►
From YouTube: IETF109-NETMOD-20201118-0730
Description
NETMOD meeting session at IETF109
2020/11/18 0730
https://datatracker.ietf.org/meeting/109/proceedings/
A
Okay,
I
think
we're
ready
to
get
started.
Welcome
to
virtual
itf
108
netmod
whoops,
I
should
have
said
109..
I
was
remembering
the
old
slide,
I'm
lou
berger.
We
have
my
co-chairs
kent
watson
and
joel
jugley
online.
We
are
watching
jabber
and
taking
minutes
through
the
online
note
taker
tool.
We
welcome
you
to
join
so
we'll
show
the
link
a
couple
of
times,
but
here's
the
link
the
first
time
it's
also
already
in
chat
session.
A
So
this
is
the
ietf
notewell.
It
covers
all
of
our
formal
ietf
meetings.
I
think
people
are
largely
familiar
with
it.
If
you're
not,
it
basically
says
that
everything
you
say
here
becomes
part
of
our
public
record
and
is
a
formal
contribution
to
the
ietf.
If
you're
not
familiar
with
the
notewell.
Please
take
a
look
at
the
link
on
the
bottom
and
read
it
at
your
leisure.
B
A
Media
echo
will
cover
cue
control,
there's
a
button
if
you'd
like
to
raise
your
hand
and
join
in
once
you're
called
on
just
unmute
yourself
and
ask
your
question
or
make
your
comment.
Meat
echo
is
automatically
taking
care
of
blue
sheets,
it's
also
a
gateway
into
jabber.
A
Again,
if
you
want
to
look
at
the
jabber
it's
for
me,
it
shows
up
as
a
little
red
icon
next
to
the
the
people
account
which
looks
like
we're
now
at
33..
Please
do
join
us
in
our
collaborative
note-taking
on
cody
md
again
the
link
is
here:
it's
been
sent
into
the
chat.
It's
also
been
sent
to
email.
It's
particularly
important,
if
you
say
something,
make
a
comment
on
at
the
our
virtual
mic
that
your
notes
are
captured.
So
please
very
comment
is
captured
in
the
notes.
A
Sorry,
please
make
sure
to
go
and
check
that
whoever
is
doing
the
note
taking
as
accurately
reflects
your
comments
the
agenda.
Actually,
I
think
we're
ready
for
that.
So,
let's
go
to
the
next
slide.
A
The
agenda
has
not
been
updated.
Although
I'll
note
that
that,
for
the
yang
versioning
update,
there's
actually
two
presentations
uploaded
two
different
powerpoints,
I
believe
jan
will
go
first
followed
by
bo
and
I
think
we
still
have
a
little
bit
of
time.
Maybe
five
minutes
that's
not
accounted
for,
even
though
we
are
a
very
short
session.
So
I
think
we
have
enough
time
for
discussion
and
although
we
always
seem
to
run
out
of
time
at
the
end,
but
we'll
try
to
keep
things
moving
along
with
the
agenda
next,
please.
A
We've
had
one
rfc
published
since
the
last
meeting.
Thank
you
all
to
who
contributed
to
that.
That's
obviously
why
we're
here
is
to
produce
the
documents
such
as
this.
So
thanks.
Thank
you
again.
We
have
two
documents
in
the
rfc.
Editor
queue,
one
is
in
off
48
and
I
think
we
have
to
poke
the
authors,
and
I
should
say
that
I
think
I'm
a
co-author
on
that
one.
A
So
we
have
to
book
the
authors
to
make
sure
that
they
respond
so
that
we
can
complete
that
one
and,
I
think
anything
noteworthy
to
say
on
syslog
model.
C
Only
that
is
dependent
on
the
work.
That's
in
that
conf.
The
client
servers
suite
address
specifically
a
tls
client
server,
which
will
hopefully
be
going
into
working
group
last
call
in
the
not
maybe
in
a
month's
time.
So
once
once
it
gets
published,
then
the
syslog
draft
will
be
unblocked.
C
A
Great
thank
you.
We
also
have
one
draft.
That's
in
iesg
processing,
the
nmda
diff,
and
we
have
several
that
are
post
last
call
we'll
see
if
rob
is
rejoined,
he's.
Actually
the
author
we're
waiting
on
for
the
the
first
two
of
these
posts.
Last
call,
obviously,
now
that
he's
in
ads
finding
time
to
wrap
things
up
is
difficult.
A
The
geo
location
we
had
been
waiting
on
an
update.
I
believe
that
has
been
updated
since
the
last
meeting
and
then
so
I
think
that's
back
with
kent
and
we
have
instance
file
format
and
we're
waiting
on
an
update
from
the
authors
on
that
and
we'd
like
to
note
that
that
is
expired.
A
So
authors,
please
do
continue.
Processing
that
I
don't
know
if
you
have
any
of
the
authors
who
want
to
make
a
comment
on
any
of
these
drafts,
please
feel
free
to
join
the
queue.
