►
From YouTube: IETF105-DETNET-20190724-1550
Description
DETNET meeting session at IETF105
2019/07/24 1550
https://datatracker.ietf.org/meeting/105/proceedings/
A
We
are
our
first
speaker
up
just
walked
in
the
room,
so
it's
perfect
timing,
I'm,
Lou
burger.
This
is
young,
far
coughs,
my
co-chair.
We
have
Ethan
Grossman
still
remote
for
the,
as
is
the
case
for
all
IETF
sessions.
Our
work
here
is
covered
by
our
contribution
guidelines.
That
is
summarized
in
the
note.
Well,
if
you're
unfamiliar
with
it,
please
visit
the
URL
on
the
bottom
of
the
page.
A
Basically,
everything
you
say
here
becomes
part
of
our
permanent
record
and
in
fact
we
are
recording
with
video
as
well
as
transmitting
we
and
we
have
collaborative
note-taking
taking
place
on
etherpad.
Please
feel
free
to
join
in
make
sure
your
comments.
Your
name
is
appropriately
captured
and
I'll
do
my
best
to
relay
comments
made
by
Jabbar
with
that
we're
gonna
jump
into
the
agenda.
First
up
is
Greg
mirskiy.
E
E
E
They
deplane
deterministic
networks
did
with
plane
and
am
so.
We
had
comments
from
Pascal.
We
addressed
them
and
discussed
and
we
took
their
death
net
or
a.m.
document
was
splitted
in
two
and
because
of
changes
in
or
how
the
impellers
data
p.m.
document
progresses.
We
removed
the
optional
years
of
Gail
gash,
so
it
uses
only
to
the
Royal
VC
C
D.
So.
E
E
To
take
control
word
ACH
from
to
the
wire
and
use
the
reserved
field
for
the
sequence
number,
so
we're
defining
that
associated
channel
header,
which
the
first
label
is
one.
Then
the
version,
the
current
version
is
zero.
The
sequence
number
is
a
food
that
would
be
monotonically,
increment
it
and
the
channel
type.
The
channel
that
we
use
from
Ayana
to
the
wire
associated
channel
called
points.
E
E
And
you
can
see
that,
basically,
with
this
sequence
number,
if
we
have
taken
to
replication
in
n1,
+
r4,
then
the
packet
that
has
the
prime
science.
So
then
they
could
go
and
elimination
in
r2
will
choose.
For
example,
the
original
packet
and
all
prime
will
be
dropped,
and
when
the
packet
elimination
function
in
n2
will
probably
choose
copy'
armed
with
two
primes:
that's
because
that
net
om
will
have
a
sequence
number,
so
this
packets
can
be
compared
and
acted
functions
will
work
on
them
as
they
work
on
adjust
net
net
flow.
E
So
what's
the
next
step,
because
there
are
documents
that
consider
of
how
MPLS
works
over
all
with
IP
UDP
tunnel
and
where
the
TSN
domain
will
continue.
This
work
to
address
specific
what
what
specifics
can
be
for
this
scenarios.
We
welcome
your
comments,
suggestions
and
questions
always
and
very
grateful
for
them,
and
we
would
like
to
ask
for
consideration
to
consider
of
working
reproduction
of
this
document.
E
A
Oh,
that's
really
funny.
I
was
looking
to
Stuart
and
then
looking
for
Andy
in
the
room
of
course,
and
he's
not
in
the
room
and
I
looked
down
and
yes
he's
at
the
my
queue.
So
I
will
read
his
question
question
for
Greg.
This
is
from
anomalous.
The
graph
needs
to
specify
if
the
sequence
number
goes
through
zero
or
not,
for
example,
255
0
1,
255
1.
This
is
currently
not
specified.
So
a
technical
comment
right.
B
E
A
A
So
the
next
question
was
how
many
read
we
already
saw
that
there's
very
few
III,
don't
think
we
can
do
much
right
now
with
the
amount
of
people
that
we've
had
read.
