►
From YouTube: NMRG Interim Meeting, 2020-02-28
Description
NMRG Interim Meeting, 2020-02-28
A
So
the
the
agenda
for
today
is
the
following,
as
you
may
have
seen
in
the
in
the
mail.
So
we'll
tell
you
a
bit
about
the
the
news
of
the
research
room,
mostly
about
the
next
meeting
that
we
will
have,
and
then
we
have
two
slots
for
basically
discussing
as
a
the
the
two
main
drafts.
Now
that
we
have
one
is
related
into
intent
classification.
A
So
I
know,
if
all
guys
join,
I'm
not
sure,
but,
as
you
may
have
seen,
there
has
been
some
initial
discussion,
starting
on
the
mailing
list
about
processing.
A
Basically,
the
commands
I
received
follow
following
the
the
call
for
adoption
that,
because
this
graph
has
not
been
updated,
but
of
course
they
will
try
to
to
to
revise
the
document
according
to
the
comment
and
have
a
new
version
available
available
in
the
next
couple
of
weeks,
I
would
say-
and
the
second
graph
is
the
one
that
has
been
that
has
been
adopted
in
the
other
as
a
research
as
a
group
document.
A
So
one
of
the
the
concept
and
definition-
and
I
think
laura
will-
will
give
some
some
some
feedback
on
the
different
comments
they
received.
And
oh,
they
want
to
proceed
in
this
draft
for
the
for
the
following
work.
A
Then
we
will
have
a
technical
presentation
from
amina
on
slice,
management,
automation
and
sli
compliance.
So
I
mean
now
you
will
have
30
minutes.
As
I
said,
if
you
need
less
it's
it's
okay
as
well
as
you
said
that
it
will
be
more
than
enough,
but
don't
be
a
don't
get
in
the
river
for
the
presentation
we
have
time,
and
then
we
have
any
other
business.
A
Of
course,
and
as
we
discussed
I
discussed
rapidly
with
lauren,
we
may
also
introduce
a
bit
what
will
the
procedure
for
starting
to
to
work
on
the
document
about
the
ia,
artificial
intelligence,
research
challenge
for
network
management?
A
A
A
So
now
we
have
two
two
hour
session,
two
two
hour
session
that
has
been
allocated
to
us
so
initial
proposal,
but
it
may
have
changed
already
it's
wednesday
afternoon
thursday
morning.
I'm
not
sure
it's
still
this
plan
because
it
can
be
changed
a
bit.
A
And
so
then,
today
we
can
discuss
what
will
be
the
content
of
this
meeting,
so
the
initial
idea
was,
of
course,
to
cover
the
free,
the
free
terms
that
we
have.
We
have
in
the
already
started,
to
discuss
during,
of
course,
the
the
reshartering
and
this
discussion
on
mailing
lists
and
the
last
iatf
meeting
so,
namely
we
have
ibn.
Of
course
we
have
the
thing
related
to
artificial
intelligence
and,
as
has
been
highlighted
last
time
in
singapore,
we
were
also
thinking
about
considering
other
network
management
frameworks.
A
So
this
is
basically
the
status
now,
and
this
is
open
to
discussion.
We,
we
actually
have
a
few
feedback
on
who
wanted
to
to
present.
I
hope
we
will
have
update
on
the
different
on
the
different
drafts.
A
A
So,
in
addition
to
the
to
the
drafts
that
you
have
currently
have
for
artificial
intelligence,
our
plan
is
to
have
more
technical
presentation.
Basically,
as
we
have
been
requested,
also
to
have
this
from
the
irtf
chair.
A
A
We
know
that
it's
not
the
best
location
for
people
to
easily
come.
But
yes,
if
you,
if
you
know
somebody
who
could
be
interested
to
to
present
in
this
aspect,
would
be
very
appreciated.
A
Of
course
really.
I
will
also
discuss
the
as
the
the
to
other
point,
which
has
the
document
that
we
would
like
to
to
have
on
the
different
research
challenges
and
also
the
status
of
the
different
competition,
I
would
say,
are
a
challenge
we
want
to
to
have
for
network
and
network
management,
and
then
it's
also
open,
as
I
said,
to
emerging
network
management
framework.
No
again
again
is
more.
We
are
also
looking
for
a
presentation
regarding
that.
A
A
Okay,
so
then
we
we
are
thinking
about
sign
meetings,
so
nothing
is
is
planned
yet,
of
course,
but
maybe
we
have
signed
an
ibn
so
related
to
to
basically
the
new
arcade
team
on
ibn
in
the
room
mat.
We
have
all
those,
maybe
more,
let's
say
session,
assignment
dedicated
to
drafting
writing
the
document
and
also
for
the
last
point
as
before.
If
it's
not
presentation
during
session,
maybe
we
can
have
side
meeting
to
discuss
about
what
will
be
the
different
different
orientation.
A
We
should
consider
for
the
new
emerging
network
management
frameworks,
so
this
is
basically
the
initial
plan
and
possible
submitting.
So
now
we
like
to
during
the
meeting
today
also
to
to
have
your
feedback
regarding
this.
This
proposed
agenda
and
to
know
if
you,
if
you
have
any
other
id,
we
would
like
to
share
what
could
be
interested
to
to
do
also
in
vancouver.
B
Want
just
a
very
minor
thing,
because
we
received
a
request
from
the
irtf
chair:
try
to
address
some
agenda
conflict
for
the
ietf
meeting.
B
A
A
B
B
C
Sorry
I
joined
late.
Can
you
hear
me.
C
B
This
is
a
situation
that
is
evolving.
We
had
some
discussion
also
in
nokia
and
some
with
some
other
companies
with
respect
to
iitf
meeting,
but
also
some
other
events
and
civilization
meetings.
The
information
I
have
so
far
is
that
for
iatf
they
stick
to
what
they
announced,
that
they
are
monitoring
the
situation.
There
is
no
consolation
announced
as
of
today.
B
There
are
some
companies
putting
travel
restrictions
for
their
employees,
for
instance,
ericsson
and
juniper
huawei
has
already
for
several
weeks
already
some
specific
traffic,
their
colleagues
from
from
china.
So
this
is
being
monitored.
We
have
you
have
to
watch
the
announcements
that
that
will
be
made
by
the
ietf
leadership
teams,
but
I
think
we
have
the
coming
week
coming
weeks
to
to
see
how
this
evolves
yeah.
A
A
Okay,
so
if
there
is
no
other
command
for
this
and
just
move
to
the
next
one,
so
we
have
also
other
plan
for
for
for
meetings.
We
we
are.
A
I've,
requested
the
norms
organization
committee,
so
namaste
is,
let's
say,
leading
conference
in
network
management,
more
academic,
I
would
say
it
will
be
in
budapest,
so
they
they
agree
to
that
we
will
prevent,
will
depend
on
there.
A
It's
we
have
been
used
to
presenting
norms,
or
I
am
so,
but
we
never
found
the
let's
say
the
right-
the
right
way
to
to
be
there,
and
so
we
tried
different
different
thing
each
time
almost
so
the
idea
here
will
be
to
have
an
introduction
of
what
is
an
energy
during
the
plenary
plenary
session
and
to
make
maybe
the
community
aware
of
an
emergency
in
order
maybe
to
attract
more
more
people,
in
particular
the
academic
people,
to
an
emergency
as
well
and
to
have
let's
say,
high
visibility.
A
They
also
propose
us
to
have
an
emergency
poster
and
of
course,
with
somebody
like,
I
will
be
there
as
well,
so
I
can
be
there
and
present
what
is
an
emoji
and
so
on.
A
A
That
will
be
fully
integrated
into
the
program
of
the
conference,
so
that,
rather
than
just
people,
see
an
emergency
on
the
program,
they
will
have
a
real
program
with
technical
presentations
that
can
be
in
their
interest
and
then
it's
also
a
way
to
to
to
bring
the
awareness
regarding
an
image
in
general,
of
course,
for
the
inviting
speaker
we
are
not
yet
we
have
not
constant
invitation.
We
are
discussing
a
bit
some
logistics
with
with
the
oc,
but
in
the
next
couple
of
days
I
would
say
you
should
have
a
final.
A
Then
there
is
a
question
to
have
a
virtual
meeting
double
because
not
everybody
can
attend
norms,
of
course,
and
I
think
we
should
not
be
able
to
provide,
and
it's
not
it's,
not
an
energy.
Let's
say
team
meeting
as
it
is
a
usual
and
we
should
not
be
able
to
provide
the
remote
participation
for
norms.
I
would
say
I'm
pretty
sure
with
that.
A
The
next
event
will
be
at
the
ipob
conference
or
if
it's
in
japan
I
just
put,
may
twice,
but
you
see
it
will
be
in
japan
here.
We
have
also
be
invited
to
organize
different
things.
One
special
session
on
ibm
with
a
is
a
call
for
for
paper
and
presentation
and
one
nmrg
meeting
as
we
as
more
regular
and
emerging
meetings.
A
That
will
be
the
day
before
the
conference,
and
we
have
been
also
invited,
lauren
me
and
to
to
laura
mito
to
make
a
keynote
most
probably
I
will
do,
and
we
do
myself
and
eventually
demo,
but
we
think
with
lauren
can
be
hard
to
present
them
over
there.