If
you'd
like
to
make
a
comment,
as
an
author.
C
We
have
rob's
in
the
queue
and
also
I
see
blush
is
queued
up
so.
D
I
I'll
just
quickly
say
on
those
two
drafts.
Yes,
I
will
try
and
get
to
those
and
finish
those
off
as
soon
as
I
can
and
again
the
same.
I
know
I
need
to
reply
back
to
andy's
comments
on
the
lna
diff
draft.
So
again,
hopefully
the
next
couple
of
days,
I'm
going
to
respond
to
that
a
bar
may
lose
a
connection
at
any
time.
Although
just
be
aware.
A
All
right,
thank
you.
Next
slide,.
A
We
haven't
had
any
liaisons
incoming
outgoing,
I
believe
rob
you
were
going
to
bring
this
up
in
the
next
session.
Is
since
you're
having
audio
issues.
Do
you
want
to
make
any
comments.
A
D
Yes,
so
so
effectively,
there's
been
some
discussion
with
in
the
packages
and
the
version
of
work
between
some
of
the
authors
about
the
yang
models
that
are
being
done
within
3gpp
and
oran,
and
some
of
the
approach
approaches
that
they're
trying
to
use
to
reuse
those
models
and
to
almost
like
standardize
deviations.
D
So
I
think
it
would
be
useful
to
have
some
discussion
on
the
list
of
of
what
they're
doing
there,
and
it
may
be
that
we
need
to
be
helpful
for
netmod
to
make
a
liaison
to
either
3gbp
and
or
oran
about
how
they're
trying
to
reuse
yang
models
and
give
them
some
violence
on
some
advice
on
how
best
to
do
that
and
what
problems
that
they
can
hit.
If
they
don't
don't
change.
Attack.
Balaj
can
probably
speak
to
this
as
well
for
3gpp.
B
A
Okay,
in
whatever
we
send
formally,
it's
very
important
that
we
reference
back
to
either
rfcs
or
graphs
covering
the
topic,
and
I
think
we
have
appropriate
text.
We
can
reference
in
on
this
topic,
but
if
we
find
that
we
don't,
we
might
have
to
generate
some
in
the
form
of
a
new
draft
or
an
updated
draft.
A
A
So
finally,
we
just
want
to
point
out
that
we
are
going
to
continue
with
this
working
remote
motive
for
at
least
a
little
while
I'm
sure
folks
have
noticed
that
iedf
110
is
going
to
be
online,
we're
not
sure
how
much
longer
it
will
continue
through
next
year,
but
clearly
our
current
mode
of
working
will
continue
into
next
year.
So
we
want
to
make
sure
to
continue
the
energy,
the
momentum
that
we
have
in
the
working
group
and
to
do
that
between
meetings-
and
we
think
you
know.
A
Obviously
our
working
group
mail
list
is
a
key
way
to
to
collaborate,
but
we
also
have
other
tools.
We
have
both
informal
and
formal
virtual
meetings
that
we
can
do.
We
have
the
webex
available.
Some
groups
are
already
making
use
of
that
if
you'd
like
to
make
use
of
it
for
informal
meetings.
A
Both
these
informal
and
virtual
meetings
are,
of
course,
announced
on
the
working
group
list
and
are
open
to
all.
That's
I'd
like
to
quickly
note
that
that's
true
even
for
these
informal
meetings,
if
you're
you
know
using
the
the
working
group
webex,
we
intend
that
we'll
make
that
open
to
everyone
and
who
knows
you
may
even
have
someone
come
and
contribute
in
a
positive
way
that
you
wouldn't
expect.
A
So
there
are
definitely
benefits
and,
of
course,
there's
the
the
most
important
benefit,
which
is
by
including
the
working
group
we're
helping
build
consensus
on
whatever
work
is
being
done.
A
With
that
leave
we
can
move
on.
Does
one
of
my
co-founders
want
to
add
anything.
C
No,
except
that
it
does
take
a
little
bit
of
time
to
flip
between
slides
and
also,
I
was
wondering
you
mentioned,
there's
two
presentations
for
the
versioning,
which
one
goes
first,
the
update
or
packages
probably
update.
So
I
think.
F
F
It's
so
I
was
picked
to
represent
the
the
group
of
people,
authors
and
non-authors
that
have
been
working
on
the
versioning
solution
set
mostly
because
of
time
zones.
I
believe,
because
I'm
in
europe
so
next
slide,
please
it
will
be
giving
a
very
short
solution
overview
here
and
then
we'll
go
into
much
versioning
semantic,
versioning
and
yang
packages
and
bo
will
present
the
last
piece
because
that's
the
most
interesting
one
where
we
are
spending
most
of
the
time
in
our
work.
F
F
So
this
giant
versioning
solution
consists
of
five
different
drafts.
One
is
about
how
we
handle
revisions.
I
mean
yang,
has
his
own
revision
scheme
already,
it's
very
useful
for
standardization
bodies,
which
typically
have
a
single
line
of
release.