The
document
so
really
like
to
ask
people
to
to
read
the
document
comment
on
the
list.
Think
of
it
talk
about
whether
you
think
it's
an
headed
in
the
right
direction
or
the
wrong
direction.
That'd
be
really
useful.
F
A
That
that's
an
interesting
idea.
The
other
thing
and
I
haven't
talked
to
Janos
about
it.
So
we
got
a
synchronize
on
this
point.
You
know
one
of
the
things
I'm
I
was
thinking
is,
is
we
could
do
an
adoption
poll
with
an
understanding
that
if
it
fails
it's
not
that
it's
dead
so
normally
when
we
do
an
adoption
poll,
if
it's
rejected,
that
topic
is
closed
and
we
don't
revisit
it,
we
could
do
an
adoption
poll
which
is
has
with
an
understanding
that
it's
it's
really
about.
A
C
A
E
A
E
E
B
So
so
only
the
current
about
congruent
routing
stuff
we
had
we
had
in
the
flow
information
whether
we
come
so
far
to
the
two
directions
to
becoming
congruent
but
and
and
introducing
parameters
for
that
like
flow
ID
management
parameters
like
floor
ID
and
so
on,
we
may
consider
extending
it
to
the
VIP
and
maybe
give
a
different
name
or
something
to
make
it
sure
or
use
those
that
the
OEM
flow
goes
along.
The
same
way
as
it's.
H
More
than
just
the
same
wake
a
David
thereby
can
concrete
example
suppose
that
the
debt
net
40
plane
is
fouled
up
at
a
node,
but
the
IP
forwarding
plane
is
not
the
OEM
traffic.
Bfd
cannot
be
betrayed.
Traceroute
in
particular,
has
got
to
go
through
the
detonate
forwarding
plane
to
tell
you
where
the
broken
node
is
because
you
send
it
through
the
IP
forwarding
plane
will
incorrectly
tell
you
everything's,
okay,
I.
H
Part
of
forwarding
plane,
yeah,
yeah
yeah-
that's
that's
that
that's
absolute
foreign
plane,
so
in
queue
speak
if
the
debt
net
queue
is
broken
in
a
node
and
the
corresponding
queue
for
that
pair
of
IP
addresses
is
functioning
and
you
send
traceroute
in
a
fashion
that
doesn't
get
into
the
decnet
queue.
Traceroute
will
fail
to
fight
will
fail
to
find
the
problem.
G
G
G
So
there's
an
oem
flow
for
debt
net
flow,
so
we
have
to
be
able
to
look
at
the
OEM
packet
and
say
that
goes
with
this
debt
net
flow
and
this
back,
and
so
it
gets
this
debt
net
flow
service
and
and
so
on,
and
the
other
part
of
that
is
that
we
have
to
allow
for
the
bandwidth
of
the
OEM
in
the
reservation
for
the
flow,
which
means
we
have
to
be
very,
very
sure
that
the
OEM
is
limited
in
bandwidth
and
you
know
conforms
to
whatever
rules
we
have
for
for
the
flow
itself.
G
G
With
the
packet
replication,
for
example,
you
can
argue
that
the
sequence
number
allows
the
data
flow
itself
to
be
the
OEM.
Now
you
don't
get
all
the
functions?
Okay,
but
you
certainly
get
connectivity
and
if
you
have
two
paths
and
I'm
receiving
it
on
one
path
and
I'm
not
receiving
it
on
the
other
path.
I
know
something's
wrong
and
it's
not
just
that
the
source
stopped
so
just
know.
I
understand
this
is
just
IP.
Sorry,
sorry,
that's
only
if
you
can
that's
only
if
you
have
the
two
paths
yeah.
Thank
you
right
again.
I.
E
They're
some
advantages,
but
in
my
experience
it's
a
good
complementary
to
it
does
not
substitute
for
active
for
am
where
the
packets
are
specially
specifically
constructed
for
the
test
purposes,
and
there
are
many
cases
for
now.
