal
procedure.
Why
did
you
have
that
I'm
going
to
come
with
the
first
lighting?
G
G
C
G
A
But
wait
a
sec
knows
are
two
different
things
right.
The
first
thing
is
what
is
explanatory
that
isn't
needed
to
be
put
into
the
charger
all
right.
So
that's
that
certainly
very
well
taken.
There
is
probably
a
lot
more
that
we
need
to
explain
to.
You
know:
participants
in
the
working
group,
then
what
you
know
the
IETF
expects
us
to
put
in
the
Charter
point
well-taken
right
so,
and
we
can
work
through
that.
We'll
just
remove
these
things
and
then
we'll
hear
back
from
you
know
which
points
we
should
put
back
there.
A
The
other
part
is
about
the
broadness
of
scope,
right
and
I.
Think
that,
basically,
is
something
where
we
try
to
be
much
more
precise
in
constraining
this
into
implementable,
and
you
know
related
to
the
anima
reference
model
and
the
components
that
we'll
want
to
do
and
excluding
the
components
we
don't
want
to
do
right.
So
if
you,
if
you
have
any
more
constraints
that
you
think
we
should
do,
that's
fine
right
just
because
we
have
many
examples.
A
There's
obviously
not
mean
that
all
of
these
will
be,
you
know,
have
good
proposals
for
work
on
them,
and
so,
from
that
perspective,
it's
equally
fine
to
remove
all
the
examples
of
the
document
and
also
move
them
out
to
an
informational
document
right,
but
I
think
the
primary
questions
field
is
what
is
the
minimum
necessary
definition
of
the
scope?
And,
if
you
have,
you
know,
opinions
about
how
to
do
that
different
or
better
than
we've
tried
to
do
it
here
that
very
much
welcome
at.
B
Right
yeah,
we
have
a
very
good
discussion.
Our
return
texts
today
are
terrific
much
longer
than
we
thought,
so
we
have
only
20
minutes
left
and
we
have
six
or
seven
presentations,
so
I
have
to
cut
it
off
and
I'm
prefer
to
give
time
to
those
who
never
it's
you
know
presented
you
our
working
group,
so
Kerry
have
the
trust
network
from
concern.
N
Yeah
yeah:
this
is
a
sanction
from
a
tree.
Actually,
this
is
our
update
from
last
July
meeting
I
wanted
to
on
trust,
networking
and
procedures
for
tonight.
Networking.
Could
you
that's
right.
Yeah
during
this
mature
meeting,
I
got
two
comments.
One
is
on
the
elaborating
more
on
differences
between
our
proposal,
trust,
economic
domain
and
enema,
a
security
framework,
and
the
second
comment
was
on
providing
implementation
experience
of
KD
based
on
anima
as
a
use
case.
N
So
so
we
describe
that
in
if
you
read
the
section
for
filling
videos
on
item
by
item
and
those
are
the
criteria
that
domain
as
a
whole
domain
members
domain,
boundary
policy
now
the
connection
to
Internet
security
model
and
any
operation.
So
for
each
criteria,
we
try
to
provide
all
key
explanation
and
also
against
any
math's
good,
very
much
okay,
and
so
then
we
will
play.
N
N
A
note
should
have
identifier
an
oath
and
compatible
ID
and
in
our
tid,
although
it
is
not
fully
comparable
with,
but
without
any
my
yet,
but
when
we
find
out
the
ad,
we
try
to
use
this
ID
as
a
words
call
self-certifying
ID,
but
there
can
be
I
think
we
can
utilize
the
Animas
local
device
ID
as
as
a
certified
body,
and
preferably
if
we
can
extend
it.
Maybe
we
can
extend
it
with
probably
keep
mission
for
the
global.
N
Technology
in
terms
of
technology
kv
allows
any
technology
right.
Then,
in
the
current
enema
framework,
the
assumes
the
idea
of
the
IP
technology,
but
on
KD
in
principle
allows
any
any
collosseum
type
you
like
there
was
or
other
analogous
to
tonight,
as
well
as
long
as
the
domain
is.
A
trust
domain
is
properly
defined,
so
only
time
for
inter
domain
communication
is
that
KK
must
aware
of
mechanisms
for
both
domain
and
then
takes
a
role
of
translation.
N
N
The
Gateway
performs
the
the
translation
between
the
local
domain
a
domain
specific
to
2d
to
the
Internet.
In
this
case
we
assume
the
Internet
as
global
global
trust
domain
and,
if,
by
by
having
separate
individual
domains
to
join
the
global
Internet
domain,
and
the
eight
ways
in
between
can
translate,
it
performs
the
making
mechanisms.
