►
From YouTube: IETF101-RTGAREA-20180322-1550
Description
RTGAREA meeting session at IETF101
2018/03/22 1550
https://datatracker.ietf.org/meeting/101/proceedings/
A
D
C
Well,
please
update
it.
It
pretty
much.
Does
the
same
thing
as
before.
The
difference
is
that
now
we
made
it
I
think
simpler
and
we
point
to
all
the
DCPS
down
at
the
bottom.
So
it's
a
lot
simpler
for
people
to
go,
find
other
things
to
refer
to,
there's
also
a
lot
of
other
information
about
the
area.
The
agenda,
the
routing
area
general
list,
which
is
the
rather
discussion
list,
etc.
C
C
The
quality
of
documents
is,
of
course,
proportional
to
the
reviews
that
we
get
hopefully,
and
we
want
to
invite
everyone
to
to
do
that,
review
the
documents
in
your
working
group
for
the
people
in
the
routing
Directorate.
Thank
you
so
much
for
all
the
work
that
you
do
for
all
the
reviews.
If
you
do
as
well
in
helping
us
getting
higher
document
quality,
we're
always
looking
for
feedback.
What
are
we
doing?
Well,
what
is
working
well
with
the
area
or
not
any
comments
that
you
have
about
us
or
about
the
area
in
general.
E
C
C
There
are
some
updates
to
the
area
that
we,
of
course
want
to
share
with
you
guys,
then
we're
going
to
have
a
talk
on
the
term
mystic
networking
the
chair
from
the
that
networking
group
are
going
to
present
that
and
then
the
always
exciting
and
never
boring
working
group
and
Boff
reports.
We
didn't
have
a
Boff
this
time,
but
you
know
all
the
working
groups
and
the
run
Directorate
report
that
John
and
Amy
put
together
and
run
all
the
time
and
then
some
open
discussion
at
the
end
that
we
have
we
got
time.
C
Okay,
I
do
actually
so
we
have
some
time
issues
and
we're
going
to
change
the
agenda
little
bit,
we're
going
to
move
the
deterministic
networking
first,
so
we're
going
to
run
that
first
and
then
we're
gonna
come
back
to
the
rest
of
the
administrivia
and
the
rest
of
the
of
the
agenda.
Anyone
opposes
to
that
no
good,
because
that's
what
we're
into
so
Janos.
G
F
So
a
bit
about
what
is
that
net
so
that
net
aims
to
deliver
guaranteed
delivery
in
terms
of
both
timing
and
and
loss?
I
mean
reliability
aspect,
so
deterministic
means
that
the
bucket
delivery
is
deterministic
in
time
and
we
are,
after
a
very
low
delay
and
delay
variation
values
and
extremely
loss,
low
loss
values.
F
F
F
That's
one
of
the
candidates,
industrial
automation,
I,
think
it's
very
important
industry
4.0
in
a
factory
floor,
there's
a
change
in
in
networking
and
they
are
looking
at
higher
early
as
well,
so
that
not
I
think
has
a
role,
power
generation
as
part
of
our
generation
of
in
wind
farms,
building,
automation
and
so
on.
So
this
is
just
a
and
Willa
station
of
the
use
cases
there's
a
link
in
the
bottom
of
the
slide
to
the
draft.
If
you
are
interested
in
more
details.
F
What
is
the
difference
between
what
is
deterministic
in
terms
of
what
is
the
difference
between
a
traditional
service
and
a
deterministic
service?
In
case
of
adolescent
and
traditional
service?
We
are
having
a
kind
of
elastic
traffic,
and
while
we
are
not
working
guys,
we
are
not
too
eccentric
in
the
sense
that
we
want
to
have
a
high
utilization
and
maximum
throughput
and
so
on.
And
of
course,
we
have
been
always
interested
in
a
good
latency,
providing
good
delay
values,
but
not
the
extreme
ones,
as
illustrated
in
the
bottom.