Ip
does,
or
at
least
appears
to
have
some
issues
that
we
need
to
discuss
and
define
and
faith
sharing
is
reported
so
also
because
there
are
some
the
number
of
documents
that
explain
how
that
let
IP
works
over
MPLS
and/or
with
TCM.
E
So
what
we
point
out
is
that
there
are
two
scenarios
to
work
in
regard
to
our
OEM,
so
first
its
if
it's
a
tunneling
and
second
one,
it's
it's
appearing,
you
can
see
inter
working.
So
we
might
have
just
look
at
TSM
or
MPLS
that
let
and
know
as
the
means
as
one
hop
or
we
might
need
to
inter
work
to
provide
end-to-end
OAM,
but
that
bit.
F
So
there
is
another
technology,
I
mean
I,
don't
really
see
how
we
solve
it
in
classical
IP
without
some
help,
somewhere
or
other,
because
we've
never
been
able
to
solve
the
problem
of
making
the
paths
congruent
with
with
IP.
But
what
we
could
do
and
what
we
probably
need
to
do
anyway
is
to
introduce
an
explicit
routing
technology
of
which
there
are
several
candidates.
F
A
I
miss
through
it
that
we've
actually
been
here
before
with
MPLS
and,
as
you
may
remember,
we
talked
about
using
sort
of
companion
labels
that
were
Co
routed
and
had
shared
resources,
and
we
talked
about
in
that
in
that
context
that
that
still
wasn't
adequate
because
you
might
have
something
in
your
your
label
database
get
corrupted.
So
just
because
you
use
the
same
forwarding
identifier
or
you
use
different
sorry,
you
use
the
same
resources
and
the
same
paths,
because
you
use
different
traffic
classification.
A
A
F
H
F
A
E
A
A
A
Forwarding
behavior
is
different
than
normal
IP,
and
so
we
definitely
have
a
different
problem
and
we
also
have
different
techniques
available
to
us
for,
for
example,
part
of
our
architecture
is
that
all
of
our
IP
six
tupled
flows
are
explicitly
controlled,
and
so
we
can.
We
can
push
some
things
on
the
control
plane
and
place
additional
requirements
on
the
control
plane.
That
doesn't,
of
course,
answer
what
we're
doing
in
the
data
plane,
but
we
we
have
some
different
tools.
Sure.
F
E
G
Norvin
I'm
very
much
interested
in
I'm,
more
interested
in
the
the
plain
IP
o
am
then
I
am
in
MPLS,
because
I
feel
like
the
MPLS.
Oem
is
not
very
hard,
so
this
is
a
lot
more
fun
problem.
So,
if,
if
so,
if
somebody
comes
up
with
a
plausible
solution,
I
want
to
hear
about
it,
but
so
I'm
just.
A
Waiting
to
hear
a
plausible
solution
so
at
this
point
like
to
suggest
that
the
folks
who
are
interested
in
working
on
it,
self-organized,
yeah
and
and
let's
see
what
they
come
up
with,
so
the
greg
has
nominated
yourself
by
writing
this
document,
okay
to
coordinate
the
activity,
but
let's
see
if
we
can
get
some
critical
mass
of
thinking
on
it
and
just
like
on
the
being
discussion.
If
you
want
to
have
a
a
the
working
group
WebEx
to
have
informal
meetings,
you
can
do
that.
A
That
book
is
saying
we
will
just
contact
the
chairs
and
we'll
set
you
up.
The
other
thing
is
is
if
you
would
like
to
ask
for
an
interim
on
this,
we
can
also
do
that,
but
we'd
like
to
ask
first
to
do
the
self
organizing
thing
and
let
us
know
how
we,
the
chairs,
can
support
the
working
group.
Thank
you
and
with
that
you're.
Actually,
at
a
time.
I
Here
is
the
background.
Actually,
although
that,
then
that
control
all
controller
plan
is
not
in
the
chatter
of
the
current
and
networking
group,
but
this
part
of
work
is
not
totally
out
of
the
sight
of
the
net,
because
in
then
at
the
architecture
there
is
a
section
about
the
controller
plan,
a
corresponding
to
the
aggregation
of
the
control
and
management
in
fc7
4:26
and
also
in
the
tenant
data
plan
draft.