I'm
not
sure
we
have
the
right
people
to
do
a
demo
there
and
last
last
event
on
this
slide.
That
will
be
will
be
the
next
next
ietf.
A
So
in
I
forgot
again
what
it
is,
but
it's
madrid
I
forgot
to
put
on
the
slide,
and
here
there
will
be-
I
mean,
of
course
we
don't
have
program
there
yet,
but
the
most
notable
part
will
be
regarding
the
acathon
that
you
would
like
to
have
be
ready
for
for
this
ietf
in.
In
madrid
in
july,
so
meaning
that
we
we
have
to
to
prepare
a
bit
beforehand
as
you,
as
we
have
already
said
before.
A
So
I
think
we
have
to
be.
We
can
also
discuss
discuss
that
in
the
last
part
of
the
meeting
aob,
but
we'll
maybe
start
to
create
a
bit
discussion,
discussion
group
on
people
who
will
be
interested
to
join
the
easy
accountant
and
see
how
we
or
we
can
advance
on
this
point.
B
So
they
are
making
progresses
in
the
team
of
walter
plus
also
the
team
of
baba
martini
from
mcnid
kind
of
joining
their
their
activities
for
proof
of
concept,
development
or
tool,
development
related
to
ibm.
So
this
is
ongoing
and
they
have
plans
that
this
should
be
mature
enough
in
a
couple
of
months,
and
so
their
plans
is
to
be
able
to
participate
to
the
to
the
madrid
meeting
with
a
proposition.
B
For,
as
I
mean,
a
kind
of
proof
of
concept
and
demo
tools
that
could
be
a
subject
to
to
work
in
the
academ
and
in
the
discussion.
I
try
also
to
investigate
a
bit
around
that
within
nokia
if
we
can
have
also
the
work
from
fred
to
be
made
available,
but
this
is
still
being
discussed
and
I'm
also
discussing
with
the
team
of
lissandro
about
the
the
activity
they
have
on.
This
interface
is
to
to
derive
the
intents.
B
If
this
can
also
be
made
available
in
in
some
form,
this
is
being
also-
I
mean
it's
not
clearly
visible,
but
just
to
report
that
there
are
first
minor
progress
being
made
to
to
prepare
participation
in
the
academy.
We
still
have
to
consolidate
everything,
but
at
least
the
first
initiatives
are
being
launched.
A
Okay,
thank
you
lauren
for
reporting
this.
This
progress,
it's
good
to
know.
A
A
So
olga
you
are
you,
you
are
the
second
one
and
in
the
agenda
regarding
the
draft
and
intended
classification.
A
So,
as
I
said,
introduction
yeah,
you
started
to
a
nice
work
with
by
starting
to
process
the
different
comments
you
received
and
enter
interaction
with
the
different.
Let's
say:
comment
author
in
general,
with
the
people
on
the
mailing
list.
Would
you
like
to
make
now
kind
of
data?
Give
us
feedback
about
the
different
comments
you
address
and
the
feedback
you
receive
and
how
you
address
them
and
so
on.
A
C
C
So
during
the
initial
stage,
we
were
replying
on
the
comments
via
emails
directly,
still
hoping
that
the
draft
would
be
approved,
but
as
it
wasn't
approved
now,
we
kind
of
have
to
move
over
then
continue
kind
of
making
modifications
to
this
version
in
order
to
be
approved
going
forward.
So
we
created
a
summary
table
with
the
comments.
C
I
have
a
another
author
that
that
adriana
ornario,
that
would
also
be
added
to
the
draft
that
is
helping
as
well
from
huawei
and
created
the
summary
table
with
comments,
and
then
we
said,
let's
start
engaging
in
the
communication
with
individual
comment:
authors
as
you
suggested
lauren.
So
there
was
a
proposed
plan,
but
I'm
not
sure
whether
that
is
the
realistic
plan.
You
know
to
review,
discuss
internally
on
the
autism
mailing
list
first,
but
in
fact
we
moved
to
the
rg
mailing
list
overall,
because
that
was
the
recommendation.
C
We
are
trying
to
engage
with
individual
comment,
authors
and
chairs
for
each
of
the
items.
First,
we
are
trying
to
address
the
items
related
to
the
purpose,
scope
and
the
goal
of
this
document.
You
know
yourself
that
I
wasn't
here
from
the
beginning
of
this
document.
So
when
I
joined
as
an
author,
you
know
I
did
some
additions
to
this
document
at
that
stage
and
some
review
comments
etc.
C
C
So
the
hope
was
to
to
look
at
that
first
and
then
the
goal
was
to
kind
of
look
at
if
it
is
possible
to
go
with
the
to
try
to
make
up
is
before
march
9th.
Is
it
march
9th?
Is
the
cut
off
date,
or
is
it
maybe
a
little
bit
later
for
the
ietf
107
lauren?
Is
it
march
9th.
C
Yes,
so
if
you
can
make
it
to
march
9,
that
would
be
our
goal,
but
it
all
depends
on
on
the
amount
of
work,
and
so
this
is
the
kind
of
a
table
summary
table
that
we
created
and
we
tried
to
kind
of.
There
were
a
lot
of
communication
on
the
emails,
a
lot
of
different
comments,
so
we
just
tried
to
summarize
it
from
all
the
emails
that
were
received.
C
So
we
identified
really
nine
comments
as
such
the
next
page,
it's
nine
in
total
and
and
some
of
those
comments
have
been
sent
from
different
different
authors.
So
the
first
one
is
really
the
one
that
we're
initially
focusing
on
the
value
of
the
document.
It
looks
like
when
we
added
those
tables
that
there
were
a
lot
of
conflict.
You
know
confusion
being
added
by
adding
those
tables
we
thought
with
addressing.
C
I
thought
in
fact
that
I
was
addressing
the
review
comments
and
that
the
problem
with
the
document
was
that
there
was
a
lot
of
text
instead
of
making
it
some
easy
to
read
form,
and
then
I
thought
maybe
tables
could
be
as
as
lauren
suggest.
But
maybe
it's
not.
Maybe
it's
not
the
right
approach
again.
Maybe
we
can
revisit
that
and
remove
them
all
together.
C
So
the
call
the
really
what
what
we
want
to
agree
here
is
the
core
value
proposition
of
the
document
and
then
focus
on
those
comments.
First,
so
we
had
a
two
candidate
approaches
about
introduction.
Like
it
doesn't
matter.
We
there
is
an
introduction
now
proposed
in
the
email
list,
and
our
goal
would
be
to
agree
that
introduction
before
doing
the
comments.
C
So
jorgen
yesterday
did
say
that
you
know
he
doesn't
agree
with
having
those
tables
at
all
and
that
he
doesn't
have
to
approve
and
whatever,
but
I
would
still
like,
even
if
he,
if
he
want,
if
you
need
to
remove
them,
that
it's
if
it's
possible
to
engage
him
as
well,
at
least
for
the
purpose
and
see
if
there
is
something
we
can
change
to
address
his
comments
for
jerome
and
lauren
like
I
would
ask
you
to
look
at
the
introduction
shed
and
give
your
opinions
if
these
tables
are
so
confusing
and
the
goal
of
these
tables
is
to
show
maybe
how
we
can
present
all
those
different
abstractions
and
try
to
end
the
inventions
and
try
to
maybe
use
them
going
forward
for
different
parks.
C
And
but
if
that
is
confusing,
maybe
we
could
focus
on
methodology
and
just
have
those
tables
on
the
website
and
remove
them
from
the
document
or
or
fill
them
in,
based
on
what
we
believe
is
inside
our
intense
scope
for
nmrg
or,
what's
not
out,
so
we
have
two
options
there.
Maybe
so
I
would
like
to
here,
like
jurgen,
would
probably
like
to
remove
them
all
together
and
then
change
the
focus,
and
I
would
like
from
lauren
and
jerome
to
hear
what
you
guys
are
thinking
and
to
address
this
comment.
First.
C
There
was,
as
you
know,
a
lot
of
discussions
in
regards
to
that
yesterday
and
there
are
differences
in
in
opinions.
What
is
the
intent
definition?
So
I
did
not.
I
do
not
want
to
to
make
it
as
a
task
for
this
draft
to
define
the
intent.
I
think
maybe
it
should
be
inside
the
other
draft
and
we
should
all
review
it
and
agree.
C
C
One
of
the
things
like
branislav
did
say
that
you
can't
really
define
intent
until
you
have
your
knowledge
base.
You
know.
So
that
is
a
valid
things
to
say.
I
have
my
knowledge
base
and
only
when
I
have
all
the
information,
then
I
can
determine
what
my
for
specific
domain
or
solution
my
intent
is,
but
in
that
case
it's
only
possible
to
define
a
methodology
for
identification
of
intents.
So
at
the
end
you
wouldn't
even
be
able
to
have
any
formal
definition
or
anything,
because
so
I'm
I'm
thinking
we
can
discuss.
C
I
still
believe
intend
definition
is
very
important
and
classification
is
very
important
and
then
we
can
look
at
formal
definitions
afterwards,
but
please,
let
me
know
the
third
comment.