F
The
delay
curves
typically
have
have
a
tail,
and
if
you,
if
you
introduce
a
bound
into
this
delay,
ground
delay
curve,
it
means
that
we
do
lose
packets
at
the
tail
of
the
of
the
curve
or
we
need
to
over
over-provision.
So
it's
not
efficient
network
utilization
as
opposed
to
it.
On
the
right
hand,
side,
the
deterministic
service
characteristics
are
discrete
and
described
as
I
imagined.
F
B
F
F
Demo,
which,
unfortunately
got
broken
during
the
dry
run,
why
we
discussed
how
to
project
it.
It
just
was
running
for
one
minute
empty
and
many
things,
good
broken
like
the
power
and
so
on.
So
if
it's
demo
day
mony,
if
some
small
piece
gets
broken,
then
it's
chain
so
I'm
a
can.
We
show
something:
okay,
I
still
keep
talking,
explain
what
it
is.
We
will
distribute
the
video
link
to
the
demo.
F
It
has
been
shown
a
month
ago
or
two
months
ago,
at
a
ninety
two
workshop
with
Article
II,
so
there's
a
video
recording
it
of
it
and
some
part
of
it
actually
has
been
shown
in
Prague.
So
I
can
explain
what
what
was
that
so
the
demo
set
up.
What
we
have
here
is
there.
Is
this
special
display,
this
rotating
display,
which
provides
the
persistence
of
vision,
a
kind
of
display
it's
a
columnar
flat,
and
it
rotates
at
a
high
rpm,
like
five
thousand,
which
gives
you
the
the
display
we,
as
illustrated
in
the
bottom.
F
We
have.
The
content
is
remote,
it
can
be
in
a
cloud
and
we
have
a
network
in
between.
So
in
this
implementation
it
is
not
in
the
club,
it's
the
far
remote
green
green
board
is,
is
the
one
sending
the
picture
and
we
have
a
simple
control
loop,
meaning
that
the
display
with
the
rotation
sends
a
trigger
to
the
to
the
cloud
and
the
content.
The
timing
and
everything
is
done
in
the
remote
part
and
the
content
is
sent
to
the
display
and
the
display
is
very
simple.
F
F
We
have,
we
demonstrate
two
aspects
with
this
demo.
Primarily
this
is
an
illustration
device
build
what
we
built
ourselves
for
showing
very
little
packet
delay
variation
like
in
the
range
of
100
microseconds.
So
the
first
part
of
the
demo
shows
that,
at
a
very
simple
case,
if
all
the
traffic
is
in
the
same
queue,
then
the
displayed
content
is
a
mess.
F
We
can
apply
very
simple
techniques
like
a
stick
priority.
We
can
give
high
priority
to
the
picture,
but
still,
if
you
have
a
background
traffic,
the
background
traffic
actually
would
be
sent
from
the
laptop
that
can
disturb
your
your
content,
because
it's
despite
of
being
lower
priority,
if
it
is
being
already
have
the
transmission,
has
been
already
started.
You
have
to
wait.
So
if
it
is
huge
packets,
then
you
have
to
wait
for
the
transmission.
F
So
this
is
where
the
delay
variation
actually
depending
on
link,
speed
and
package
size,
that's
what
it
depends
on
and
what
we
would
show
is
a
kind
of
200
microseconds
deliberation,
which
means
the
image
is
shaking.
So
it
is
circus
you
so
partially
it
circles
what
we
use
and
the
image
shakes
if
it.
If
there's
a
delay
variation,
if
you
imagine
this
is
in
a
factory
like
it,
it
a
machine
like
creating
a
particular
surface.
This
delay
variation
is
not
tolerable
if
coming
back
to
the
use
cases.
F
Provide
deterministic
latency
is
preemption,
which
is
a
simple
concept.
It
is
a
link-local
here,
and
an
express
packet
can
preamp
the
transmission
of
lower
priority
or
less
important
preemptable
traffic.
So,
though,
you
don't
have
to
wait.
