►
From YouTube: IETF92-SACM-20150327-0900
Description
SACM meeting session at IETF92
2015/03/27 0900
B
So,
instead
of
talking
about
three
different
things
will
just
talk
about
the
work
that
we
did
during
the
week
and
have
a
discussion
around
it.
Then
we'll
have
a
presentation
about
the
open
group
and
sockem
etsi.
I
guess
tony
is
going
to
be
presenting
TC
cyber
and
Sock'em
and
3gpp
siii
and
Sock'em,
and
then
we'll
talk
a
little
bit
about
milestones
and
decide
a
way
forward.
A
A
B
Right,
let's
see
what
do
we
want
to
go
here?
Okay,
so
basically
we
met
three
times
and
had
a
bunch
of
discussions
along
the
way
at
first
weeks.
You
know
that
during
the
first
session
we
talked
a
little
bit
about
scope
and
then
I
said
wait.
That
is
a
bad
board.
So
let's
talk
about
focus
and
and
then
ultimately
we
started
talking
about
prioritization.
So
we
had
a
very
long
prioritization
discussion
and
I.
B
We
also
from
that
had
a
state
discussion
and
it
feels
like
we
have
some
consensus
that
were
interested
in
state
/
behavior.
So,
for
example,
we're
not
going
to
go,
and
it's
not
our
goal
to
verify
that
a
web
server
will
not
negotiate
a
week.
Cipher
suite,
though
it's
configured
to
not
negotiate
a
week,
cipher
suite
we'll
look
at
the
configuration,
but
we
won't
necessarily
go
and
interact
with
that
web
server
to
find
out
what
it
will
actually
do.
B
A
B
We
definitely
recognize
that
we
need
to
drive
the
information
model
home,
which
basically
means
get
it
to
a
working
group.
Glass
call
state.
So
then
we
can
work
on
submitting
it
to
iesg,
and
yesterday
we
came
up
with
a
list
of
things
that
we
might
need
for
the
information
model
to
be
more
or
less
complete.
We
need
models
to,
and
these
are
kind
of
in
plain
terms
just
so
we
can
start
with
more
or
less
shared
understanding.
B
So,
even
in
the
quote,
agent
lists,
you
know
if
you're
interacting
with
a
network
device,
for
example-
and
you
connect
to
a
shell
that
she'll
is
really
the
client
that
you're
interacting
with
to
get
the
data.
That's
what
we
mean
by
client,
it's
not
necessarily
a
second
component.
That's
doing
the
assessment
that
would
have
to
be
installed
on
that
thing.
B
Okay!
Next
sorry,.
A
A
C
B
Think
that
is
a
by
ok,
so
my
impression
at
this
point
is
that
most
of
these
things
at
this
level
would
be
in
the
information
model,
so
the
information
model
kind
of
needs
to
have
these
submodels
in
them
more
or
less,
or
we
could
have
separate
models.
I
think
that
how
they
break
out
and
drops
is
it
is
that
your
question
no.
C
D
D
B
A
F
Dave
ball
tomorrow,
I'd
also
like
to
+1,
Jessica
and
and
then
address
Jim's
question
I
think
we
really
need
two
things
we
need.
We
needed
a
listing
of
all
of
the
available
attributes
that
we
can
can
collect
and
evaluate
for
a
given
end
point
whether
or
not
they're
defined
or
defaulted
or,
however,
that
that
mechanism
would
work
on
on
a
given
endpoint,
because
nine
times
out
of
ten
we're
not
going
to
be
interested
in
collecting
the
full
information
that
is
available.
On
that
end,
point
we're
generally
going
to
be
interested
in.
F
You
know
some
subset
that
is
relevant
to
you
know
to
the
organization's
use.
So
we
need
to
be
able
to
select
from
the
menu
of
attributes,
essentially
that
we
want
to
collect
and
evaluate,
and
then
additionally,
we
need
another
method
of
representing
the
actual
values
that
exists
on
the
endpoint
for
those
for
those
attributes.
So
I
think
we
really
need
both
and
there,
and
both
things
are
doing
slightly
different
things.
A
B
G
G
Instead,
we
should
understand
that,
as
the
requirements
are
written
now
and
a
better
way
to
potentially
make
progress
is
to
accept
that
we're
going
to
accept
a
lot
of
different
standards
operating
in
in
the
space,
so
sticks
taxi
I,
oh
def,
IP
fix
oval,
you
no
soup
du
jour.
All
of
them
are
going
to
be
there.
Well,
we
should
focus
on
is
just
the
key
elements
we
need
to
be
able
to
connect.
G
You
know
the
key
identifier
is
going
from
an
endpoint
to
the
collector,
from
the
collector,
to
whatever
repository
from
the
query
capability
to
the
repository
such
that
those
things
can
then
actually
operate
on
the
data
and
understand
what
data
is
available
right
and
that
data
can
be
in
any
of
those
formats.