So
one
example
is
in
the
next
slide
shows
that,
as
you
can
see
here,
we
have
a
to
trust
domain
and
the
in
between.
N
We
have
a
cholera
internet
and
from
each
trust
domain
the
news
reading
they
trust
domain
to
communicate
each
other
through
this
global
internet.
It
has
go
through
the
it
with
interest
domain,
it
has
its
own
private,
ID,
ID
and
when
it
goes
out
to
the
global
Internet
is
global
ID.
In
this
case,
the
just
centric
domain
itself
is
also
IP
based
version
that
then,
then
it
uses
private
IP
as
internal
ID
and
then,
when
it
goes
out,
uses
the
public
IP.
N
So
mapping
between
the
translation
mapping
between
them
is
done
by
the
domain
gating,
but,
as
I
said
just
just
to
me
can
be
technology
independent.
Although
this
is
example
shows
up
the
private
IP
cases,
but
actually
we
can.
We
can
also
have
different
kinds
of
technologies
locked
and
the
right
side
of
how
these
mapping
is.
Okay,
I.
N
Think,
having
too
much
time,
I
guess
so:
scared,
just
a
private
private
model.
I
think
we
can
forget
this.
You
can't
read
an
operation
wise
and
I
think
you
cannot
refer
to
them
and
for
the
second
comment,
we
have
have
a
prototype
implementation
on
our
just
autonomic
domain.
This
is
a
the
software
mojo
that
we
designed
the
next
and
description
on
that
and
this
sequence
flow
diagram
that
how
its
components
interact
each
other,
and
we
also
built
our
test
bed
to
to
test
our
components-
are
that
we
prototyped
a
big
build.
N
B
P
P
So
this
is
about
the
constraint:
don't
proxy:
it
fits
in
the
Brisky
escape
proposal
where
Brisky
uses
EST,
HTTP
and
TLS.
We
have
a
constraint
approach
where
the
don't
proxy
is
part
of
that
and
actually
it
uses.
What
it
does
is
that
the
certain
proxy
which
you
will
find
in
the
TFS
16
draft
will
be
replaced
by
the
poxy
which
will
be
finding
next.
P
So
this
is
in
graphic
gray
that
you
will
see
the
pledge,
the
one
which
wants
to
start
be
part
of
the
network
and
you
have
the
HT
server
and
in
between
there
is
this
circuit
proxy
in
the
HTTP
TLS
ROM
and
for
the
constrained
device
we
talking
about
join
proxy
I
talks
lindlar
go
to
the
pledge
and
it
talks
global
to
the
eesti
server
and
it
uses
co-op
and
DTLS
instead
of
the
HTTP
TLS.
Yes,
please,
it's
fast
enough.
P
The
constraints,
so
the
transport
format
is
that
what
you
want
to
do
between
the
pledge
and
and
the
proxy
is
that
between
the
joint
proxy
and
HT
server,
you
want
to
remember
what
actually
the
link
local
address
from
the
from
the
pledge
boss.
Well,
the
joint
proxy
and
the
HT
server
Tory's
out
of
his
global
addresses.
So
what
we
have
sent
is
from
the
pledge
to
the
jumper
see
we
have
in
detail
s
payload
the
Detailers
payload
is
encrypted.
We
don't
want
to
join
proxy
to
decrypt
it
and
encrypt
it
again.
P
So
what
we
will
have
is
the
DTLS
payload
will
be
unchanged
sent
over
from
the
joint
party
to
is
this
over,
and
so
what
you
do
with
in
packets,
on
the
joint
proxy
to
eesti
server,
in
which
we
also
keep
the
link
local
addresses
and
some
other
information
about
the
place,
and
that
is
then
sent
over
to
the
HD
server.
What
do
you
know
useful?
P
That
is
a
new
format
which
is
defined
recently
by
the
coworker
group,
is
that
you
will
have
the
possibilities
which
was
not
there
before
and
which
actually
was
supported
by
HTTP
but
present-day.
But
my
coop
is
that
you
can
have
different
media
formats
and
in
this
case,
what
we
will
do
is
that
we
will
define
several
media
formats
in
the
in
role
in
the
payload
image.
Some
of
those
actually
specifies
the
the
address
of
the
page
and
in
the
other
one
you
have
to
modify
details
payload.
P
Clear,
okay,
so
maybe
Michael
already
explained
it
that
we
have
to
those
draft
relations.
So
there
is
the
Brisky
which
is
piece
which
is
part
on
which
this
is
all
based.