The
transmission
of
a
whole
lower
lower
priority
traffic
by
the
Express
that
gets
transmitted
and
the
demo
shows
that
the
image
becomes
still
again.
So
you
can
see
it
on
the
video,
as
I
said,
the
link
will
be
distributed.
F
So
this
is
the
latency
aspect
and
we
also
implemented
and
can
demonstrate
and
a
demonstrated
in
progress,
the
protection
aspect
for
the
reliability,
with
write
about
reliability,
very
low
loss.
This
is
a
per
packet
protection.
We
call
it
a
packet,
replication
and
elimination
in
in
that
net,
so
it
is
a
kind
of
fun
pass
on
redundancy
when
you
we
send
duplicates
of
packets
in
the
network.
As
you
see,
there
are
two
disjoint
paths.
F
We
use
the
lack
of
robot
to
demonstrate
it
before,
even
in
the
video
recording
we
will
distribute
or
the
last
Prague
meeting
in.
In
that
case,
the
Contra
and
the
the
application
is
that
the
balancing
logic
is
not
on
the
de
balancing
leg
or
like
a
robot,
but
in
the
cloud,
and
if
there
is
a
loss
like
moving
on
to
the
particular
before
that,
the
the
data
plane
we
implemented.
It
is
the
data
plane,
solution,
0
0
status,
and
that
was
prepared
for
the
Prague
meeting.
F
It
is
there,
it
is
in
the
MPLS
encapsulation,
with
a
transport
labor
service,
labor
and
that
control
word.
So
that's
what
it
is
on
the
on
those
species.
Actually,
the
PCs
implemented
the
ring
topology
I
show
in
the
figure
in
below,
and
here
we
we
investigate
or
did
the
two
scenarios,
a
simple
link:
failure
when
classic
techniques
provide
50
millisecond,
and
this
in
this
setup
means
like
losing
two
or
three
packets
and
with
the
logo
application.
F
It
is
too
much
for
the
balancing
logic
having
run
over
the
network.
The
Allegro
falls
over
with
this
display.
What
we
could
show
is
that
the
image
gets
completely
crappy
and
the
other
case
is
a
link
flapping
when
wet
which
we
often
do
not
detect
with
classic
techniques.
So
the
linear
protection
is
based
on
an
OEM
monitoring
and
you
switch
over
if
your
link
flapping
is
like
20
millisecond
or
it's
below
the
time
scale.
F
You
do
not
even
notice,
but
if
you
have
this
very
sensitive
control
application,
one
of
those
or
then
that
we
will
detect
an
and
and
actually
the
effect
is
the
same.
Like
your
control,
loop
is
broken,
Lego
falls
over
image
gets
broken,
so
that's
that's
the
state
status
before
that
net
and
what
we
have
in
that
net,
this
packet
replication.
F
If
we
turn
this
on
due
to
the
packet
per
packet
protection,
even
if
we
have
a
link
failure
in
one
one
of
the
parts,
there
is
zero
packet
loss
actually
because
the
there
is
no
switch
over
time,
no
detection
time.
The
packet
just
gets
to
the
elimination
point
and
it
sends
over
CentOS
and
onward.
I
am
very,
very
sorry
for
not
being
able
to
show
this
demo
or
can
we
show
anything
just.
F
F
Okay
enough
things
have
got
broken
today,
I
mean
even
just
power
supply
very
strange
things.
Nothing,
no,
no
fresh,
nothing!
It's
just
so
this!
This
is
the
what
we
want
to
have
and
if
we
have
a
failure,
wait,
15
minutes
a
failover,
then
actually
it's
we
will
have
multiple
failures
to
be
able
to
see
repeated
in
one
second,
so
50
millisecond
outages
repeated
in
1
seconds.
F
So
you
see
there
is
a
white
column
in
the
beginning
of
the
picture
and
that
that
screws
up
the
image.