We
can
see
in
the
previous
presentations
that
there
are
a
lot
of
controller
plan
considerations
included
in
the
current
draft.
I
So
we
are
not
this
working.
We
are
not
zero.
We
have
something,
but
the
purpose
of
this
draft
is
that
we
want
to
summary
all
these
efforts
about
general
controller
plan
and
also
the
requirement
in
one
place,
and
we
want
to
provide
an
overview
of
the
possible
solutions
for
the
net
control
plan,
especially
in
the
architecture.
It
can
be
a
distributed
system
with
signaling
protocol.
I
It
can
be
a
fully
centralized
the
controller
weather
controller
as
the
in
system,
or
it
can
be
a
hybrid
control
plan,
a
signal
signaling
protocol
and
also
have
a
controller
they
work
together.
This
also
aligns
with
a
work
in
actuallyI.
The
draft
QC
c
is
also
have
the
similar
structure,
these
three
class
classes,
and
also
we
want
to
discuss
some
control
plant
issues
that
are
unique
to
that.
I
For
example,
we
have
to
build
up
explicit
pathway,
resource
reservation,
or
we
have
to
build
up
explicit
paths
with
PR
yo
F,
and
we
have
also
some
discussions
about
done
as
integration
with
the
existing
control
plans.
For
example,
the
RMP
is
IP
or
as
our
seminal
routing
things
so
and
also
a
Greg
have
some
very
great
work
about
this
part.
The
management
plan
issues
unique
to
ten.
That
should
also
be
considered.
I
I
So
we
are
wondering
whether
the
controller
plan
solution
is
also
needed
in
the
working
group
and
though
whether
the
working
group
is
going
to
reach
had
her
about
this
part
of
work
to
make
this
discussions
happen
in
the
working
group,
and
so
we,
of
course,
the
comments,
review
or
questions
are
welcome
and
also
we
want
some
feedback
from
the
working
group.
What
will
what
should
be
done?
Next?
That's
all
so.
A
From
a
I
just
looked
at
the
architect
that
they
sorry
the
Charter
again
to
make
sure
that
what
I
was
going
to
say
is
right,
I
think
it
is,
and
I'll
also
look
to
Deborah
re
D
to
respond
to
this-
maybe
not
right
now,
but
afterwards
after
she
takes
a
look
at
the
Charter
again
and
thanks
about
it.
But
as
I
read
the
Charter,
we
are
certainly
not
chartered
to
do
control
playing
solutions.
A
I
A
And
identifying
the
possibilities
and
the
possible
solutions,
sorry,
the
possible
ways
you
know
distributed
or
centralized
in
multiple
ways
of
controlling
debt
net
I
read
our
charter
is
allowing
that
in
in
terms
of
even
identifying
gaps
in
control,
plane,
solutions,
I
think
is,
is
within
our
charter,
at
least
as
I.
Read
it
and
I'm
I'm,
making
sure
our
ad
is
listening.
So
she
can.
She
can
tell
me
I'm
wrong
now.
If
we
get
to
the
point
of
saying
we've
identified
this
and
here's
our
solution,
I
think
we've
gone
too
far.
A
What
we
would
need
to
do
is
go
to
the
right
working
groups
and
talk
to
those
group
groups
about
the
solutions.
So
in
terms
of
talking
about
the
part
of
the
solution
space,
the
architecture
for
control-
that's
certainly
there
talking
about
what
specific
technologies
are
available,
I
think
I
think
we're
on
safe
ground
and
you're
doing
that
so
I
think
most
of
this
document
were
in
good
in
okay
place.
To
have
that
discussion.
There
are
definitely
places
where
you're
saying,
let's
solve
it.
This
way
that
yeah
that's
too
far.
Okay,.