We
have
the
voucher
document,
which
is
also
here
in
animal,
which
is
well
well
enter
Tommy,
cheese
and
IRC
these
days
and
then
in
ace.
P
We
have
something
which
is
called
easty
coop
s,
which
defines
how
you
do
est
from
a
pledge
to
the
to
the
HT
server,
and
we
have
an
other
one
which
is
going
to
be
discussed
by
Michael,
which
tells
you
about
the
constraint
voucher,
and
it's
tells
you
also,
which
are
the
Biscay
extension
to
the
est
functions.
How
that
is
going
to
be
done.
I
won't
say
anything
about
that.
P
B
P
F
B
F
Swensen
we're
doing
on
a
secure
home
gateway
and
with
Syrah,
and
one
of
the
things
we
did
not
solve
was
how
to
initially
cook
up
a
phone
and
there
the
Gateway
next
slide
next
line
so
Gateway
my
ot
device
and
as
firewalls
all
things.
We
have
an
app
which
is
shown
on
an
iPad
here
and
it
needs
to
do
a
secure
connection
to
the
Gateway
effectively.
It
needs
to
enroll
in
the
domain
next
slide.
F
F
Of
course,
smartphones
get
run
over
dropped,
just
died,
having
the
root
of
all
trust
on
your
vulnerable
device
might
seem
like
a
reasonable
policy
from
gee.
It's
it's
closest
to
me
and
front
most
friendly
to
interface,
to
it's,
not
so
good
from
a
resiliency
point
of
views
with
real
people
and
real
stuff.
So
this
proposal
basically
says
well.
No
look!
We're
gonna,
put
the
the
registrar
on
the
device,
that's
going
to
be
the
kind
of
infrastructure
for
the
home,
but
since
it
doesn't
have
internet,
yet
it
can't
talk
to
the
masa.
F
So
the
idea
is,
we
actually
have
a
next
slide.
Yeah
there's
the
idea.
Next
slide
next
slide
there.
So,
basically,
we
just
brewski
have
a
request:
a
votre
request,
turning
the
what
was
a
forward
process
where
the
pledge
talked
to
the
Registrar,
which
talks
the
Massa
and
then
answers
into
one
where
the
pledge
talks
to
the
registrar
and
then
relays
the
request
to
the
basa
and
release
data
back
and
don't
think
in
this
time.
I
can
explain
to
why
DPP
was
important.
F
Doing
a
ideally
like
to
leverage
it
also,
the
crypto
was
identical
such
that
when
we
can't
have
smartphones
that
have
interfaces
that
they
can
speak,
the
right
management
frames
that
DPP
requires
that
it's
not
kind
of
all
just
falls
out
and
becomes
that
today
we
can't
do
that,
and
so
this
is
essentially
a
DBP
doesn't
use
masses
and
other
things
like
this,
and
so
there
is
some
changes,
but
the
idea
is
that
the
proof
of
who's?
What
and
the
QR
codes
that
DPP
use
would
be
leveraged
to
do
things.
F
We
I
thought
a
lot
about
using
EAP
noob,
but
none
of
the
devices
I
care
about
have
any
way
to
import
or
output
large
QR
codes.
So
specifically
the
Registrar
does
not
ability
to
do
that,
and
while
we
could
make
something
up
new
involving
flash,
the
LEDs
in
the
front,
I
decided
that
was
harder
than
I
wanted
to
do
next
slide.
Please
there's
some
time
sequence
diagrams.
If
you
really
want
to
go
into
it,
we
can
discuss
it.
No
time
right
now,
keep
going
and.
F
Just
some
other
little
notes,
if
you're
curious
about
how
we
are
securing
the
connection.
So
fundamentally,
this
connection
from
your
smart
phone
to
this
to
this
router
is
a
TLS
connection
over
link,
look-over
or
actually
ula
addresses,
and
we
are
intend
to
ship
the
routers
with
certificates
built
with
their
ula
address,
already
pre-selected
and
DNS,
already
populated,
with
a
name
and
therefore
were
able
to
put
something
in
a
certificate
on
the
device.
So
that's
the
TLS
server
certificate
side
of
things.
How
is
that
side
authenticated?
F
You
know
it's
real
device
and
the
name
matches,
and
so
the
problem
is:
how
do
we
get
the
TLS
client
certificate?
And
why
are
we
using
a
TLS
client
certificate
because
we
don't
want
any
passwords
next
slide?
I,
don't
really
think
it's
belongs
in
anima,
yet
it
leverages
all
the
basically
an
animal
protocol.
So
if
not
here,
then,
where
and
I'd
only
have
time
for
that
special
either
thanks
thanks.