Jurgen
was
saying:
is
it
a
survey
so
he's
just
said:
it's
not
a
survey,
so
the
fourth
comment
again
from
jurgen.
C
It
was
about
the
table,
so
we
thought
if
we
introduced
introduction
and
if
we
introduce
methodology,
which
is
the
next
thing
we
would
like
to
do,
then
tables
could
be
used
in
the
context
of
those
two,
but
we
can
potentially,
if
majority
believes
that
they
are
not
useful,
we
can
remove
them,
they
were
not
there
in.
In
the
previous
version,
they
were
added
the
result
of
the
comments
that
we
had
at
the
meeting.
D
C
Time
point
five
unclear
how
the
categories
in
the
tables
have
been
identified
selected
and
the
document
is
missing,
a
more
rigorous
approach,
methodology
to
the
classification
work,
so
we
are
proposing
here
to
add
a
methodology
section
in
section
five
that
would
propose
the
process
and
that
would
also
include
the
process
of
how
we
initially
came
with
this
draft
table,
but
also
tables
and
also
how
we
we
will
extend
it
in
a
way
going
forward
by
either
adding
or
removing
different
things
from
the
from
the
scope.
C
The
issue
with
the
issue
with
fully
knowledge
based
driven
would
be
if
I
have
to
go
to
the
domain,
and
I
have
to
go
and
get
knowledge
base
for
that
domain.
The
problem
for
me
would
be
that
the
knowledge
is
so
big
that
so
so
our
tables
already
kind
of
produce
outcome
of
that
process.
Where
you
looked
at
several
solutions,
you
know
the
knowledge
you
you
have
experience,
you
worked
on
it
for
decades,
and
now
you
are
saying
this
is
what
we
believe
the
intent
should
be
like
references
in
terms
of
references.
C
How
we
are
proposing
is
that
I
don't
know
like
there
are
huge
amount
of
references
that
are
solutions
from
different
customers
and
different.
You
know
china
telecom.
There
gave
their
input
what
we
gave
our
input,
like
the
solutions,
would
be
common
between
different
vendors,
different
carriers.
So
you
know
it's
not
something
that
we
reinvented
here,
but
I
do
believe
that
methodology
section
would
help.
But
I
don't
know
if
that
is
sufficient,
or
you
know
would
like
your
opinion
on
on
that.
C
But
after
we
review
introduction,
we
would
share
some
ideas
about
what
we
can
do.
Maybe
in
that
methodology
section
the
next
comment.
It
was
about
policy
continuum
and-
and
you
know
that
it
fits
here
or
whether
it
should
go
in
in
some
other
document.
So
we
have
two
candidate
options.
C
We
can
either
remove
it
from
classification
document
and
use
it
in
other
documents
or
or
we
can
expand
the
classification
document
and
add
better
connection
with
the
the
policy
continuum,
maybe
at
this
point
and
that
one
would
be
probably
the
better
one
because
of
the
time.
So
if
that,
if
natal
bob,
would
the
next
comment?
Number
seven
is
so
please,
if
you
have
any
comments
or
if
you
have
any
opinions,
stop
me
sorry.
C
C
So
the
next
comment,
seven
is
about
definitions
in
the
draft.
Do
not
bring
clarity
in
terms
of
differences
between
intent
and
policy.
So
there
is
a
confusion
that
we
have
some
sections,
because
there
are
multiple
authors.
There
are
some
sections,
four,
four,
five,
four
six
that
are
using
policy
terminology
instead
of
intent
terminology.
C
So
we
want
to
address
that,
and
but
we
also
want
to
refer
to
the
claim
which
describes
differences
between
intents
and
policies,
but
we
saw
even
yesterday
during
the
discussion
that
it
is
a
it's
a
big
topic
and
there
are
people
who
have
different
opinions,
and
you
know
I
personally
think
what
we
should
try
to
do
is
try
to
agree
some
terms
for
the
nmrg
and
some
agreements
for
nmrg
and
then
try
to
kind
of
bring
the
highest
value
through
the
you
know,
the
methodologies
or
through
the
classifications
through
the
the
architectures
through
the
different
dsls
going
forward,
and
I
do
believe
this
topic-
I've
been
involved
with
different
stakeholders
about
this
topic
now
for
years
same
as
with
the
topics
of
service
resource
customers
facing
resource
facing
services,
it's
very
difficult
to
always
get
consensus
about
all.
C
So
if
there
is
a
way
that
we
can
get
consensus
in
nmrg,
that
would
be
the
beneficial
and
you
know,
even
if
the
declarative
policy
is
something
for
somebody,
but
something
this
else
for
the
other
person
or
for
other
stakeholders.
I
do
believe
it's
more
important
to
find
a
common
ground
and
come
up
with
some
definition
that
we
can
agree
on
then
whether
something
is
perfect
or
not,
but
that's
my.
D
C
Is
kind
of
quite
negative,
I
did
ask
him
if
he
would
be
interested
to
contribute
because
I
thought
his
idea
of
knowledge
base.
What
was
interesting,
and
you
know
that
maybe
if
he
can
contribute
to
methodology,
maybe
we
can
get
some
value,
but
he
unfortunately
he
he
says
it
would
take
too
much
time
so
he
wasn't
able
to.
C
But
my
understanding
of
this
comment
is
that
he
thinks
that
you
cannot
identify
define
intent
until
you
have
knowledge
base
for
the
specific
domain
where
you
can
then
obstruct
your
intent
from
it.
But
it
means
that
you
can
only
have
a
process
of
identification
and
but
what
would
your
guys
are?
And
what
would
your
view
be
on
this
comment
about
knowledge
base,
driven
obstruction.
B
Yeah,
I
think
it's
not
easy
to
handle
the
way.
Branislav
is
communicating,
but
it's
some
some
input,
some
some
data
points
yeah.
So
for
me,
I
try
to
reply
a
bit
to
that
in
one
of
the
comments,
because
you
you
pointed
out
that
there
are
some
inconsistency
in
in
the
definition
the
current
definition
we
have
in
the
draft
the
claim
draft,
because
for
me
the
it's
a
challenging
task
because
going
and
ending
with
one
intent,
comprehensive
intent,
definition,
I'm
not
even
sure
it's
feasible
or
even
it's
advisable
to
have
that.
B
But
what
we
could
have
is
a
definition
or
definition
in
plural,
that
captures
what
are
the
essential
characteristics
of
different
forms
of
intent
and
the
way
I
will
join
to
the
comment.
Eight
is
that
for
sure
I
mean,
if
we
generalize
too
much,
then
we
will
have
like
a
dictionary
definition
of
what
an
intent
is,
but
this
is
not
applicable
to
the
an
emergency
scope
or
to
any
scope.
So
it's
too
generic.
B
A
bit
like
what
the
discussion
we
can
have
with
john
straster
about
these
domain
specific
languages.
But
I
think
this
is
for
me
a
bit
also
the.
How
can
I
say
it's
the
resulting
it's
it's.
What
is
at
the
end
of
the
discussion?
B
Okay,
when
you
start
talking
about
dsl,
you
already
made
all
the
processing
before
that
to
come
to
the
conclusion.
What
is
the
dsl?
I
need
to
design
or
use
to
apply
internet
based
networking
in
this
domain,
but
I
think
at
the
stage
where
we
are
in
energy
is
try
to
clarify
everything
that
comes
before
that
point,
and
I
think
this
is
why
we
have
the
work
in
the
draft,
on
definition
and
the
draft
and
classification
to
try
to
identify
what
are
the
the
key
elements:
the
key
characteristics,
information,
the
intent
types.
B
What
is
common
between
the
different
types?
What
is
specialized,
how
you
need
to
use
things?
Are
you
if
you
have
different
parts
in
the
definition
you
have
the
user,
you
have
the
target
system,
you
have
the
transformation,
so
different
elements
of
the
definition,
how
you
can
use
those
parts
and
what
you
need
to
do
on
those
parts,
how
you
specialize
them,
how
you
immediately
some
parts
you
need
to
validate
some
parts.
You
need
to
provide
information
about.
So
for
me
this
is
you
see
this
first
phase
before
we?
B
We
are
at
the
point
where
we
can
say
we
will
use
this
language
or
we
will.
This
is
the
form
of
intent
we
will
use.
We
are
not
yet.
At
that
point,
we
need
to
do
all
the
processing,
for
that.
This
is
the
work
of
draft
definition
and
draft
classification,
and
in
your
draft
you,
I
think
what
you
did
also.
Is
you
looked
at
the
different
intent
solutions,
so
you
already
try.
B
Also
when
I
said
in
my
comments
when
I
said
you
you
did
bottom
up,
is
that
you
looked
at
some
application
areas,
and
you
say
if
it
is
a
enterprise
network,
my
user
is
this
and
my
intent
is
in
this
form,
and
the
typical
target
system
is
this.
So
this
is
domain
specific
and
you
try
to
extract
out
of
that
key
elements,
and
you
put
that
into
a
instance
of
often
intent
types,
and
I
think
we
need
to
do
this
kind
of
process.
C
Because
from
I
agree
completely
thanks,
because
I
do
believe
that
was
my
understanding
that
we
are
starting
with
these
two
drafts
and
then
you
will
end
up
with
one
or
many
dsls.