Release
train
if
you
like
everything,
is
a
very
linear
model,
but
many
of
the
vendors
that
are
using
yang
find
that
model
is
somewhat
troublesome
because
we
have
several
release
trains
going
out
in
parallel.
F
So,
like
you
have
the
five
dot
x
line
of
products
and
the
six
dot
x
line
of
products
going
out
at
the
same
time,
so
how
we
handle
revisions
when
there
are
several
living
release
trains
at
the
same
time.
That's
the
question
there
and
number
two
is
about:
how
do
you
name
those
versions?
So
when
do
you,
how
do
you
use
version
numbers
to
reflect?
F
F
Then
we
are
spending
most
of
their
time
right
now
on
packages.
So
how
can
you
make
release
releases
of
sets
of
modules
at
same
time
and
version
them
as
a
group,
and
then
we
have
potential
operations
for
a
client
to
choose,
which
version
of
a
yang
interface
that
this
client
is
prepared
to
use
in
the
case
that
the
server
supports
more
than
one
and
then
some
tooling,
for
how
do
you
compare
two
yang
malls
and
see
what
is
compatible
and
not
and
so
on?
F
We
have
bumped
the
comparison
and
the
versioning
drafts,
but
we
haven't
really
added
any
real
information
in
them.
It's
just
some
editorial
details
and
updating
the
xml
format
and
so
on,
and
even
though
we
spent
considerable
time
on
packages,
we
still
see,
there's
lots
left
to
go
there
and
that's
where
our
focus
will
will
remain
for
a
while.
We
think
next,
please.
F
And
we
are
very
interested
to
have
everybody.
That's
interested.
We
are
interested
to
have
interested
parties,
participate
and
contribute
to
this
work,
and
we
are
having
weekly
calls
on
tuesdays,
2
pm
uk
time.
Anybody
can
can
come
in
and
contribute
their
authors
and
others.
We
have
a
github
issue
tracker.
F
So
one
of
the
things
we've
been
talking
about
is
white
space
changes.
It's
we
have
seen
in
packages
and
the
when
we're
discussing
that
it
can
be
convenient
to
have
a
checksum
of
the
packages
and
to
check
some
of
the
modules
so
that
we
can
see
that
things
have
or
have
not.
F
The
white
space
changed
or
not,
and
if
we
want
to
have
checksums
on
modules,
probably
white
space
changes
would
constitute
a
version
change,
at
least
at
the
editorial
level,
but
we
can
also
say
then
that
deleting
a
space
or
adding
a
blank
line
should
not
really
it
doesn't
change
the
interface.
It
does
change
the
model,
the
yang
model,
but
it
does
not
really
change
the
interface
that
the
young
world
describes.
So
how
do
we
treat
versioning
as
related
to
that?
F
F
F
Well,
module
a
could
import
module
b
or
version
200
or
derive
it,
for
example,
saying
that
two
one
five
or
two
to
six
could
work,
but
this
is
proving
somewhat
problematic
when
there
might
not
be
any
200
revision
and
there
might
be
a
three
or
a
revision
which
might
not
be
compatible.
F
So
how
should
the
compiler
or
whoever
is
doing
this
import,
find
out
what
we
are
actually
importing,
and
this
is
actually
where
one
of
the
sticky
points
that
we
decided
to.
Let's
do
packages
first,
because
packages
might
specify
exactly
which
modules
that
are
being
used
or
something.
So
that's
one
of
those
areas
where
we
said
that
resolving
packages
might
actually
resolve
this.
C
As
chair
not
not
on
this
one,
but
on
the
list
on
chat
window
tom
hill's
asking
some
questions
tom,
can
you
speak
to
the
mike?
Please.
G
Yes,
hello,
can
you
hear
me.
G
G
F
Way,
yes,
I
think
it
makes
good
sense
to
treat
even
white
space
changes
as
a
version
change
at
least
an
editorial
change
for,
for
many
reasons,
like
version
control
and
for
keeping
track
of
well.
If
somebody
is
cleaning
up
line
endings
or
whatever
to
make
it
proper
module,
everybody
else
should
be
basing
their
work
out
of
that.
So
then
it
needs
to
be
treated
and
like
inversion,
so
it
makes
good
sense.
From
that
point
of
view,
I
believe,
sir.
F
Yeah,
that's
fine!
I
don't
know
if
I
should
looking
at
the
participant
list
through
the
chat
now
with
okay.
So
when
it
comes
to
this
import
statement
that
is
importing
not
by
revision,
but
by
revision
or
derived,
we
come
to
the
question
whether
a
change
of
the
module
that
is
being
imported
or,
if
the,
even,
if
the,
if
you
change
the
import
statement,
so
that
your.
F
So
we
have
option
a
that.
Any
change
of
the
module
is
incompatible
or
option
b
that
yeah
as
long
as
the
resulting
schema
is
not
changing.