So
if
you
imagine
this
as
a
machine
in
a
factory,
that's
that's
too
bad
or
if
we
have
the
link
flap
in
case
when
we
have
this
on
the
right
hand,
side
Illustrated
on
the
figure
link
goes
up
and
down
again
it's
it's,
it's
very
bad
for
the
for
the
picture.
F
This
is
without
proof.
This
is
classical
linear
protection
and
if
we
turn
on
the
packet,
replication
and
illumination
function
of
of
that
nut,
now
we
turn
on.
It
means
that
actually,
we
need
to
reboot
one
of
the
virtual
switches,
so
those
four
pieces
are
all
which
was
which
is
implementing
this
function
and
the
MPLS
data
pane.
F
So
if
we,
if
we
turn
it
on
and
do
the
same
thing
either
one,
let's
do
the
link
failure,
then
you
don't
even
notice
anything
because
one
of
the
packets
goes
to
so
in
the
video
recording
it's
the
Lego
robot.
This
was
we
wanted
to
have
only
one
device
in
this
small
table.
That's
why
we
brought
this.
Is
this
special
one
prepared
for
the
latency,
so
moving
on
I?
Think
time
is
running.
I
would
like
to
give
the
word
to
Lou.
H
So
first
I'd
like
to
thank
yo
and
his
colleague
for
doing
the
demonstration
and
one
of
the
nice
things
about
this.
Is
you
know
it's
not
a
can
demo.
You
know
you
go
to
so
many
demos
and
everything
works
flawlessly.
Anyhow.
Do
you
know
what's
really
working
well,
you
know
this
is
a
good
way
to
show
that
we're
actually
doing
something.
So,
unfortunately,
when
everything
works,
when
you
have
when
you're
running
the
protection
function-
and
you
have
the
the
links
breaking
back
and
forth,
you
just
get
a
solid
picture.
F
F
The
image
that
was
the
absolute
congestion
and
here
comes
the
stick
priority
when
the
people
are
sitting
closer,
see
that
the
image
is
shaking.
That's
where
the
rotation
is
very
important,
to
have
a
high
rotation
and
divide
the
picture,
because,
because
you
see
the
shaking
better
and
if
we,
if
we
apply
the
protector,
sorry
preemption,
then
the
than
the
shaking
disappears
so
there
so.
H
H
We
think
that
that's
you
know
what
we'll
come
and
solve
that
and
maybe
10
to
20
years,
but
maybe
that's
too
long,
I
don't
know,
but
eventually
we'll
get
there
in
order
to
achieve
something
real
well,
we
have
a
scope
of,
as
he
mentioned,
a
single
administrative
domain,
but
we
actually
also
allow
for
multiple
administrative
domains
been
there
when
they
are
cooperating.
So
this
could
be,
for
example,
a
customer
with
its
provider
cooperating
to
deliver
a
service
between
multiple
sites
that
that
is
certainly
within
our
scope
as
well
and
I.
H
Think
that's
an
important
distinction
in
terms
of
what
we're
trying
to
deliver.
There's
excuse
me
three
core
functions
or
services.
The
first
is
protection
from
congestion,
and
the
thing
I
always
like
to
talk
about
or
is
something
that
norm
Finn.
Who
was
one
of
the
people
who
work
on
this
early
talks
about
is
zero
congestion
loss,
so
we
want
to
make
it
so
that
when
you
have
a
debt
net
service,
when
there
is
contention
in
the
network,
the
services
that
receive
the
zero
congestion
loss
service
don't
see
any
packet
loss
whatsoever.
H
Now
we
can't
fix
the
line
problem
with
with
good
queuing.
We
can
fix
it
with
protection
like
you've,
seen
here
with
pref,
which
is
actually
that
the
last
one
was
service
protection
along
the
way.
We
also
have
latency
control.
You
saw
their
need
for
that
in
the
demo
already
so
I
don't
need
to
spend
too
much
time
on
it
and
the
last
part
that
we
think
it's
important
is
pin
droughts
or
explicit
routes.