You
know
you
may
end
up
with
yourselves
per
solution
per
user.
Even
you
know
it's
quite
possible
that
you
will
have
to
end
up,
but
we
I
thought
it
was
outside
of
the
scope.
Another
comment
I
have
about
this
is
that
you
see
if
you
are
starting
from
knowledge
based
knowledge
bases.
We
know,
for
our
networks
are
huge
huge.
C
So
what
we
did
is
kind
of.
We
proposed
subset
already
obstruction
of
that
knowledge
base
that
we
believe
is
relevant
for
the
intent
space.
But
if
you
start
from
the
beginning
from
from
thousands
of
attributes
and
thousands
of
properties
in
each
of
those
solutions
it,
it
would
be
much
bigger
exercise
to
do
so.
C
Yeah,
okay
and
the
next
was
jerome
saying
that
maybe
some
examples
would
be
useful.
So
it's
that
that's
a
very
valuable
comment,
and
there
are
some
examples
already
in
some
of
the
previous
sections
in
the
general
text,
but
I
do
agree
that
it
would
be
if
we
are
keeping
the
tables.
If
we
agree
that
we
can
keep
tables
even
in
annex
or
something
then
we
would
add.
A
Okay,
just
as
a
speaking
as
a
person,
so
as
I
I
told
you
by
email,
I
just
just
returning
from
holiday,
so
I
didn't
have
read
all
the
the
email
thread
still
so.
Regarding
the
the
first
comment,
so
I
will
have
a
detailed
look
at
what
has
been
discussed
on
mailing
list
and
provide,
if
needed,
more
feedback.
Regarding
this
first
comment
by
email,
I
think
I
will
do
my
best
to
give
you
by
by
mid
next
week.
C
And
and
just
one
small
comment
like
in
terms
of
the
intent
definition,
you
know
I
I
do
understand
it's
difficult
to
have
a
specific
definition,
but
there
is
a
lot
of
confusion
and
I'm
getting
always
what
is
intent.
What
is
intent
and
I've
seen
from
so
many
different
stakeholders
that
they
do
have
different
completely
different
views
of
what
intent
is.
So
I
still
believe,
even
if
it's
a
very
abstract
definition,
you
know,
even
if
it
is
definition
that
we
agree,
that
it
is
a
declarative
policy.
C
You
know
the
current
definition,
I'm
not
sure
about
the
rules,
one
that
you
have,
but
the
other
one
if
we
agree
as
an
nmrg.
This
is
our
definition
and
still
it's
still
more
valuable
than
if
it
is
a
perfect
one.
You
know
whether
we
said
objective
or
whether
we
said
declarative
policy
of
it.
We
said
at
least
we
have
it.
You
can
send
it
to
someone
and
say
this
is
our
starting
point.
The
starting
point
is
that
we
are
doing
it
in
the
way
that
we
are
saying
we
are
looking
at
the
user.
C
We
are
looking
at
what
what
user
wants
from
the
network
and
not
how
network
delivers
it,
and
it
can
still
be
a
definition
that
is
nmrg
definition
and
then
later
on.
We
can
go
and
do
more
specific
definitions
in
dsls,
but
still
believe
it
is
important
to
have
van
learned.
Otherwise,
you
know
we
won't
be
able
to
answer
the
question
to
the
audience.
What
do
you
guys
think
intent
is?
B
I
agree
olga
this
is,
for
me,
I
mean,
even
if
the
draft
has
been
recently
adopted
as
a
research
group
document.
For
me
this
is
again
there
are
some
content,
but
we
still
have
clear
work
to
do
on.
I
mean
the
intent
definition
work,
but
also
clarifying
the
different
underlying
concepts.
So
for
me
this
is,
we
are
not
at
the
end
of
the
story.
It
doesn't
need
the
one-year
work,
but
I
mean
this
is
for
me
this
is
not
a
down
deal.
B
And
just
comment
also
back
to
to
your
draft
and
classification
and
the
relationship
with
the
draft.
On
definition,
we
had
discussion
in
the
early
stage
of
the
documents
that
we
wanted
to
keep
the
two
documents,
even
if
we
know
they
are
related
because
we
wanted
to
foster
the
the
content
development.
We
didn't
want
to
say
how
this
should
be
structured
first,
but
first
put
the
content
and
then
see
if
we
need
to
split
merges.
B
We
are
still
at
this
stage,
but
I
think
there
are
a
strong
relationship
between
the
what
what
the
two
drafts
I
mean.
All
the
two
drafts
gets
developed
in
the
sense
that,
as
you
said,
you
have
dependency
to
intend
definition
from
draft
claim
in
your
document.
B
But
I
think
what
the
what
you
with
the
analysis
you
can
do
in
the
draft
classification,
will
also
be
useful
as
input
back
to
the
to
the
definition,
because.
B
Bit?
Yes,
so
that's
why
again,
it's
not
it's
not
simplifying
the
the
approach,
but
I
think
we
have
to
to
see
it
iteratively
and,
as
you
say,
it's
circular
so
to
work
together
in
trying
to
to
bring
what
is
good
in
both
to
and
to
align
our
our
text,
our
understanding.
B
So
for
me
this
is
part
of
the
the
work
in
progress.
So
I
see
a
strong
relationship
at
this
level
between
the
two
drafts
and
on
the
classification
work,
because
you
had
some
earlier
part
of
the
comments
on
the
methodology
for
the
classification
and
whether
we
should
keep
the
tables
or
how
to
use
that
for.
D
B
B
B
So
it's
just
to
document
a
bit
the
approaches,
so
it
doesn't,
it
can
be
very
formal
or
it
can
be
just
a
good
explanation,
but
I
think
this
will
be
useful
for
an
external
reader
to
know
how
we
ended
up
with
this
proposal
and
then
the
the
use
of
the
methodology
for
me
is
also
this
document
when
it
will
be
finalized
to
be
to
be
consistent
to
show
that
there
is
an
approach
to
derive
the
different
categories,
but
I
think
it's
it
will
be
also
useful
after
the
document
is
finalized,
because
for
me
the
the
purpose
of
the
classification
is
also
that,
if
in
two
months
or
if
someone
else
comes
with
another
application
area
or
another
smart
idea
about
how
to
use
intent,
I
want
him
to
or
him
or
her
to
be
available
to
be
able
to
insert
their
novelty
in
the
the
classification,
because
it
relies
on
some
known
criteria.
B
You
see
it's
like.
If
you
add
a
new
species,
you
discover
new
species,
you
don't
want
to
re-architect
the
entire
living
taxonomy,
but
then
you
can
see
if
this
is
a
branch
is
from.
I
don't
know,
then.
Okay,
you
understand
a
bit.
What
I
say
is
because
this
for
me
will
create
a
long-lasting
value
to
the
document.
C
Something
like
talk
of
type
or
something
like
that
where
we
would
just
propose
a
process
of,
we
will
see
whether
it's
one
thing
or
two
things.
You
know
whether
extension
and
creation
should
be
one
process,
but
hopefully
it
should
be
one
process.
So
I
I
do
agree.
It
would
be
valuable
and
my
hope
is
just
that
by
adding
this
section,
which
may
be
one
or
two
pages.
C
C
So
so
I
think
introduction
that
we
added
it
does
talk
about
the
document
being
used
in
the
methodology
to
explain
how
we
came
up
with
this
table
these
categories
and
tables
or
types,
because
you
know
we
have
some
solutions.
Absolutely
it
would
be
other
solutions,
you
know,
so
you
can
spend
two
two
years
now
discussing.
Oh,
is
it
all
only
these,
or
do
we
have
more
even
better?
C
These
three
are
the
right
ones
shouldn't
we
have
five
instead
of
three,
and
I
don't
believe
this
is
the
purpose
you
know
so
so
this
is
just
a
descriptive
way
of
saying
these
are
the
solutions
and
actors
that
are
there
at
the
moment
in
the
industry,
and
you
know
you
can
add
new
solutions,
you
can
you
add.
New
applications
in
fact
was
our
application.
We
only
had
the
one
like
controller,
but
you
can
add
new
dimensions.
C
You
know
new
intent
types
going
forward
or
in
your
solutions
you
can
remove
some
because
they
would
not
be
relevant
for
you.
Therefore,
you
will
end
up
with
completely
different
intent
knowledge
based
on
on
your
individual
product
solutions
or
domains,
and
I
do
agree
that
we
can
try
to
propose
some
approach
that
could
be
reused
both
for
what
explain
how
we
came
up
with
the
initial
draft
and
how
this
can
be
extended
going
forward
for
different
applications.
C
So
I
do
believe
that
that
is
connected,
but
my
only
worries
would
be
the
the
amount
of
notifications
to
the
overall
draft.
C
You
know
I
would
prefer
that
we
can
communicate
that
through
addition
of
the
section
in
clarified
and
good
introduction,
and
that
we
don't
have
to
do
complete
too
many
changes
in
the
draft
because,
as
you
said,
these
are
the
initial
versions,
we
will
improve
them
through
time.
So
I
do
believe
at
this
point.