It's
considered
to
be
a
backwards,
compatible
change
and
that's
discussion
has
been
going
back
and
forth
and
there's
many
small
things
to
trip
over
here,
but
the
recommendation
from
the
group
is
probably
option
b,
as
I
have
seen.
I
think
it's
on
the
next
slide.
Actually,
next,
please.
F
Yeah,
so
the
rough
consensus
is
option
b
and
we
think
that
as
long
as
schema
is
not
changing,
a
change
of
the
of
the
import
statement
itself
should
be
considered
editorial
change.
But
further
discussion
is
needed
and
if
you
have
any
comments
on
this
either
now
or
on
the
list,
that
would
be.
F
F
F
F
F
When
we
submit
rfcs
with
yang
modules
to
ayana,
ayana
has
some
instructions
about
assigning
rfc
numbers
and
things
like
that,
and
one
of
the
things
that
needs
to
be
added
to
rfcs
as
they're
being
released.
Is
this
statement
whether
a
particular
release
is
backwards
compatible
or
non-backers
compatible,
and
it
is
of
course
possible
that
ayanna
would
add
this,
but
we
feel
that's
probably
something
like
for
for
itf's
purposes
anyway,
that
the
young
doctors
group
or
something
would
be
the
group
to
decide
about
non-backers
change
or
whether
to
call
a
particular
release.
Non-Backers
compatible
or
not.
F
F
F
G
Thank
you,
tom
hill
from
bt.
I
didn't
mention
that
last
time
I
have
a.
I
have
a
big
problem
with
the
idea
of
having
one
module
being
used
for
the
exports
of
telemetry,
for
example,
and
then
wanting
to
upgrade
to
a
newer
model
that
has
a
new
feature
or
fixes
something,
and
it
will
take
months
possibly
years
to
upgrade
thousands
of
routers.
G
So
it
is
very
useful
for
us
to
know
if
there
is
any
incompatibility
between
versions
of
models
extremely
useful,
and
there
will
also
be
you
know.
This
is
almost
a
flag
as
well
for
us
to
be
able
to
say
you
know,
this
is
going
to
be
a
hard
convert
between
two
models
and
we're
going
to
have
to
put
some
effort
into
converting
for
the
various
consumers
of
that
telemetry.
F
Very
good,
I
think
very
few
would
argue
with
that.
Nobody,
I
heard,
but
the
devil
comes
in
the
details.
So,
if
you
are,
it
would
be
interesting
to
hear
your
views
more
in
detail
in
the
tuesdays
discussions
that
we
have,
because
determining
what
is
backwards
compatible
and
not
compatible
may
be
somewhat
subjective,
understandable.
C
C
C
Thank
you.
Yes,
he
is
go
ahead.
D
Can
I
just
jump
in
on
that
that
previous
point,
I
I
sort
of
agree
where
you're
coming
from.
I
think
the
tricky
case
is
that
there
are
some
yang
modules
that
are
derived
from
other
registries.
I
think
you
have
like
a
one
of
the
bgp
ones
is
so
whenever
they
add
a
new,
I
don't
know,
address
family
or
something
to
bgp
or
something
like
that.
Then
you
end
up
creating
a
yang
module
at
the
same
time.
D
So
I
think
from
an
I
a
point
of
view,
there
needs
to
be
some
consideration
about
what
happens
for
those
sort
of
auto
generated
yang
models,
and
there
have
been
some
cases
recently
where
the,
where
there's
been
a
desire
to
have
some
non-banks
compatible
changes
to
those
I.e,
fixing
the
description
statements
or,
in
some
cases,
even
changing
some
of
the
names.
So
I
think
that,
whilst
for
most
modules,
I
think
I
agree
it's
the
authors
or
the
young
doctors
that
should
be
driving
this.
D
C
H
Yeah-
okay,
I'm
just
good
here
to
give
some
quick
updates
on
the
young
packages
next
slide.
Please
yeah
jen
just
mentioned
there's
a
lot
of
time
spent
on
package
discussion,
but
we
right
now.
H
We
just
made
some
changes
to
the
other
issues
that
we
currently
discussed
in
the
core
meeting
and
the
the
major
three
parts
that
we
changed
since
108
is
the
first
one
that
we
added
a
new
division
rule
for
the
package
definition
and
then
the
next
one
is
a
new
section
that
just
this
one
is
related
to
rob
just
and
blush
just
mentioned
about
the
different
sdos
deviations
handling.
H
So
this
is
a
new
section
to
to
related
to
that,
and
then
it's
mainly
the
editorial
changes
and
all
the
discussion
is
ongoing
and
the
gcap
is
the
link
is
below
next.
Please
and
here's
a
the
package
usage
guidance
related
to
just
rob
and
blush
mentioned
that
packages
is
recommended
to
publish
for
better
interoperability
and
deviation
should
be
reserved
for
implementation
rather
than
specification
and
publishing
a
package
is
based
on
another
standard
package
with
some
division
is
very
dangerous
because
they
may
lead
to
some.