F
A
Have
we
have
three
or
four
working
groups
who
have
a
charter
of
doing
control,
playing
an
outing
area,
so
I
think
we
can
probably
ask
them,
say
you
own
this
protocol,
we
see
a
gap,
do
you
want
to
fill
it
and
it
certainly
if
they
come
back
and
say
no
I
think
we
have
a
really
strong
reason
to
go
to
the
isg
and
say
we
should
change
our
charter
to
allow
us
to
do
it.
Okay,.
A
F
A
A
So
one
of
the
interesting
things
that
happened
or
earlier
this
week
in
tease
is
there
was
a
proposal
for
doing
traffic
engineering
for
IP
flows
and
I
asked
about
bringing
in
supporting
the
debt
net
IPE
flow,
and
the
authors
were
receptive.
So
we
may
actually
end
up
with
a
solution
that
we
can
use
before.
We've
even
talked
about
the
framework
for
it,
so.
J
J
Correct
in
scope,
its
its
scope
is
98%
correct.
It's
not
it's
more
to
do
to
the
document,
but
it's
a
good
document.
The
the
problem
I
have
here
is
when
Lou
says
we
should
give
it
to
those
are
the
working
groups.
What
is
the
is
it?
The
debt
networking
group
shares
that
actually
going
to
take
that
initiative
or
not.
If.
A
A
Not
quite
there,
yet
if,
if
now
we're
at
the
point
where
people
are
bringing
contributions
of
control
playing
solution
documents,
we
can
have
that
the
discussion
in
specific,
rather
than
abstract,
by
the
way
Andy
malice,
had
one
comment
in
response
to
I
think
what
I
was
talking
about
earlier
with
the
Charter
his
feeling
was,
he
was
being
conservative
in
the
reading
of
the
Charter.
Were
it
since
control
playing
solutions
were
not
there
I
guess
he
was
reading
it
as
we
couldn't
even
talk
about
control
plane,
which
you
know
that
the
community,
the
just
to
be.
A
You
know
clear.
You
start
out
and
talk
about
the
second
line
in
the
Charter.
It
says
the
work
encompasses:
the
data
playing
the
data
playing
OAM
time,
synchronization
management
control
and
security
aspects,
referring
of
debt
net,
so
control
management
controls
what
we've
called
the
controller
plane.
So
you
know
from
the
architecture
level
from
the
high
level
that
is
certainly
in
scope,
so
I'm.
J
Still
a
little
bit
concerned
that
you're
actually
pushing
it
way
too
out
I'll
see
the
two
sections
that
you
pointed
out
that
were
given
too
far.
They
are
an
indication
that
this
is
already
happening
now,
so
it
should
be
proactive
and
take
that
decision.
Take
that
action
now
to
start
discussing
within
the
working
group
that
we
actually
need
to
be
involved.
I
I.
A
Think
the
next
step
is
to
get
some
proposed,
some
proposals
on
the
table
around
rather
than
talk
and
theoretical
x'.
Let's
see
the
documents
and
figure
out
where,
where
they
belong.
So
if
someone
has
a
control,
plane
solution,
they've
been
hanging
back,
holding
back
on,
bring
it
forward
and
then
we'll
figure
out
where
it
where
it
goes,
contribute
it
somewhere
could
bring
it
into
the
IETF
and
if
it's
in
the
wrong
working
group
we'll
get
into
the
right
working
group.
As
you
said,
let's,
let's
not,
let's
not
hold
up
on
process.
K
So
Erik
gray,
Erickson
I'm
gonna
point
out
the
elephant
that
we're
all
dancing
around
lose
already
kind
of
hinted
at
it.
We
we
could
possibly
have
this
work
already
beginning
to
start
in
the
TS
working
group.
That's
Lou,
talking
to
Luke,
so
I
mean
really.
This
is
not
as
complicated
as
we
seem
to
be
trying
to
make.
A
A
Does
like
to
gauge
the
feeling
of
the
working
group
on
whether
the
working
group
should
be
spending
time
on
the
architecture
and
framework
level
discussions
of
the
controller
plane.