H
We've
been
doing
that
so
long
in
traffic
engineering,
then
I,
don't
think
we
have
to
talk
too
much
about
it,
but
we
think
that
that's
a
necessary
component,
particularly
when
you
want
low
loss
without
doing
the
1+1
protection
we're.
Looking
at
now
from
a
architecture
standpoint
Dennett
wants
to
deliver
service
protection.
The
initial
solution
to
that
is
a
one
plus
one
protection
mode
called
pref,
but
we,
the
architecture,
allows
for
other
modes.
H
For
example,
there's
work
going
on
in
the
network
coding
research
group
which
maybe
will
deliver
something
which
allow
us
to
give
the
the
same
level
of
protection
without
sending
a
hundred
percent
more
of
traffic.
So
from
an
architectural
standpoint,
we
allow
that,
if
you're
interested
more
in
more
details
that
link
that's
going
to
be
impossible
to
read,
but
you
download
it,
you
can
find
our
architecture
and
it
covers
the
whole
picture.
I
encourage
our
I.
H
Ask
that
folks
review
it
we're
actually
getting
ready
to
last
call
it
it's
it's
pretty
mature,
there's,
maybe
a
paragraph
or
two
that'll
change
between
the
current
version
and
last
call
so
we'd
love
for
additional
review.
I
think
the
80s
mentioned.
You
know
please
review
documents,
so
this
is
an
additional
plea
for
that.
H
What
are
some
of
the
example
scenarios
were
trying
to
cover
so
there's
something
inside
going
on
in
the
I
Triple
E
called
802
one
TSN,
this
TSN.
It
builds
on
the
work
that
was
done
in
802,
a
VB
audio-visual
bridging,
which
talked
about
how
to
control
traffic
within
a
LAN
for
AV
per
tog
production.
You
know
there
was
a
pretty.
It
was
pretty
publicized
that
ESPN
redid
their
production
CBO
years
ago
and
threw
in
you
know
one
of
our
large
vendors
switching
gear
to
fully
distribute
production
within
a
single
building.
That's
great!
H
You
can
do
that
with
a
LAN
technology,
but
the
problem
there
is
is
you're
its
limited
scope.
That
net
wants
to
marry
with
that
technology
to
extend
the
technology
like
we
do
with
l2
VPNs
and
be
able
to
send
what
is
defined
in
I
Triple
E
is
really
a
a
single
site
technology
into
a
multi-site
technology.
Our
view
is,
is
we're
going
to
do
that
using
the
current
l2
VPN
technology,
with
some
additional
tying
together
with
an
understanding
of
what
the
TSN
control
is
into
our
TSN
control?
H
Excuse
me
and
flows,
and
tying
that
into
the
debt
net
flows
and
control.
Stewart
actually
talked
about
that.
Well,
I
thought
so
various
he
talked
about
that
type
of
service.
In
the
last
meeting
in
the
routing
working
group,
his
discussion
is
was
very
focused
on
sort
of
the
big
picture-
architectural
framework-
it's
great.
It's
supporting,
but
we're
we're
happy
to
take
solutions
right
now
in
debt
net.
If
you
have
them
and
but-
and
this
is
an
important
use
case
for
us-
but
it's
just
an
example
of
one.
H
We
also
you
know
we're.
We
do
IP
here
right,
so
we
want
to
give
a
native
IP
debt
net
solution.
So
this
is,
you
can
take
a
host,
it's
an
IP
host.
It
goes
through
routers.
You
have
an
interconnected
network.
You
will
want
to
have
subnet
technology
that
can
deliver
the
services
that
are
required
by
debt
net
and
not
all
of
it.
All
networks
may
be
equal.
For
example,
a
service
may
be
ultra
high,
latency,
sorry
ultra
high
ultra-low
latency.
We
don't
like
ultra
high
in
order
to
achieve
that.
H
You'll
need
a
certain
level
of
a
certain
level
of
control
in
the
sub
network,
but
you
might
be
able
to
achieve
lower
congestion
loss
without
all
the
TSN
services.