H
Some
cannot
be
changed
after
the
publish
the
standardization
and
the
public
standard
should
make
use
of
property
feature
statements
to
provide
flexibility.
So
this
is
a
the
summary
of
the
guideline
from
that
section
next
piece
and
then,
with
the
open
issues
we
we
mainly
have
the
main
four
open
issues.
The
first
one
is
game
amount
and
packages.
H
They
also
think
that
it
can
contain
module
with
mount
points
that
should
be
additional
information
for
the
package
to
have
to
to
carry
some
points.
Information
to
be
augmented
with
like
a
list
of
package
could
be
mounted
on
the
mount
point.
So
so,
but
currently
we
are
still
in
the
discussion
about
that.
So
we
would
like
to
hear
more
feedback
from
the
working
group
if
on
the
mailing
list
or
right
now,
it
could
be,
it
could
be
appreciated.
H
We
don't
mandate
that
package
have
the
revision
date,
but
young
allows
module
without
a
revision
date.
So
we
we
we'd
like
to
hear
more
feedback
from
the
working
group.
Whether
there
could
be
some
options,
some
some
suggestions
on
this,
but
there
could
be
some
issues
that
it
could
be.
There
be
more
multiple
versions
without
revision
identified
for
some
young
audio.
H
Then
better
allow
a
circular
young
package
inclusion
with
we.
We
discussed
this,
but
right
now
we
don't
have
a
question
an
announcer
yet,
but
we
think
blocking
circular
inclusion
may
prevent
different
groups
from
independently
defending
their
packages
like,
for
example,
give
an
example
about
oem
and
routing
it
could
be
could
be
depending
on
each
other.
But
if
we
block
that,
then
it
should
be
coordinated
between
the
two
working
groups
to
define
those
packages.
H
This
is
the
last
open
issue
for
this
time
and
we
just
gave
the
the
guidance,
but
we
think
that
might
be
also
applies
to
the
one
individual
young
module
so
right
now
we
think
our
proposal
is
a
define
new
section
in
the
packages
drop
to
give
the
guidelines
on
the
division
and
feature
usage,
but
right
now
we
we
are
happy
to
hear
more
feedback
on
this
from
the
working
group.
So
that's
all
the
presentation
you
know.
A
Bo
a
question
on
the
last
slide
yeah:
I
did
you
talk
about
other
alternatives
and
the
obvious
take
to
8407
to
the
guidelines.
Privacy.
H
Right
now
it's
in
the
new
section
in
the
packages
draft,
but
the
rob
think
if
it's
it's
in
the
like
the
previous
published
author
and
young
guidelines,
it
could
be
quite
separate,
I
mean
rob,
can
you
add
something
here?
I
think
we
will
discuss
this.
D
Yes,
so
I
so
so
lou
the
answer
is
yes,
we
did
discuss
this
that
are
confirmed
with
adding
into
the
the
yang
author
guidelines
is
currently
that
document
is
focused
on
only
modules,
so
it
felt
that,
and
the
topic
is
about
guidelines
for
module
authors
as
to
whether
that
would
be
overreaching
the
scope
of
that
document,
but
certainly
happy
to
have
a
discussion
as
to
where
the
best
place
is
to
put
that
text.
D
A
That
that's
sensible
and
we
could
always
use
the
updates
approach.
We're
saying
that
this
section
updates
that
document,
so
that
readers
of
that
document
in
the
future
of
8407
know
to
look
to
this
new
rfc.
D
That
makes
sense.
So
can
I
just
ask
a
separate
question
because
I
obviously
missed
a
little
bit
of
the
of
the
meeting
earlier
on.
Well,
has
there
been
any
discussion
about
us
holding
virtual
interim
meetings
and
is
that
worth
something
that
we
should
discuss
now
at
all.
A
We
did
mention
that
that's
possible.
We
said
that
those
could
be
initiated
either
by
the
chairs
or
at
the
request
of
the
working
group.
Are
you
requesting
a
virtual
on
this
versioning
topic?
Is
that
what
you're,
where
you're
headed.
D
Yes,
I
think
so
we
have
discussed
this
briefly,
the
last
few
days
between
the
authors
and
they're,
all
quite
keen
that
they
think
that
that
would
be
good
to
have
a
few
virtual
meetings,
a
virtual
interim
meeting,
setting
up
focusing
on
on
trying
to
resolve
specific
issues.
So
maybe
we
can
follow
up
with
you
on
the
chairs
offline
and
try
and
get
those
set
up.
Please.
A
Yeah,
I
think
that
would
be
great
and
perhaps
yeah
we
can.
We
can
sync
up
and
maybe
send
something
to
the
list
done
some
proposed
times.
Okay,
thank
you.
I
I
Yeah
for
people
who
don't
know
what
is
this
java
about?
Actually
this
child,
they
define
the
data
object
tag.
Actually,
the
user
tag
is
actually
widely.