So
again,
this
is
something
that's
in
scope,
revert
Charter!
Do
we
want
to
spend
time
on
it
who's
interested
in
doing
that
I'm,
not
even
to
look
down
to
see
if
jabber
for
Andy's
hand
is
up
I'm
sure
it
is
so,
should
we
be
working
on
the
topic
of
controller
plane,
it's
a
similar
number
to
the
previous
one
and
a
similar
comment.
A
A
Worry
about
charter.
We
worry
about
the
technology
in
this
document
feel
free
to
talk
about
what
what
is
Miss
Singh
mm-hmm,
don't
put
solutions
in
this
document.
Okay,
do
put
solution
if
you
have
a
solution
to
brain
put
it
in
another
document,
okay,
contribute
it
to
one
of
the
working
groups
that
you
think
is
appropriate
and
the
chairs
will
figure
out
where
it
goes
and
between
the
ad
will
figure
out
where
it
goes.
The
biggest
thing
is
contribute
it
bring
it
to
the
table.
Let's
get
the
discussions
going.
I
I
A
We've
had
other
documents
in
other
work
groups,
I've
even
written
some,
which
go
through
and
say
here's
the
set
of
things
we
need
from
the
control
plane
and
here's
what's
missing,
so
it
almost
becomes
a
gap
analysis
document.
Those
are
very
useful
documents.
Ok,
so
there's
nothing
and
you're
doing
that
you're
talking
and
talking
about
what
the
options
are
and
where
the
gaps
are
that
helps
guide
the
work
that
gets
done
so.
I
A
L
I'm
leon
from
china,
mobile
and
one
of
the
author
of
the
ring
topology
draft
and
today
I'm,
going
to
present
as
the
author
and
with
the
update
from
last
night's
meeting.
So
the
update
in
this
fourth
zero
full
version
mainly
is
the
alignment
of
this
draft
OS,
the
architectural
draft.
So
basically,
we
use
the
terms
that
are
defined
in
internet
architecture
so
and
it's
also
support
the
packet
world
of
function
and
use
the
defined
F
label,
s
label,
service
level
and
the
control
world
in
data
net
architecture.
L
The
presentation
will
be
very
short
because
I'm
going
to
only
talk
about
the
updates,
so
the
first
one
is
in
the
single
ring
situation,
where
you
have
a
point-to-multipoint
detonate
service.
The
document
basically
show
in
the
ingress
and
egress
a
node
of
the
Ring.
What
type
of
function
in
terms
of
the
replication,
elimination
and
reordering
is
needed.
So
first
for
this
particular
example,
and
in
the
ingress
node,
which
is
a
in
here,
you
will
need
to
support
the
nodes
of
support
the
replication
function
and
for
for
the
leaf
node
in
C,
D
and
F.
L
L
On
the
right
hand,
side
on
the
counterclockwise
wrote
and
the
other
mechanism
are
all
the
same
for
single
ring
for
the
packet
on
the
rain,
either
on
the
left,
hand,
side
or
ring
on
the
right
hand,
side
ray
yep.
So
summary,
the
ring
topology
is
very
important
for
for
long-haul
transmission,
basically
we're
going
to
see,
and
these
type
of
ring
topology
very
frequently
in
a
network
for
long
haul.
L
So
we
think
it
is
very
important
for
data
net
up
to
four
week
or
for
people
to
understand
and
to
know
how
we
should
configure
the
ring
topology
for
detonates
services
and
basically
rain
topology.
Intrinsically
give
you
the
protection
with
clockwise
and
counterclockwise
route
at
the
same
time
and
and
it's
more
efficient,
because
it's
only
two
copies
of
the
packets
to
be
sent
and
to
guarantee
your
performance,
and
we
think
after
last
meetings,
discussion
and
the
revised
version
in
this
couple
of
months
effort.
L
B
A
question
for
verification.