It
would
be
more
important
to
have
two
approved
that
people
would
be
able
to
refer
to
and
then
work
on
the
new
versions
and
or
even
emerges,
if
agreed
so.
C
Thanks
so
please
everyone
send
comments
and
engage
in
the
mailing
list.
Any
comments
welcome
and
also
if
anyone
wants
to
contribute
or.
B
Just
a
question:
this
is
not
mandatory
at
all.
As
you
may
be
aware,
we
have
now
a
github
page
for
the
energy.
We
have
a
policies
on
the
accepted
research
group
document.
We
have
an
official
repository
for
the
draft
that
is
public
and
available.
We
also
offer
the
possibility
for
the
individual
draft
related
to
the
work
of
energy
to
have
a
repository
there.
B
So
my
question
is,
since
we
are
also
entering
a
bit
more
active,
I
will
say
evolution
of
the
document
with
comments
and
your
proposal
on
how
to
to
evolve
this.
Will
you
be
willing
to
have
a
repository
on
the
on
the
github?
Is
it
something,
of
course
you
can
discuss
with
your
computer?
So
that's
great.
C
A
C
C
F
Everyone,
this
is,
will
speaking,
I'm
so
sorry
that
I
just
finished
the
previous
meeting,
and
I
got
a
message
from
china
telecom,
the
author
of
actually
the
first
officer.
He
said
he
can.
He
said
that
he
doesn't
know
what
happened,
but
he
cannot
join
this
meeting.
F
A
So
is
there
any
other
comment
now
regarding
regarding
this
draft.
B
Updates,
yes,
so
I
don't
have
any
supporting
slides,
because
this
would
be
a
very
quick
update.
B
We
are
working
with
the
other
authors
to
address
the
comments
that
were
raised
during
the
adoption
process
and
also
to
address
the
plan
we
advertise
to
make
the
evolution
of
the
document,
so
this
is
essentially
to
there
is
some
updates
on
the
life
cycle
diagram
some
editing
through
the
text
that
is
being
ongoing.
B
We
have
also
the
aim
to
add
a
security
consideration,
so
this
is
the
main
active
discussion
we
have
with
the
co-authors.
This
is
not
yet
public
because
we
try
to
stabilize
a
bit
the
update,
but
we
will
generate
a
new
version
very
soon
and
also
more
recently
through
the
discussion
with
the
internet
classification
draft.
B
Thanks
to
that,
we,
it
was
spotted
some
inconsistency
in
the
draft
based
on
the
different
definitions
terms
that
we
have
in
different
places
of
the
draft
on
definition
and
concept.
So
this
is
a
new
new
issue.
We
have
to
fix,
probably
with
the
next
release
of
the
draft.
So
I
think
in
the
next
two
before
the
deadline,
we
hope
to
be
able
to
provide
a
good
new
update
of
the
draft
in
the
different
sections
to
to
go
towards
the
plan
we
we
have
and
we
are
keep.
B
Of
course
we
keep
open
to
reviews
and
receiving
additional
comments
how
to
evolve
this
document.
This
is
a
research
group
document,
so
we
are
really
here
to
to
make
the
necessary
work
to
bring
it
to
a
very
good,
stable
document.
G
G
So,
thank
you
for
the
invitation.
It's
always
a.
G
A
way
to
to
discuss
our
current
work,
mainly
on
on
network
management
and
in
the
context
of
network
slicing
and
automation
of
management
and
the
interest
that
we
have
in
in
the
group
is
to
foster
our
comprehension
and
going
beyond
a
comprehension
to
integrate
ibn,
maybe
a
new
wave
in
ibm
in
network
management
and
more
in
network
exposure,
and
the
objective
of
this
presentation
is
first
to
share
with
you
our
work
on.
G
As
I
said,
slice
management
with
onap
and
it's
issued
from
two
research
papers
published
lightly.
One
was
presented
in
network
of
the
future
conference,
late,
2019
and
another
one
that
was
presented
actually
this
week
in
ithin
2020.
G
So
the
okay.
I
move
in
I'm
playing
three
three
sections
of
the
presentation,
so
the
first
one
will
focus
on
the
network,
slice,
lifecycle,
management,
automation
and
our
main
tool.
Main
focus
actually
is
on
app
open
network
automation,
platform,
the
second
part
and
actually
it's
linked.
G
But
I
I
wanted
to
highlight
the
the
sla
focus
if
they
related
to
running
slices
and
monitoring
to
ensure
sla
compliance,
and
the
last
section
is
quite
open
with
challenges
for
intent,
based
what
I
call
what
I
want
to
call
intent
based
network
exposure,
because
we
find
the
literature
more
things
on
internet
based
networking
and
interfaces
inside
the
network.
But
what
we
target
is
interfaces
towards
the
customer
final
customer.
G
We
also
talk
about
instantiation.
All
this
have
different
definitions,
but
what
we
want
to
to
have
in
mind
is
the
a
double
or
triple
process
of
deployment
and
then,
once
the
all,
the
software
breaks
are
on
a
virtualized
on
in
an
environment
that
is
capable
of
executing
them
with
the
right
resources,
then
the
slice
can
be
activated
and
then
during
the
runtime
we,
this
is
classical
actually,
but
we
activate
the
the
monitoring
during
runtime.
G
So
what
we
focus
on
actually
is
the
deployment.
As
I
said
it's
in
it
includes
three
three
steps
and
and
the
design
so
the
design
and
the
deployment-
and
this
slide
shows
the
the
workflow
the
main
steps
from
the
design
to
the
slice
deployment.
G
So
the
first
step
is
fully
software,
with
vnf
images
described
in
heat
files,
the
packaging
of
the
vns
and
then
the
design
of
a
composition
of
them.
We
assume
this
is
actually
very
important.
We
we
are
behind
the
customer.
We
do
not
have
a
way
to
give
have
a
real
tool
to
make
us
make
this
composition.
We
really
do
our
compositions,
assuming
that
we
have
a
request,
and
this
is
our
big
gap.
G
Actually
so
we
create
our
service,
there
is
a
software
validation,
we
have
the
requirements
for
each
vns
and
then
for
the
set
of
vns
for
the
network
of
the
global
network
service.
So
there
is
a
software
validation
for
okay
having
all
each
component
have
its
interfaces
and
all
components
are
interoperable
and
the
global
service
is
a
meaningful
service.
G
Then
there
is
the
cloud
os
preparation
where
we
we
create
the
the
virtualized
hosting
environment.
G
We
mainly
use
openstack
that
we
also
have
in
our
labs
now
kubernetes
for
for
more
agile
deployments
after
the
cloud
os
preparation,
we
we
start
the
slice
onboarding,
so
it's
the
the
set
of
dns
the
global
service
with
creation
and
distribution
of
services
over
the
infrastructure,
with
with
an
app
and
the
actual
action
of
putting
the
services
using
on
app
in
in
the
openstack
environment,
is
what
we
call
the
the
final
step
of
the
deployment.
G
So
these
steps
are
inside
an
architecture
that
we
well.
We
are
using
we're
considering.
So
we
assume
there
is
a
network
orchestrator
for
us.
It's
an
app
a
network,
orchestrator
orchestrator.
G
With
a
companion
with
a
network
slice
manager
and
this
network
slice
manager
can
be
the
operator,
but
they
can
be
also
another
actor,
and
this
network
slice
manager
is
responsible
of
getting
the
the
request
from
a
communication
service
or
actually
any
any
any
service.
Any
applicative
service,
a
high
level
service
towards
the
network
factory
network
industry,
guided
by
honest.
So
the
the
global
scheme
is
that
okay,
we
have
a
request.
G
So
then
the
the
monitoring,
and
so
the
the
idea
is
how
to
maintain.
Since
there
is
a
request
and
it's
for
a
dedicated
service
that
we
are
deploying
such
as
life,
we
need
to
maintain
and
negotiate
a
previously
negotiated
sla
and
have
the
enforcement
of
these
slas
with
policy.
G
And
here
also,
I
have
something
to
emphasize
that
from
what
I've
read
from
intense
days
is
that
policies
we
need
to
focus
on
a
difference
between
what
we
call
policies
and
what
we
call
intents
and
and
today
we
are
in
our
work
at
least
we
are
still
talking
about
policies
and
using
policy.
G
So
it's
not
it
can.
These
policies
are
can
be
even
based
by
they
are
not
really
what
we
call
demand
based
and
that's
what
we
put
behind
from
our
comprehension
what
we
put
behind
the
the
intent.
G
So
it's
not
either
device
specific.
Yes,.
B
Please
finish:
I
have
a
more
question
on
the
on
the
slide,
but
you
may
finish
your
explanation.
First,
I
thought
you.
G
G
So
we
have
so
the
tools
with
onap,
further
monitoring
for
the
collecting
the
some
parameters,
mainly
on
the
resource
resource
conception
and
we're
trying
to
put
some
analytics
for
the
evaluation
of
the
reports
coming
from
the
monitoring
do.
B
More
on
the
first
part,
because
the
first
paragraph
you
mentioned
that
there
is
an
external
entity
which
interacts
with
the
various
orchestration
platforms
for
the
end-to-end
network,
slides
just
for
my
clarification.
Do
you
consider?