You
know
used
for
classification
organization
and
not
only
in
the
idea
protocol,
but
also
in
the
internet
and
on
module
tag.
Actually
is
one
of
example.
He
provides
a
young
model.
Classification
can
classify
young
model
into
network
animal
element
model
and
a
network
service
model
user
group
model,
and
we
borrow
the
same
concept.
I
Actually,
we
introduce
self-describing
data
object,
tag
which
you
can
classify
the
management
and
operation
data
into
several
categories.
From
this
category
we
can
identify
this
characteristic
data,
so
the
typical
cases
is
using
in
the
streaming
animation.
We
can
filter
the
query
and
reduce
the
amount
of
data
to
export
to
the
client.
I
I
So
so
what
is
the
current
status
of
this
job
that
this
job
has
been
present
for
many
times?
Actually,
we
ipr
account,
has
on
this.
Job
has
been
issued
and
we
have
confirmed,
and
we
also
request
the
working
adoption
call
on
version
five,
and
we
got
a
lot
of
support
and
also
a
lot
of
comments,
and
many
thanks.
I
You
can
yukon
and
many
other
colleagues,
and
it
was
a
great
suggest
actually
by
the
working
group
and
for
operation
operation
type,
maybe
should
be
defined
in
specific
device
device
models
such
as
minimize
minimal
value,
maximum
value
average
value.
We
think
this
makes
sense
and
for
this
job
they
should
focus
on
object,
specific
attack
and
which
provide
the
general
utility.
I
So
in
this
latest
version
version
6,
we
actually
remove
operation
type
extension
statement
and
also
we
actually
remove
other
complex
tags
like
a
metric
precision,
magic
scale
and,
in
addition,
we,
you
know,
take
out
a
service
attack
and
task
attack
extension
because
for
multi-tag
actually
can
be
reused
for
this
same
purpose
and-
and
we
also
rewrite
the
use
cases,
try
to
clarify
the
relationship
between
the
object
and
the
property
separation
and
measure
object
and
give
the
example
how
to
use
them
and-
and
we
also
have
some
other
editorial
changes
and
next.
I
This
is
three
category
and
in
this
in
the
left,
we
show
the
relationship
between
these
three
categories.
Actually,
you
can
see
that
one
object
can
contain
well,
two
or
many
property
sub
objects,
and
one
object
can
also
have
one
too
many
metric
sub
objects
for
object.
It
usually
defined
as
a
container
or
at
least
leave
list
for
property.
It
is
actually
diff
node
and
the
magic
sub
object
that
we
think
it
can
can
be
the
leave
list
and
list
and
also
can
be
the
the
leaf.
I
So
in
the
right,
we
give
the
example
how
we
can
tag
them.
The
data
objects
are
defined
in
the
in
in
the
young
data
model.
For
example,
we
have
a
tunnel
pm
model
and
we
have
tunnel
service
root
node
and
we
also
have
name
create
time
and
average
latency
as
a
child
child
node
to
the
tunnel
service
router
node.
I
Next,
so,
in
addition
to
define
the
con
the
opm
tag,
we
also
introduced
two
context
information,
since
we
already
classified
the
management
algorithm
data
into
three
categories,
which
include
the
metric
sample
object.
We
can
use
the
metric
group
to
further
classify
his
metric
separate
object
into,
for
example,
loss
related
or
cheater
related
and
delay
related
category
and
another
context.
Information
tag
is
multi-source
tag.
Actually
this
will
use
for
the
multi-source
aggregation.
So
we
support
two
different
types
of
aggregation.
Type.
One
is
aggregate,
the
other
is
no
aggregator.
I
I
I
A
Who
is
contributor?
I
did
ask
a
question
on
the
list
that
didn't
get
answered
I'll
ask
again
and
just
be
interested
in
seeing
the
response.
I
Yeah
yeah,
I
I
just
say
you
asked
this
question
and
we
actually,
I
I
see
some
colleagues
actually
respond
to
anna
for
for
people
who
are
interested
and
we
actually
you.
We
can
comment
on
that
and
get
more
feedback.
C
I
Okay,
this
is
ching
again
and
I
will
present
this
easy
policy.
Imagine
draft
update
and
I
will
try
to
make
sure,
because
we
run
out
of
time
and
next
and
actually,
what
is
this
job
about?
Actually
easy
is
considered
as
a
special
case
of
the
policy
management
and
it
includes
it
comprise
the
event,
condition
action
three
k
elements,
so
it
can
enable
the
event-based
management
provider,
a
rapid
response
to
the
network
events,
and
we
use
a
young
data
model
language
to
express
these
key
elements
in
the
easy
policy.
I
Actually,
this
easy
can
be
implemented
in
two
different
ways.
One
is
centralized
network
management.
The
second
is
device
safer
management
device.
Several
management
can
also
see,
as
distributed
policy
management
for
centralized
nano
management.