So
what
makes
this
so
specific
in
the
sense
that
we
now
have
the
emptyness
data
plane
that
describes
engine
generic
I
mean
the
data
template
opinion,
generic
brillo
fans,
control,
hood
and
everything,
and
we
have
a
number
of
solutions
for
ring.
Topologies
of
work
is
going
on
currently
as
well
so
I
mean.
L
A
We
have
two
layers
right:
you
have
the
forty
layer
and
the
service
layer
and
the
free
off
lives
at
the
service
layer.
Have
you
looked
at
taking
advantage
of
the
ring?
The
existing
MPLS
ring
solutions
at
the
forwarding
layer
have
have
we
looked
at
using
the
existing
ring
solutions
as
solutions
at
the
forwarding
layer
right?
That's
a
good
question.
So.
J
You
know
I
know
I,
understand
if
I
read
this
correctly
and
it's
it
I
haven't
read
the
previous
version
of
the
draft.
I
haven't
read
the
new
one,
but
I
think
this
is
are
more
for
the
forwarding
layer
and
then
you
need
to
add
the
application
and
elimination
and
reordering
for
the
service
layer.
So
you
shouldn't
specify
anything
new
for
the
the
forbidding
layer.
You
should
use
our
mrs
at
this.
J
L
J
So
one
of
the
advantages
of
our
Mar
2
to
it
does
this:
you
know
by
advertising
stuff
into
ice,
ice
or
SPF,
and
it
determines
this
automatically
versus
needing
any
configuration.
The
other
thing
is
in
this
two
ring
case.
You
said
the
upper
node
and
the
lower
node,
which
makes
sense,
but
you
need
to
do
this
in
terms
of
you
know.
J
In
this
case,
the
upper
node
is
i1
if
you're
going
clockwise
and
for
if
you're
going
counterclockwise
it's
I
to
you
need
to
have
this
either
be
algorithmically
done
or
again,
you're
going
to
use
configuration.
The
whole
thing
mode
rings
is:
if
you're
going
to
have
lots
of
rings,
making
sure
you
configure
all
the
Rings
correctly
can
be
a
big
problem,
so
the
more
you
can
make
the
Rings
do
this
either
algorithmically
or
by
discovery
or
something
the
better.
You
are
in
terms
of
making
this
a
scalable
solution
from
a
management
point
of
view.
J
A
N
From
a
tree,
the
there
are
many
there
are
many
ring
protection
mechanism
for
various
transport
network
technologies,
including
MPLS
ARIMA,
on
those
photos,
rim,
protection
mechanisms.
We
cannot
avoid
traffic
disruption
in
case
of
data
failure,
but
with
this
the
net
ring,
then
the
ring
technology
here
for
we
can.
We
can
I
mean
in
the
event
of
NATO
failure.
We
we
can
avoid
the
traffic
disruption.
I
mean
that
we
cannot,
we
don't
lose
any
packet.
So,
yes,
that's
the
advantage
of
this.
This
document,
yes.
B
B
Technologies
and
as
for
the
P
of
I,
wanted
to
also
mention
that
what
you
can
find
is
we
call
it
a
load
or
topology,
which
is
comprised
of
things,
and
there
were
in
802
dot
on
CB.
What
example
is
explained
in
detail
where
to
place
application
of
elimination
points
and
so
on
how
to
leverage
that
affect
their
artifact
of
topology.
So
it's
a
combination
of
the
series,
informally.
N
Right
exactly
I
mean
this
I
mean
I,
heard
a
question
about
how
you
kind
of
use
a
conventional
reading
protection
mechanism
with
is
that
net
technology
I
need
the
net
are
in
protection.
My
answer
is
that
or
when
you
use
this
the
net
type
of
protection
mechanism,
you
disable
that
conventional
protection
mechanism
in
fording
plan
not
for
the
fourth,
so.
N
A
I
think
you're,
suggesting
that
there
there's
a
room
for
a
pre
off
Oh,
Optima,
sorry
and
optimized
pre
off
aware
ring
protection
scheme,
and
you
know
that
that
may
be
the
case
and
I
think
you
need
to
articulate
it
a
little
bit
better
and
in
particular,
if
talk
about
how
you
want
to
take
advantage
of
our
mr
or
why
you
think
you
can't
take
advantage
of
our.