Because
if
you
consider
end-to-end
network
slice,
usually
you
have
different
networks,
domains
or
network
segments.
Essentially
the
access
exactly.
B
G
So
yeah,
okay,
we
we
also
focus
on
on
up
as
an
orchestrator
of
of
a
multi-domain
network
slide.
Another
slide
deployed
over
multiple
domains,
that
is,
for
us,
feasible,
but
based
on
the
choices
that
the
network's
life
manager
can
make,
he
may
choose
another
domain,
which
the
orchestrator
is
something
else,
and
I
can.
I
can
make
a
reference
to
a
5g
eve
project
which
focus
on
this
domains:
interoperability
with
a
different
orchestrators
and,
for
example,
partners.
I
can,
I
can
speak
of
orange.
G
We
have
the
the
labs.
The
platforms
by
orange
are
using
on
app
and
other
partners
using
osm,
and
the
idea
is
to
make
that
a
global
orchestrator.
There
is
a
way
to
to
to
deploy
the
slices
on
these
different
domains,
even
if
each
domain
has
its
own
orchestrator.
G
G
So
we
I
mean
well
well,
I
was
going
to
say
I've
already
presented
some
ontologies
that
we
have
included
actually
in
the
definition
of
of
network
slices,
and
here
it's
it's
a
big
picture
of
design,
time
and
runtime
phases.
What
we
call
on-app
compliance.
Actually
we
defined
a
set,
an
anthology
for
web
autology
for
network
slices
compatible
with
what
one
app
is
capable
of
supporting.
G
So
there
are
already
modules
defined
by
onap
for
the
design
time,
so
the
the
on
the
right
and
it's
the
it's
linked
directly
to
the
modeling.
At
this
phase
we
are
modeling
the
slice
template,
if
you
can
call
it
this
way,
but
the
this
modeling
considers
so
all
the
elements
from
the
from
the
customer
and
from
the
provider.
G
So
from
from
the
the
the
customer
side
is
the
mainly
the
the
request,
a
request
for
the
network,
slides
or
a
request
for
network
service
actually
and
the
dsla
and
the
provider
is
the
one
who
will
define.
Then
the
policies
find
the
right
network
functions.
G
Actually,
the
customer,
depending
on
it
on
his
profile,
would
say
more
or
less
things
on
on
the
network
service.
He
can
go
into
details
of
network
functions
if
his
profile
allows
him
to
do
so,
or
just
ask
for
network
global
network
service
to
support
his
application
service.
So
the
provider
here
has
more
entities
involved
in
the
modeling,
mainly
everything
related
to
a
policy
policy
definition
and
his
well.
G
We
assume
also
that
he
is
the
one
having
owning
the
the
vns
and
and
pns
the
descriptors
and
then
at
the
runtime
after
this
design,
well
composition.
So
there's
the
create
after
the
modeling,
with
the
the
the
the
cool
animation
here
is
the
so
the
creation
of
of
the
model,
the
and
then
the
deployment
and
it's
as
it's
a
cycle.
Well,
it's
it's!
It's
supposed
to
be
a
loop,
as
I
saw
as
I
showed
on
on
one
of
the
first
slides.
G
We
are
supposed
to
take
into
consideration
any
changes
either
that
comes
from
the
monitoring,
so
like
alerts
or
things
that
to
which
we
need
to
reconfigure,
for
example,
or
modify
or
add
for
scaling
or,
for
example,
if
less
resources,
if
the
the
sla
towards
the
infrastructure
is
not,
is
not
respected,
then
these
are
things
alert
that
comes
from
the
monitoring,
but
we
should
be
also
able
to
take
into
account
new
well
changes
in
the
service
itself.
G
For
example,
if
the
customer
is
capable
of
asking
for
a
new
algorithm
for
authentication,
for
example,
or
any
other
network
function
more
specific,
then
we
should
be
able
to
add
it
into
the
chain
into
the
slides
that
it
already
deployed
and
reconfigure
if
necessary,
and
also
change
the
policy
and
enforce
it
further
for
for
the
global
service
for
the
network,
slides
if,
if
necessary
as
well,
our
there
is,
you
can
see
here
a
component
that
is
very
important,
which
is
a
clamp
which
is
control,
loop,
automation
for
management,
sorry
control,
loop,
automation.
G
Oh
sorry,
I
forgot
the
ali
amp.
What
was
the
mp4,
but
it's
the
monitoring?
Sorry
so
it's
the
the
the
the
the
loop
for
the
control
and
and
here
and
this
work
is
very
important.
It's
ongoing.
G
There
are
many
gaps
that
have
we
haven't
identified
in
onap
to
ensure
that
we
can
do
all
all
the
things
that
we
are
saying
here
to
integrate
both
both
changes
and
and
and
be
really.
Actually
it's
like
a
a
map
loop.
So
it's
monitoring
analysis,
planning
and
execution.
G
It's
exactly
this
and
the
the
knowledge
base
is
the
active
and
available
inventory
on
the
right
side
on
the
top
right
side,
and
but
this
inventory
is
not
enough
alone,
so
we
also
added
databases
for
the
network
services,
so
we
have
a
database
for
the
resources
and
database
for
the
network
services.
G
And
here
so
to
illustrate
to
illustrate-
and
you
can
find
it
in
also
in
the
paper
references
we-
we
have
considered
a
mobile
mobile
coin
network
provided
by
beacon,
if
you
know
become,
which
is
an
institute
in
france
and
the
the
components
of
the
vns
have
been
deployed.
Based
on
the
models
we
have
described.
G
The
descriptors
also
have
been
modified
to
cope
with
our
ontology
proposed
in
in
previous
work
and
the
deployment
of
all
these
components
with
a
separate
data
and
control
planes,
which
is
also
included
in
in
the
developments
of
the
vns
themselves,
which
is
a
huge
work
actually
in
the
design.
We
can
call
it
an
offline
design
phase,
but
it's
not
based
only
on
the
composition,
but
it's
the
the
functional
decomposition.
G
When
we
go
from
the
3gbp
functional
descriptions
towards
more
modular
functions,
this
work
is
is
really
a
huge
work.
What
we
are
doing
on
on
5g
is
to
look
at
the
specifications
not
going
from
the
code
itself,
but
going
from
the
theoretical
description
from
specifications
and
developing
the
components
to
have
to
go
with
5g
modular
functions,
to
cope
really
with
the
challenges
that
we
want
to
express
here.
So
it
was
more
to
illustrate
and
also
to
to
to
make
sure
our
lab
on
the
on
the
automation
of
network
slides.
G
So
these
are
the
example
of
network
lines
that
we
we're
showing
often
a
private
mobile
coin
network.
We
also
have
the
access
actually
from
opera
air
interface,
but
what
we
also
well,
what
we
wanted
to
include
is
this
definition
of
of
a
customer
facing
service
and
resource
facing
service,
so
the
the
network,
slice
manager
have
two
ways
of
expressing,
and
we
believe
that
in
this
it's
in
this
layer
that
our
internal
exposure
should
be
should
be
should
be
developed
and
implemented.
G
So,
regarding
the
the
sli,
we
again
we'll
say
it's
an
sla
regarding
the
infrastructure,
so
the
the
automated
loops
that
we
have
allow
us
to
check
the
the
the
needs
of
our
service
as
defined
by
the
the
designer,
so
here
the
provider
of
the
network
service.
So
we
know
how
much
resources
all
type
of
resources
we
need
for
the
global
service
based
on
each
vns
on
the
links
as
well,
and
we
are
clients
of
the
infrastructure.
G
G
We
assume
that
an
application
service
is
asking
for
this
bandwidth
for
this
speed
rate
for
this
we're
giving
this
availability,
for
example,
we're
giving
the
the
the
requirements
actually,
but
we
express
them
in
the
way.
We
know
we
are
experts
of
the
network
and
we
know
how
to
express
them.
The
difficulty
is
to
to
to
find
a
way
to
give
the
the
the
user
a
way
to
express
his
his
need
without
knowing
any
anything
of
the
network
of
of
or
or
even
network
services.
G
So
here
the
work
we've
done
on
a
policy-based
sla
compliance
is
made
regarding
some
main
scalability
for
the
mobility
management.
Here,
for
example,
is
regarding
the
infrastructure
and
if
we're
owning
the
infrastructure-
and
that's
it's-
it's
quite
easy,
but
if
not,
then
there
should
be
also
here,
an
intent
based
layer
that
would
allow
network
service
to
ask
for
the
right
resources
in
a
very
abstracted
way
from
the
infrastructure.
G
But
our
focus,
as
I
said,
is,
is
the
client.
What
can
I
say?
Here's
a
this
is
a
detailed
picture
of
including
the
the
the
use
case
that
we
have
considered
so
the
the
4g
core
mobile
network
and
with
all
with
all
the
descriptors
and
the
the
stakeholders.
So
the
customer
and
the
provider-
and
this
is
just
the
the
the
description
of
the
whole
slide
and
all
this
will
be-
is
configured
for
the
for
the
for
the
deployment
and
mainly
for
the
monitoring.
G
Vfx,
I
don't
know
if
it's
very
interesting
to
present
this
slide,
but
this
this
is
our
a
picture
of
the
the
lab,
the
open
lab
we
have
in
in
orange,
using
optostack
just.