I
It
can
provide
a
nanowide
visibility
to
water
happening
in
a
networker,
but
it
may
require
the
huge
resource
consumption
when
facing
the
massive
data
connection
and
processing,
and
it
may
have
slow
reaction
to
the
network
change
to
address
these
challenges.
We
introduce
these
device
server
management.
I
So
what
is
the
current
status
of
this
job?
These
jobs
actually
has
been
around
for
a
while.
It's
harmonized
with
several
com
problems,
easy
related
work
and
we
present
in
106,
and
we
actually
get
agreed
to
adopt
this
worker
and
I
suggest
to
define
this
model
more
independent
from
the
nether
comp
protocol.
So
we
did
a
request
working
with
dobson
on
version
six,
and
we
got
a
lot
of
comments
and
also
there's
some
concern
on
this
essay
model
design.
I
So
it
was
suggested
to
use
expats
for
logical
expression
and
also
it
was
suggested
to
use
combination
of
x
paths,
policy,
variable
new
xpath
function
and
to
bind.
This
is
a
common
to
scripting
language,
so
we
actually
after
ietf
for
100
107
meeting.
Actually,
we
discussed
with
many
problems
and
we
also
get
andy
on
board
and
discussed
with
many
other
companies
like
eager
hank,
chevron
and
void,
and
we
set
up
some
conference
meeting
to
discuss
this
issue.
We
also
exchanged
email,
offline
and
actually
latest
version.
I
We
get
a
review
by
jonathan,
but
you
only
focus
on
some
abstract
section
or
introduction
section.
So
in
the
latest
version
we
make
a
lot
of
changes.
The
big
changes
will
rewrite
as
a
easy
model
usage
use
usage
example.
We
also
added
two
new
usage
example:
one
is
related
to
the
cpu
utilization
shop
shooting.
The
second
is
really
to
the
root
log
thumb
and
and
based
on
the
discussion
and
many
this
discussion.
I
We
think
it's
more
reasonable
to
reintroduce
some
function
called
rpc
core
and
to
allow
his
computation
logic
to
work
and-
and
we
also
move
the
notification
operation
as
separate
notification,
since
we
believe
the
notification
is
changed
between
the
remote
client
and
the
server
and
also
we
add,
is
a
function
library
list
in
the
essay
model.
Next.
I
So,
to
make
sure
everybody
on
the
same
page
actually
will
introduce
several
terminal
terminologies.
The
key
technology
we
introduce
is
easy
event.
It
will
use
a
capture.
The
telemetry
subscription
data
source
use
this
as
input
to
the
easy
logic
and
for
events
stream.
Subsequently,
we
introduce
server
event
which
can
trigger
to
generator
notification
for
event
stream
subscription
for
data
store
subscription.
I
We
introduce
a
datastore
event
which
can
trigger
the
notification
to
be
generated
for
datastore
subscription.
We
also
introduce
a
timer
event
and
a
diagnostic
event.
The
timer
event
actually
can
allow
you
to
call
the
ese
logic
periodically,
which
is
similar
to
the
crown
job
task
management
for
a
diagnostic
event.
Actually,
it
were
used
to
test
the
easy
logic
logic
to
make
sure
it
works
properly.
I
I
So
what
is
the
use
cases
for
these
several
device,
separate
management?
So
in
the
next
we
introduce
this
architecture,
so
the
easy
model
will
be
populated
in
the
server.
The
server
will
act
as
a
local
client
to
do
the
safer
management.
It
will
keep
track
of
the
data,
object,
change
and
will
trigger
the
action
when
a
certain
condition
to
be
met,
and
so
we
have
several
use
cases.
I
The
typical
use
case
is
a
smart
filter
which
actually
will
use
the
server
event,
and
in
this
case
actually,
for
example,
you
can
scan
specific
type
of
interface
every
five
seconds
and
you
can
check
the
counter
status
and
return
an
area
of
the
interface
entries
that
match
the
search.
In
this
case,
we
may
need
to
store
some
data.
For
example,
you
can
check
whether
the
same
event
occur
twice,
so
you
need
to
define
some
variable.
We
call
the
policy
variable
to
store
this
state,
and
another
case
is,
is
about
the
cpu
utilization
troubleshooting.
I
Actually,
this
actually
doesn't
require
any
state
to
be
stored.
For
example,
when
cpu
utilizing
utilization
goes
beyond
60
and
you
can
output
a
stack,
cpu
fan
information
to
the
flash,
and
then
we
also
introduce
a
timer-based
event
and
a
diagnostic
event
for
timer
event.
For
example,
you
can
use
the
watchdog
to
dump
the
router
log
every
three
minutes
to
the
flash
so
diagnostic
event
just
used
to
test
the
easy
logic:
freedom
of
smart
field
logic
to
make
sure
it
worked
properly.
I
I
I
In
some
cases,
you
need
to
multiple
step
interaction
between
the
local
client
and
server.
You
can
use
this
policy
variable
to
store
this
state
and
also
we
actually,
you
know,
model
their
relationship,
for
example,
for
each
esa
entry.