Excuse
me
or
all
the
sophisticated
queuing.
You
would
need
on
a
link
to
achieve
that,
so
you
know
having
your
te
network
today
could
deliver
certain
classes
of
debt
net
services,
one
of
the
things
that
we've
talked
about
rubber.
If
it's
on
the
it
comes
up
later.
H
One
of
the
things
we've
been
talking
a
lot
about
is
what
that
IP
host
should
do
in
order
to
support
that
net
service,
and
one
of
the
things
that
became
really
difficult
is
thinking
about.
How
do
we
provide
this
one,
plus
one
protection
from
a
IP
host
and
fit
into
the
normal
stacks,
and
that
caused
us
a
huge
amount
of
heartburn
and
cost
quite
a
bit
of
time
and
we've
come
up
with
an
approach
which
is
we
call
the
simplified
approach?
H
It
came
up
in
one
of
our
interims
we're
hoping
to
reach
agreement
that
this
is
really
the
way
we're
going
on
Friday.
So
if
you're
interested
in
this
come
to
Friday,
where
we
give
up
on
sort
of
the
image
at
the
objective
of
full,
end-to-end,
1+1
and
now
just
do
it
by
subnets,
so
you
can
use
leverage
the
TSN
technology
that
allows
you
to
do
that
and
if
we're
going
across
MPLS,
we
have
an
MPLS
solution
based
on
control
words.
You
know
some.
H
They
look
like
pseudo
wires,
but
you
know
is
an
EVP
n
control,
we're
at
a
pseudo
era.
I,
don't
know
it
doesn't
really
matter
from
a
forwarding
standpoint.
We
know
how
to
deal
with
them,
so
we
have
a
solution
for
MPLS
that
goes
MPLS
edge
to
edge.
We
have
a
solution
on
a
subnet
of
TS
n.
We
think
that's
good
enough,
at
least
for
an
initial
case.
It
doesn't
preclude
doing
something
fancier
if
you're
interested
come
on
Friday
and
talk
to
us
about
it.
H
Oh,
we
have
some
other
things
that
are
going
on
in
the
working
group.
We
have
a
security
model,
it's
actually
a
pretty
good
document.
Take
a
look,
please
comment.
One
of
the
things
that's
going
to
be
really
important
is
for
us
to
define
what
our
service
parameters
are.
We
have
a
good
start
on
that.
We
haven't
finalized,
that
we
want
to
be
able
to
have
a
wide
range
of
services
that
can
be
supported
from
those
services
that
are
need
all
TRO
control,
but
also
those
that
don't
we
think
we're
getting
a
handle
on
that.
H
But
this
is
still
evolving
work.
Eventually
we're
going
to
have
a
yang
model
that
gives
us
the
configuration
parameters
for
the
service
as
well
as
what
happens
on
a
node.
We
have
an
individual
document.
I
think
we're
going
to
hopefully
adopt
that,
but
we
have
to
see
how
the
workgroup
goes,
but
once
we
get
that
we
are,
we
think
we
have
enough
information
to
enable
control,
plane,
work
where
that
control
plane
work
happens.
H
Well,
there's
other
groups
that
do
control
plane
work.
They
should
do
those
we
think
they
may
give
them
some
requirements,
which
will
require
some
discussion,
our
working
group,
but
right
now
we're
thinking
that
that
goes
elsewhere,
but
we'll
see
what
happens
based
on
actual
discussions,
I'm
sure
I'm
over
time.
So
I'll
close
with
again
a
writer
come
join
us
on
Friday
we
actually
are
going
to
have
an
extended
session.
Are
the
first
session
is
a
normal
session.
It'll
feel
like
a
normal
working
group.
C
C
C
Again,
we're
gonna
go
back
to
the
administrative
part
now
that
we
solved
the
timing
issue
here,
so
we've
been
busy
the
last
cycle
a
little
bit
busy.