Mr,
I
think
that's
a
good
discussion.
C
A
Because
we've
we've
worked
since
we
it's
a
technology.
We've
worked
on.
We
want
to
make
sure
that
if
we
can
leverage
it,
we
can't
we
can
and
if
we
can't-
and
it
makes
sense
to
do
a
pre
off
aware
ring
solution.
Well,
we
can
go.
We
can
spend,
go
to
a
bit
deeper
on
that
one
of
the
things
that
of
course,
will
come
up
is
how
to
skit
make
that
scalable.
A
Where
that
your
ring
protection
scheme
is
where
each
each
ring
you
don't
have
to
have
the
same
number
of
pre-op
functions
as
you
have
in
hand,
if
there's
a
way
to
do
that,
if
there's
some
way
to
do
some
shared
protection,
who
do
you
if
your
only
solution,
if
your
solution
is,
is
that
at
every
basically
every
ring
becomes
a
segment
and
you
do
free
off
within
that
segment
for
every
flow?
That's
that's
gonna,
be
a
very
expensive
solution,
so
maybe
there's
an
opportunity
to
do
something
better.
A
C
J
Guess
my
my
community
compared
my
my
comment
is
the
way
that
they're
using
this
in
the
ring
since
you're
multitasking
on
both
both
paths-
you're
not
doing
protection,
really,
what
you're
saying
is:
if
a
link
fails,
you
still
get
the
packet
become
going
the
other
way
around,
but
armor
has
two
parts
to
it.
One
is
the
self-configuring
of
rings
and
the
others
action
I
think
you
don't
need
the
protection
part
here,
but
you
do
need
the
self-configuring
to
make
it
scalable
from
a
management
point
of
view.
J
Laura
Anderson
I
got
up
here
because
I
thought
Lou
was
a
little
bit
too
vague.
When
he
said
you
take
a
look
at
the
forum
or
well,
yes,
that
you
came
back
and
said
that
they
actually
have
to
show
that
they
can't
use
on
Rory
but
I
think
that's
the
case.
We
need
to
look
at
or
an
hour
and
see
if
there
is
a
synergy
here
and
if
we
can
use
it,
we
should
use
it
in.
A
General
we
try
not
I
mean
I,
know
some
groups
don't
follow
this,
but
in
general
we
try
not
to
have
two
solutions
for
exactly
the
same
problem.
So
that's
all
I'm,
saying
I'm
not
saying
anything
more
than
let's
follow
our
standard
practice
of.
If
there's
an
existing
solution.
That
is
usable,
let's
use
it.
N
Okay,
jungle,
you
again
about
self
auto,
auto
configuration
I
mean
they
here.
The
example
is
4p2
MP
multicast
the
services
and
the
the
bridge
node
over
or
to
know
that
h22
to
each
leaf
is
aware
of
that.
I
mean
that
I
have
I,
have
a
subscriber
so
there
that
there
can.
In
my
understanding,
that
is
done
by
any
any
multicast
a
protocol
but
case
configuration
protocol
and
then
maybe
that
once
subscriber
is
attached
EDA
and
then
that
can
get
the
information
I
mean
a
signal
and
then
enable
packet
replication
functionality.
A
L
A
And-
and
we
look
forward
to
seeing
the
update
and
wherever
you
end
up,
based
on
your
your
exploration
of
the
RMR
yeah,
what
you
can
use
what
you
can
use
yeah
all
right
with
with
that
we're
actually
over
time
we
managed
to
use
all
our
time,
even
though
we
had
hoped
we'd
have
some
additional
conversation
on
SR
v6
and
bounded
latency
requirements.
If
you
have
comments
on
those,
please
take
them
to
the
list
and
continue
the
discussion
there
and
thank
you
very
much
and
we
will
see
you
online
and
then
in
Singapore.