C
G
They
are
actually
they
are.
If
there
is
only
one
here,
we
we
it's,
I
can
say
it's,
it's
a
it's
an
easy
example,
because
we
only
have
one
one
graph,
one
one
chain.
C
C
In
the
previous
slide,
where
we
were
showing
the
model,
you
had
network
service
with
virtual
function
and
policy,
but
I
didn't
see
kind
of
forwarding,
graphs
or
graphs
of
network
services.
So
I.
G
So
yeah
this
picture
of
open
level,
and
this
workflow
is
it's
one
at
least,
is
a
zoom
of
what
we.
What
I
presented
on
on
the
lifecycle
management
and
the
the
the
design
here
is
it's
more
detail.
So
it's
on
the
on
the
on
the
on
the
model
itself
on
the
vnf's
themselves.
G
Some
things
were
developed
by
by
orange
other
things
by
the
community
in
in
on-app,
and
the
the
design,
as
I
said,
is
a
very
software
validation
phase.
The
testing,
then,
is
a
test
of
intercom
interoperability
so
that
all
the
vnfs
are
compatible.
The
the
the
whole
sets
actually
it's
it's
an
open
source,
so
everything
is
on
the
gitlab.
You
have
the
link
here,
it's
reusable.
We
have.
G
We
used
actually
things
from
from
the
community
and
then
once
the
the
the
decomposition
is
very
verified
certified
then,
because
the
descriptions
actually
comes
from
the
vns
providers
and
once
the
the
global
service
is
created
and
submitted,
there
is
a
test
of
the
service
to
show
that
it
really
does
its
own
functionality,
the
governor,
actually
it's
linked
to
to
the
license
and
how
to
distribute
the
service
so
that
they,
the
operator
that
is
owning
the
license
to
to
to
use
this
dns
and
then
actually,
the
installation
of
the
service
is
more
details
than
one
one
wants
that
also.
G
I
don't
know
if
it
is
good
to
go
into
such
details
for
the
for
the
platform,
but
this
is
just
some
snapshots
from
the
the
vns
that
has
been
onboarded,
so
the
4g
core
control
control
plane
and
that
airplane,
with
some
functions
from
the
opener
interface
on
the
axis.
D
G
There
are
other
functions,
but
the
ones
that
are
dealing
with
the
use
case
when
we
show
the
networking
the
links
that
are
between
these
these
elements
of
the
control
plane.
So
you
can
see
the
elements
from
4g
chord,
the
validation,
for
example.
You
can
see
here
certified
the
validation
goes
with
the
description
of
the
software
itself
and
then
its
interfaces,
functional
and
testing
the
control
and
the
data
plane.
G
So
it's
not,
I
would
say,
if
it's
not
a
functional
validation
of
each
function,
because
this
is
done
one
other
development,
but
more
ways
to
ensure
that
the
design
over
the
software
design
level
in
onap
is
correct.
Each
each
component
has
its
links
as
as
as
as
defined
in
the
template.
G
Also
some
some
examples
of
the
of
the
composition
and
then
once
they
are.
This
is
the
design
validation,
so
one
once
validated
and
35
there
are
in
in
the
in
the
catalog
of
the
well.
Here
is
its
own
app.
So
it's
a
catalog
for
for
the
deployment.
G
And
once
on,
actually,
we've
tested
the
the
attachment
of
the
user
to
to
the
chain
to
the
to
the
slide,
and
we
can
see
that
status
is
on
here
with
also
the
ues
attached.
G
The
when
I
wanted
to-
and
maybe
I
I-
how
can
I
say
so
this
this
this
slide-
is
from
walter,
which
is
in
the
next
room,
actually
there's
a
parallel
meeting
and
the
idea
I
wanted
to
show
this
slide.
G
As
the
main
point
here
is
the
automated
resource,
management
and
authorization
platform
which
we
consider
as
being
onap
for
us-
and
there
is
here,
okay,
a
user
admin,
but
for
us
it
will
be
the
this
customer
layer
and
the
customer
layer
can
be
folded
into
two,
so
the
network
management
network,
slides
manager
or
network
service
manager
and
the
client
itself-
and
here
the
well
from
from
my
point
of
view,
the
intent-based
programme,
northwest
interfaces
are
mainly
used
for
exactly
the
the
the
the
the
the
interfaces
with
the
infrastructure,
these
controllers
from
network
or
cloud
from
infrastructure,
I.t
or
network,
and
it's
well.
G
It's
a
name.
It's
networking,
it's
interest-based
networking
and
for
us
the
more
higher
we
consider
this
n-type
concept,
the
more
we
make
it
easy
to
talk
to
give
the
hand
to
clients
to
talk
or
give
the
voice
to
hand
to
clients
to
talk.
So
I'm
taking
back
this
picture.
G
So
we're
absolutely
new
to
to
the
e
intent
definitions
and
what's
what
we
hope
from
getting
with
the
with
the
group,
is
to
have
organizations
to
develop
some
methodologies
to
define
maybe
abstractions
what
what
we
know
today
is
descriptions
of
vns
descriptions
of
global
services.
But
how
can
we
trigger
that?
How
can
we
present
them
in
a
very
simple
catalog
to
end
users,
for
example,
we
can
model
with
we're
working
also
actually
on
modeling
the
profiles
of
the
users.
G
A
profile
of
it
can
be
just
three
three
families
of
users,
but
let's
start
small
and
based
on
requirements,
for
example,
on
the
range
of
requirements
and
how
these,
how
the
clients
can
ask
for
a
network
slide
for
network
service
without
getting
into
any
details
of
even
slices,
actually
only
network
slides.
I
mean
no,
no
service,
sorry.
So
this
is
our
big
question
mark
and
we're
just
starting
to
work
with
this.
G
We're
also
working
on
unless,
with
with
nokia,
to
define
some
the
right
abstraction
having
even
base
event
base
is
okay,
because
it's
not
a
sequence.
It's
not
it.
It's
not
fixed
as
a
as
a
sequence,
but
if
it's
demand
based,
then
it's
much
better
and
moving
from
in
that
base
networking
so
we're
less
interested,
very
much
less
interested
into
a
lower
level.
Intent.
G
Integration
application
but
more
to
the
network
exposure
based
on
on
intent.
So,
as
I
said,
the
multi
the
multi
domain
here
is
also
at
all.
Levels
can
be
considered,
but
the
the
multi
domain
can
be
triggered
by
by
onap
or
by
another
orchestrator
in
order
to
fulfill
just
to
fulfill
the
the
network,
the
the
request
for
network
service.
So
actually
I
don't
have
more
slides.
So
I
I
was
writing
some
some
trained.
G
I
was
trying
to
to
make
a
big
picture
of
our
questions,
but
didn't
manage
to
to
finish,
but
I
think
my
main
messages
are.
I
delivered
my
messages.
So
if,
if
you
have
questions-
and
I
I
will
be
glad
to
have
discussions
on
how
we
can
move
forward
with
your
expertise
on
the
subject.
B
Thank
you,
amina
for
the
presentation,
so
speaking
more
as
a
participant
on
the
last
part
where
you
I'm
curious
about
the
the
group
can
help,
I
think
it
will
be.
The
first
point
would
be
to
maybe
list
some
of
the
key
points
to
to
address.
B
I
mean
go
to
the
first
next
level
of
detail
behind
your
your
question
to
the
group
so
that
it
will
be
maybe
easier
to
track
the
scope
of
what
we
can
where
we
can
help
on
the
when
you
are
mentioning
this
network
slice
level
for
for
the
intent.
I
know
this
is
also
joint
work
with
some
of
the
teams
in
nokia,
but
I
think
there
is
another
team,
the
one
with
sabine
and
fred
that
have
been
working
on
the
intent
for
network
slicing.
B
They
have
presented
a
demo
last
year
in
the
group,
so
they
they
made
some
hypothesis
on
how
to
use
the
network
slicing
templates
and
and
map
it
to
a
kind
of
intent-based
interface.
So
I
think
this
could
be
a
starting
point
also
for
for
discussion
about
the
linking
intent
and
network
slice.
G
Okay,
yeah
I've
pushed
in
the
in
the
meeting.
Where
was
it
no
yeah?
Well,
it
was
in
april,
but
some
some
some
questions
on
on
the
cross
domain
management
with
intent
race.
Actually
it's
it's
not
far
from
that,
but
maybe
it's
now
that's
the
time
to
to
maybe
merge
the
the
two
questions.
The
two
set
of
questions
into
one
and
present
them
to
the
group.
G
B
And
also
because
you
mentioned
this
exposure
framework
because
for
me
there
are
clearly
a
in
all
you
presented.
It's
you
go
from
the
customer
request
to
create
a
slide
out
of
the
template
and
instantiate
the
slice.
B
So
I
think
there
are
some
elements
of
answers
that
they
could
be
investigated
in
the
research
group
on
the
forms
of
those
intent
and
what
are
the
or
it
can
be
supported,
but
to
relate
to
what
you
mentioned
about
this
exposure
framework.
I
think
this
can
be
a
very
important
component
in
the
intent
functionality
so
that,
if
you
have
an
intent,
I
mean
an
incoming
intent
expression.