Yes,
so
working
group
changes
we
closed.
Three
working
groups
is
iOS
PFR,
no
more!
That,
of
course,
doesn't
mean
that
they
don't
exist,
so
we
don't
care
because
we
chartered
the
new
working
group
called
LSR
for
leaks
a-crowding.
So
now
we
have
both
of
them
in
the
same
working
group.
We
also
closed
trill
this
week
on
Monday,
oh
yeah,
one
of
the
last
thing
she
did
was
close
trail.
C
We
want
to
thank
everyone
in
the
trill
working
here
who
has
been
working
for
many
years
on
technology
and
especially
for
the
last
few
months
as
well,
for
the
getting
out
the
work
and
and
special
things
to
the
chairs,
to
John
and
Sue,
and
to
Donald
with
a
lot
of
shepherding
and
all
of
the
work.
If
you
hear
please
stand
up,
thank
you.
So
much.
C
I
guess
there's
the
very
tired
because
of
all
the
work,
so
they
didn't.
We
also
recharter.
Two
working
groups
would
return
it
as
a
C
and
B
R,
which
is
now
working
on
standard
track
documents.
This
is
a
big,
significant
change
for
beer.
Thank
you,
therefore,
everyone
who
has
been
working
on
that
besides,
we
also
had
a
lot
of
cheer
changes.
C
Some
of
them
because
of
Martin
becoming
an
ad
invest.
We
actually
changed
both
chairs,
both
Martin
and
Thomas.
Back
in
November
decided.
Let's
continue
so
it's
the
fun
who's
standing
right.
There
is
the
new
best
here
and
Matthew
Bochy,
who
is
over
there
he's
gonna
stand
up
now
is
the
other
janitor
welcome.
Matthew
has
already
been
chair,
he's
already
chair
of
in
vo3
in
that
net,
because
Pat
is
retiring.
Janos
is
a
new
chair.
You
can
send
up.
Yes,
Janos
is
the
new
chair.
You
already
saw
him
up
here.
C
He
seems
to
know
what
he's
doing
so
we're
okay.
There
we
have
two
new
chairs
in
illis
are
which
are
the
old
chairs,
AC
and
Chris
AC
you
get
up
just
in
case
Chris.
It's
probably
hanging
out
with
the
troll
guys
talking
about
bias
and
stuff.
That's
at
least
what
I
hope.
That's
the
only
excuse,
I'm
gonna,
buy
or
they're,
probably
filling
out
the
working
group
report
in
the
wiki,
which
I'm
sure
all
the
chairs
filled
out
already.
C
We
have
two
new
working
groups
that
are
focusing
on
data
center:
routing
LSB,
our
league
state
vector
routing
back
way
back
there.
Gunther
and
Victor
are
the
two
you
cheer
for
that
welcome.
This
is
Victor's.
First
working
group
and
I
was
first
in
the
routing
area
and
we
have
also
rift
routing
in
factories
Jeff,
since
we're
also
chairs
our
twg
is
here
and
Jeffrey
is
over.
C
Bruno
is
Rob's
co-chair
in
spring.
This
is
called
solidarity.
It's
called
teamwork.
This
is
great.
Thank
you.
This
is
what
we
want
to
see
in
the
routing
area.
So
this
is
the
new,
updated
work,
distribution,
80
responsibilities.
So
in
case
you
don't
remember
an
El
burro,
that's
Debra,
and
this
is
Martin.
G
G
G
C
I'm
gonna
stand
up
for
this
one,
so
we've
been
saying
all
week
and
finally,
it's
over
with
Aliyah
finished
her
four-year
period
or
two
tweeter
periods
last
night
at
the
IU
meeting
when
she
gave
her
got
to
chew
on.
We
don't
have
enough
words
to
thank
her
for
all
the
work
that
she's
done
in
the
roaning
area.
She's
been
a
mentor
to
both
there
were
and
I.
Wouldn't
we
came
a
DS.