We
will
conclude
one
event
that
can
also
contain
multiple
condition:
action
pairs
for
each
connection,
action
pairs.
It
will
include
multiple
conditions
and
one
action.
Therefore,
one
action
can
it
can
have
multiple
action
items.
This
is
how
we
design
this
easy
model.
Next.
I
So
here's
a
typical
example:
you
may
need
to
suppress
some
duplicate
event,
for
example,
for
internet
status
down
you
without
easy
logic.
You
may
send
this
event
multiple
times
and
so
with
the
easy
logic
we
can
design
you
can
you
can
subscribe
this
freedom
interface
model
to
to
to
satisfy
specific
type
of
interface
and
and
you
can
scan
or
interf
internet
interface
check,
whether
the
interface
stats
is
done.
Also,
you
can
create
the
event
occurrence
counter
to
keep
track
of
the
occurrence
of
the
same
event.
I
If
it
happen,
you
know.
I
If,
if
the
occurrence
time
you
see
the
pre-configure
the
threshold,
you
will
suppress
the
event
and
next
so
how.
C
I
It
may
be
evaluated
to
be
false,
so
we
can
raise
the
e-say
exception
if
the
expanse
evaluator
is
true
and
we
can
jump
to
the
easy
action
execution
triggers
appropriate
action
to
be
executed.
If
the
action
is
executed,
you
know
not
successfully,
we
will
raise
the
easy
exception
again.
If
it
can
be
executed
successfully,
we
we
may
trigger
to
send
the
event
notification
to
the
to
the
remote
client.
I
So
for
smart
field
cases
you
can
see.
Actually
we
introduced
two
policy
variable.
We
can
use
the
event
repeater
counter
to
keep
track
of
the
the
the
the
duplicated
event
the
occurrence
time-
and
we
also
you
know,
define
the
interface
statics
event
and
to
to
keep
track
of
the
status
down
interface.
I
So
also
we
can
map
to
this
condition.
We
can
enumerate
all
the
condition
and
we
can
design
the
easy
logic
in
the
left
and
we
can
make
it
a
worker
actually
next.
I
I
don't
want
to,
you
know,
repeat
all
these
issues,
because
I
already
post
all
these
four
issues
on
the
list.
I
haven't
got
feedback
and
actually
all
these
issues
we
already
you
know
addressed
in
this
draft
and
we
want
to
hear
your
feedback
and
can
you
go
back?
Can
you
move
to
to
the
last
slide.
C
Yeah
yeah
am
I
on
the
right
side
now.
I
Yeah
yeah
correct,
maybe
move
to
the
last
slide,
because
all
these
four
issues
I
already
raised
on
the
list
haven't
get
feedback.
Maybe
yeah
people
can
take
a
look
at
that,
and
and
so
we
will
keep
on
to
improve
this
job,
and
I
want
to
hear
your
feedback.
C
Okay,
so
as
chair
we're
pretty
much
at
the
end
of
the
meeting
now,
but
you
said
that
you
wanted
to
take
some
things
to
the
list
and
should
that
be
done
prior
to
call
for
adoption.
I
Yeah
just
to
make
sure
everybody
you
know,
raise
this
open
issue,
and
actually
we
already
addressed
this
urban
issue.
This
just
gave
the
update.
I
think
we.
We
also
believe
it's
ready
to
go
for
adoption
and
if
anybody
can
raise
addiction
comments,
we
are
welcome.
C
C
C
C
Okay,
I
say
this
because
it
would
be
this.
These
two
drafts
have
both
been
up
for
adoption
before
and
I
think
that
there's
I
don't
know
if
it's
an
official
or
whatever
rule,
but
there's
no
rule,
but
you
know
it's
kind
of
the
case
that
if
something's
come
up
to
the
bat
multiple
times
and
it
never,
you
know,
there's
no
interest,
then
it
gets
dropped.
So
it
may
be
better
to
try
to
resolve
any.
C
I
I
The
the
concern
raised
actually
by
the
eugene
I
did
check
the
you
can
actually
about
previous
job
there
and
he
seems
okay
for
the
second
job,
that
we
work
together
with
andy
and,
I
think,
probably
yeah.
We
we
can
discuss
with
andy.
If
andy
has
some
comments
on
these.
C
Okay,
then
I
think
then,
what
I'm
hearing
you
say
is
that
you'd
like
to
have
the
adoption
call
initiated
for
these
two
drafts
that
all
the
issues
have
been
addressed,
and
so
the
chairs
can
initiate
that
and
we'll
see
how
it
goes.
C
Okay,
so
we're
a
little
bit
over
time,
but
it
concludes
our
meeting.
Thank
you.
Everyone
for
joining
in
the
comments
shared
we'll
be
posting
minutes
as
soon
as
we
have
the
time
to
do
that
and
follow
up
on
the
actions
that
were
discussed
during
this
meeting.