I
I
want
to
do
something
with
my
infrastructure.
B
There
is
an
important
step
of
mapping
or
matching
the
the
parts
of
the
intent
to
the
capabilities
of
the
infrastructure
in
terms
of
functionality
in
terms
of
performance
in
terms
of
topology
etc.
So
the
I
think
there
is
also
things
that
can
come
from
your
work
on
the
the
exposure
framework
has
a
functionality
in
the
intent
architecture
to
say
these
are
the
capabilities
of
what
is
underneath
the
interface,
the
intent
interface,
and
then
we
can
go
into
kind
of.
B
B
So
I
think
you
see
there
are
questions
about
when
you
go
top
down
how
to
express
an
intent
and
what
will
be
the
mapping
to
the
the
elements
related
to
slices,
but
I
think,
on
the
other
hand,
there
is
also
a
step
where
the
capability
exposure
framework
needs
to
push
up
advertise.
What
are
the
capabilities
in
home
of
a
graph
or
a
catalog
whatever?
And
then
this
this
this?
These
capabilities
are
used
in
the
mapping
function
to
understand
the
elements
coming
from
the
slice
request
up
to
what
kind
of
capabilities
it
can
be
mapped.
G
On
putting
all
the
resources,
the
exposure,
yeah
you're
right
so
there's
we
already
considered-
I
didn't
present
this
here,
but
in
in
other
works
the
exposure
of
the
network
towards,
let's
say
the
application
and
the
the
the
resources
other
resources
towards
different
management
frameworks,
management,
infrastructure
management
framework.
So
actually
we
we,
we
consider
this
dual
exposure
yeah.
So
it's
a
good.
G
C
C
G
We
yeah
as
a
main
as
a
main
view.
We
were
looking
at
this
exp
the
exposure
of
network
functions
globally
to
network
services.
Actually,
if
it
slices
or
not,
it's
not
that
important,
but
we're
going
towards
this
full
virtualized
core
and
access
network.
So
it's
in
the
context
of
virtual
network
virtualization
and
the
value
we're
pushing,
is
okay,
a
network
service
is
a
network
slide,
but
we
from
from
from
the
work
from
from
the
project
we're
working
in
it's
this
view
right,
the
other.
G
The
other
way
to
see
things
if
I
understand,
is,
for
example,
a
request
for
an
operation
and
not
mainly
for
a
functionality,
and
is
this
the
the
other?
The
other
view.
C
I'm
thinking
from
our
perspective,
we
are
saying
okay
for
intense
different
uses.
There
are
different
intents
okay.
This
is
the
intent
for
the
user,
who
needs
a
network
slice,
but
there
are
the
users
who
would
you
know,
administer
business
policy,
security
policies
overall
on
the
network,
and
then
you
would
need
to
do
the
heavy
design
of
your
service.
You
know
so
so
there
is
a
strategy,
intense,
potentially
the
ones
which,
which
would
include
how
you
did
some
composing
policies
and
things
like
that
or
higher
level.
C
From
that
perspective,
I'm
not
asking
maybe
just
about
your
project
but
as
on
up
as
a
whole,
if
you
are
familiar,
whether
they
are
considering
those
is
intense
as
well
or
whether
it's
just
a
slice.
G
Yeah,
I'm
not
sure
onap
itself
is,
is
it's
focusing
on
and
we're
using
on
app
in
a
global
context
of
exposure,
but
for
sure,
if
it's
security
or
even
monitoring,
for
example,
can
be,
can
be
defined
as
as
a
function
in
a
catalog
and
and
then
the
the
intent
will
trigger
to
call
such
such
such
a
function.
So
it's
it's,
I'm
just
thinking
thinking
loud,
so
it's
maybe
it's
not
that
different
and
after
all
slices
is,
is
nothing
well,
it's
nothing.
G
The
slices
is
empty
and,
and
we
we
will
be,
we
will
be
putting
the
functionalities
that
are
requested
in
in
it.
So
if,
if
the
slice
is
100
security
functionalities
and
then
why
not
or
100
management
functionality,
then
maybe
maybe
different
monitoring
systems,
then.
C
Like
in
your
case,
in
your
project,
if
there
is
some,
if
there
is
some
request
from
the
user,
ensure
that
slices
originating
here
here,
don't
go,
you
know
or
ensure
that
these
kind
of
base
stations
have
these
kind
of
poly.
You
know
something
that
is
not
per
specific
slice,
but
something.
G
D
G
All
this
is
not
yet
structured.
Well
in
our
concrete
outcomes,
there
are
things
that
exist
here
and
there,
but
it's
not
well
that
well
structured
and
also
the
languages
in
which
languages
can
we
write
this?
What
what
what
format,
what
data
models.
G
Yeah
we're
using
what
we
have
from
open
source
and
from
what
we
can
develop
internally,
and
but
it's
that
there
are
different
questions
actually
that
still
remain
open,
but
the
time
the
time
it
takes
to
to
go
and
develop
the
tools
that
allow
us
to
have
the
right
abstractions.
G
Some
other
groups
well
groups,
but
projects.
We
focus
on
the
separation
of
so
each
it's
like
separation
of
concern
to
to
to
tackle
the
abstraction
to
other
abstractions.
It's
not
it's
at
the
beginning,
but
we
we're
trying
to
have
this
approach
as.
A
A
A
I
don't
know
if
you
recall
the
last
the
last
status,
but
the
idea
is
that
we
identify
in
the
previous
ietf
meeting
and
different
challenges
that
could
serve
as
a
basis
of
indifferent,
basically
section
of
the
document,
different
challenges
that
we
are
regarding
as
it
is,
for
example,
the
problem
of
having
data
the
problem
of
taking
action
and
not
only
and
doing
analysis
and
so
on,
and
then
once
we
define
the
structure
of
the
documents.
A
Is
this
outline
then
also
we'll
have
some,
let's
say,
let's
say
some
use
case
or
scenarios
to
illustrate
all
the
different
challenges
and
why
we
are
not
yet
at
the
point
to
resolve
the
challenges.
A
Of
course,
we're
doing
also
a
bit
of
state
of
the
art
and
several
what
are
the
current
proposal
and
until
which
point
they
as
they
are
currently
so
we
we,
we
will
have
to
have
this
more
concrete,
let's
say
outline
or
drive
a
footline
ready,
almost
I
think,
beginning
of
the
year.
Unfortunately,
we
I
personally
did
not
have
enough
time
to
do
because
I
promise
to
eat,
but
I
will.
A
I
will
of
course,
work
on
it
as
soon
as
I
can-
and
I
I
hope
in
the
next
in
the
next
week,
to
do
that
and
they'll
share
with
people
who
express
their
interest
into
participating
to
writing
this
document.
B
As
we
discussed,
I
think
it's
we
received
a
several
expression
of
interest
to
participate
such
an
initiative.
I
think
what
remains
to
be
to
be
seen
is
how
we
want
to
drive
it.
I
mean,
and
I
don't
want
to
prescribe
any
specific
approach,
but
I
would
suggest
that
I
mean
having
a
way
so
that
several
people
can
contribute
easily
to
to
provide
inputs,
feedback
on
the
or
even
suggest
some
challenges,
and
then
there
are
so
maybe
some
owners
that
try
to
organize
it
into
a
meaningful
structure.
B
I
mean
this.
We
should
try
to
lower
the
bar
so
that
people
can
contribute
easily,
because
we
have
some
contributors
that
are
not
so
used
to
writing
a
draft
or
even
use
the
ietf
tools.
Of
course,
yes,
but
I
think
on
the
other
hand,
we
have
also
some
people
that
are
active
in
the
research
group
on
ietf
and
that
are
expecting
this
to
be
visible
or
accessible
in
some
places,
not
only
like
a
google
drive
document
or
as
a
work
in
progress
as
a
publication.
B
So
I
think
we
need
to
find
a
way
to
to
support
both.
Don't
have
a
clear
answer
now
but,
as
we
discussed,
I
think
this
should
be.
I
mean
I
think
people
are
ready
to
help.
We
should
maybe
just
give
them
the
opportunity
to.
A
So
so
so
maybe
my
my
idea
that
I
have
in
mind,
but
of
course
it
has
to
be
discussed,
as
you
say,
we
don't
want
to
to
force.
Some
kind
of
approach
is
that
I
understand
the
need
for
the
visibility
and
that
a
draft
can
be
a
good
way,
but
at
the
first
point
maybe
is
a
kind
of
yeah,
google,
doc
or
whatever,
and
just
to
maybe
share
what
what's?
A
You
don't
know
if
you
see
what
I
mean
just
asking
people,
what
are
the
important
challenges
for
you
just
write
a
small
paragraph
for
each
of
them
and
then
you
can
try
to
to
consolidate,
because
of
course,
then
there
will
be
some
some
effort
to
do
to
have
something
integrated
all
together,
but
I
I'm
not
sure
we
can
start
directly
with
with
the
writing
address.
It's
my.
A
A
So,
thank
you
all
for
for
participating,
yeah.
We
are
the
end
that
hope
to
to
see
you
talk
to
you
soon
in
the
next
meeting.
Thank
you
very
much.