It
was
a
really
hard
job
for
her,
especially
because
of
that's
just
really
because
she
had
to
wait
these.
C
This
is
the
first
time
when
the
routing
area
went
from
two,
ladies
to
three
and
alia
had
the
job
term
under
both
of
us.
At
the
same
time,
we
have
learned
a
lot
from
her.
She
has
been
not
only
her
guide
through
this,
but
she
has
been
a
voice
for
routing
in
the
IAG
and
across
across
various.
This
is
her
her
positive
outlook
always
wanting
to
work
with
others,
and
collaborate
has
been
an
example.
I
think
for
all
of
us,
Aliyah
is
unique.
C
C
C
If
you
want
to
show
it
to
everyone
sure
everyone
could
come
later
and
tickler,
one
of
her
I
think
the
same
machine
characteristic
Cecilia
likes
to
talk
a
lot
to
other
people.
Right,
not
just
you
know
you
guys
heard
so.
She
she
likes
to
communicate
a
lot.
She
likes
teamwork.
She
likes
to
foster
new
ideas,
new
people,
new
working
groups,
new
everything.
That
is
something
that
that
we've
seen
not
just
in
the
area,
but
she
helped
went
well.
C
She
started
when
she
became
a
lady
cheer
chats,
for
example,
so
we
said
you
know
about
once
a
month
or
every
other
month
with
the
chairs.
We
talked
about
chair
issues,
we
do
training
for
them,
etc,
and
this
has
created
a
good,
a
really
good
community.
You
just
saw
Bruno
come
up
here
and
cover
for
his
for
his
co-chair.
That's
the
kind
of
chair
bonding
that
we
see
in
the
routing
area,
so,
okay,
yeah,
you
can
take
pictures
pictures
as
well.
C
Aliyah
has
fostered
a
lot
of
communication
across
the
area
in
the
IAG.
She
has
looked
for
working
and
doing
outreach.
She
works
with
the
ad
team.
She
has
formed
a
local
community
in
the
Boston
area
that
you're
gonna
is
periodically
for
IT
efforts
to
meet
and
and
talk
about,
different
stuff
out
there.
C
She
is
on
experiments
with
the
and
we
were
three
chairs,
for
example,
on
better
ways
to
organize
rooms
and
your
performance
to
to
foster
communication
things
along
those
lines
so
to
chew
on
our
aliha,
we're
going
to
do
an
experiment
today
and
and
what
the
experiment
is,
is
we're
going
to
eat
the
cake,
so
we
have
some
plates
over
there
and
we're
going
to
eat
the
cake,
but
we
still
have
to
go.
Do
the
working
group
updates
and
John
and
Amy
have
put
together
some
slides
that,
unfortunately,
no
joke
on
the
road
in
Directorate?
C
So
what
the
experiment
is
this
you
want
you
to
come
to
the
front
eat
the
cake
with
us,
find
other
people
communicate,
foster
the
teamwork.
Tell
other
people
about
your
working
group.
Stressing
people.
Do
you
don't
know
if
you're,
not
the
chair
doesn't
matter
tell
people
about
what
you're
doing
tell
them
about
how
fabulous
your
work
is
and
invite
them
to
your
working
group.
C
So
we
want
to
foster
their
communication
to
extend
one
of
the
the
the
big
characteristics
that
alia
has
has
taught
us
of
better
teamwork
and
better
communication
through
that
personal
interaction,
so
yeah,
thank
you
so
much
we're
going
to
definitely
miss
you
and
you
have
a
great
impact
on
the
whole
later.
Thank
you.
C
So
so
keep
in
mind
that
the
meeting
is
not
over
until
5:50
I'm
serious.
This
is
a
serious
experiment,
we're
going
to
measure
the
communication
and
make
sure
that
everyone
talks
to
everyone
else.
Again
we
have
some
plates
there
you
can
come.
Maybe
you
want
to
look
at
the
cake
before
Chris
drops
it
on
the
floor
and
thank
you
so
much.