So
the
data
model
is
data
models
from
other
systems.
G
Unless
we're
going
to
tackle
like
kind
of
building
the
one
true
ring
but
I,
don't
think
we
should
I
think
we
should
go
with
the
scooby-doo
decoder
ring
model
and
we
should
focus
on
doing
that
piece
instead
right,
which
means
that
all
those
underlying
data
models
are
welcome.
They're
already
done
whatever
vendor
thinks
that
you
know
you
know
whatever
data
model
they
have
is
the
best
thing
since
sliced
bread
good
for
them
work
on
an
interoperability
layer
that
lets
those
things
communicate.
E
G
G
So
I
think
what
I
I'm
effectively
saying
is
you
need
to
have.
We
would
need
to
be
able
to
define
enough
of
an
information
model
and
data
model
to
be
able
to
define
the
metadata
of
what
gets
put
into
the
repository
such
that
you
can
arbitrarily
determine.
What's
there
that
you
could
use,
but
then
how
you
act
on
that.
That's
a
whole
nother
problem
that
we
just
kind
of
don't
solve
necessarily
so.
E
I
think
this
is
really
hard.
So
looking
at
you
know
a
few
of
us
in
the
room,
we
have
to
worry
about
different
things.
So,
taking
off
my
ad
hat
at
emc,
we
have
to
worry
about
storage
and
that,
right
now
the
most
popular
interface
is
a
restful
interface
with
me.
Just
a
second
recall,
with
this
I
would
let
a
whole
project
on
this.
No
wait,
one,
no,
no,
no,
the
the
data
model.
E
C
But
we
should
not
dunno
much
kind
of
speaking.
So
should
not
forget
that
we
are
talking
here
about
the
information
model
which
is
kind
of
the
higher
level
protocol
independent,
while
the
real
interoperability
will
be
measured
at
the
data
model
level,
with
instantiates
in
specific
data,
modeling
languages,
information
model,
yeah.
H
So
I
think
I
want
to
try
and
regurgitate
because
I,
oh
sorry,
I'm
Nancy,
what's
my
last
camwyn
jet,
like
I'm,
not
awake
yet
so
I
think
I
agree
with
Chris
in
the
sense
that
the
way
I
think
of
an
information
model
is
the
abstraction
of
the
different
types
of
information
and
the
interactions
or
operations
that
need
to
occur.
How
they
get
instantiated
into
a
particular
data
model.
I
believe,
is
a
group.
We
have
agreed
that
there
will
be
many
and
again
to
reiterate
to
Kathleen's
point.
H
The
data
model
is
yet
another
level
of
extraction
that
out
of
scope
for
us
in
the
group.
How
we
do
the
storage
considerations
underneath
is
up
to
the
implementers,
because
there's
going
to
be
optimizations
on
data
size
deployment,
size
and
all
of
that.
So,
while
I
like
the
focus
and
there's
these
lists
of
things
right,
I
just
want
to
make
sure
that
we're
staying
still
for
the
information
model,
we're
staying
at
the
abstraction
to
allow
for
the
sticks.
The
text,
sorry
Thanks.
D
A
F
Agree
with
all
of
that
too
and
I
think
that's
actually
somewhat
what
is
meant
by
by
this
right,
because
we're
not
saying
that
we're
going
to
enumerate
how
you
check
for
minimum
password
length
for
microsoft,
windows
or
red
hat
linux,
or
you
know
all
of
the
other
variations
that
exist
on
that
on
that
theme
right.
What
we're
talking
about
here
is
how
do
we
generalize
a
way
to
record
to
allow
those
things
to
be
instantiated
in
a
model
that
would
allow
for
some
interoperability,
I.
G
So
Kristen
hacia
I,
don't
want
necessarily
to
create
interoperability
at
a
data
model
level.
I
think
the
reality
is.
If
we're
going
to
make
progress,
it's
going
to
be
caveat,
Emperor
a
little
bit
of
buyer.
Beware,
because
if
you're
going
to
accept
multiple
data
models
into
your
ecosystem,
then
you
know
interaction
and
multiple
data
models
is
just
too
hard.
It's
going
to
take
too
long,
and
to
that
point
it's
not
clear
to
me
that
you
need
to
identify
what
a
hardware
how
to
do
or
handle
a
hardware
inventory
and
I.
G
Don't
necessarily
think
that
we
need
to
understand
or
do
how
to
identify
a
software
inventory.
What
we
potentially
need
to
do
is
understand
that
there
are
code
points
that,
oh
you
know
we
could
have
Swedes,
we
could
have
CPEs,
we
should
have
an
information
model
and
our
data
model
should
allow
the
exchange
of
those
types
of
data
points
to
between
sockem
components.
And
you
know
how
does
that
interoperate?