►
From YouTube: IETF114 LISP 20220727 1730
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Okay,
now
it's
what
it
works,
so
welcome
everybody
to
the
lisp
session
working
group
meeting.
We
have
a
one
hour
meeting
the
usual
note
well,
which
you
are
supposed
to
have
read
it
during
your
registration
procedure.
A
I
know
you
are
familiar
with
every
single
line
of
this,
not
well
few
tips
for
if
you
are
in
the
room,
you
join
the
mitiko
with
the
local
tool.
Okay,
so
that
in
this
way
you
can
sign
the
blue
sheets.
Also,
okay,
right,
you
can
arise
your
hand
if
you
want
to
go
to
the
mic.
Okay
and
wear
a
mask
unless
you
you
are
at
the
mic
and
you
are
speaking-
sometimes
is
better
that
way,
otherwise
for
people
that
are
in
remote,
the
usual
procedure
with
mitiga
in
remote.
A
Okay,
this
is
a
slide
that
we
have
already
the
last
time.
I
think
it's
important
the
code
of
conduct
also,
I
keep
it
are
expected
to
to
treat
each
other
with
courtesy,
which
means
also
to
to
wear
a
mask
because
it's
protecting
each
other.
Okay,
we
have
technical
discussions
here,
not
personal
discussions.
A
A
If
you
go
to
the
agenda
the
main
agenda,
you
have
directly
the
link
that
brings
you
to
the
eater
part,
the
equivalent
of.
B
A
B
A
Okay,
a
quick
update
on
the
status,
so
we
are
progressing
in
the
last
two
months
that
there
was
quite
some
some
work
to
do.
We
have
eight
documents
now
in
the
rfc
editor
queue,
which
is
good
news,
so
the
beast
documents
basically
are
through
lisptec,
okay.
A
That
was
quite
some
discussions
on
that
one,
but
yeah
we
we
seem
to
to
to
to
ready
to
actually
also
unblock
some
old
documents
like
the
the
introduction
document,
which
is
stuck
in
the
queue
for
I
don't
know
how
long
anymore,
but
now
with
the
the
the
beast
document
going
through,
we
should
be
good.
Is
there
any.
B
Means
you're
almost
done
what
you've
got.
What
you've
got
in
front
of
you
is
when
they
get
to
it.
The
rf
rpc,
the
production
center
will
do
their
job
when
they
are
done,
they
will
send
to
you
an
email
that
says
auth48.
Here's
what
we
did
to
your
document,
please
review
and
approve.
There
is
a
list
of
questions.
Please
answer
them
do
so
promptly.
Please
I've
had
authors
in
other
contexts.
B
B
D
So
hard
to
say
something:
no,
I
forgot
what
I
was
going
to
say.
D
Oh
the
documents
keep
in
mind
that
we
requested
expedited
processing
for
the
whole
cluster,
because
at
the
request
of
I
ko
right
and
there
was
no
way
to
request
processing
for
just
two
so
for
the
whole
cluster,
all
seven
or
eight
documents.
D
So
what
that
means
is
that
please,
as
joel
said,
reply
to
the
auth
48,
ideally
in
48
hours,
I
think
that's
what
that's
supposed
to
mean
so
that
we
can
get
these
documents
out,
because
all
of
them
are
probably
going
to
come
out
together
in
one
block,
and
so
we
need
them
to
to
get
out.
So
we
can
meet
what
iko
wants
us
to
do
so
between
luigi.
F
A
Yes,
good
plan:
we
have
another
couple
of
documents
that
they
they
passed.
The
working
group
last
call
we're
working
on
that.
Okay,
the
there
is
the
young
model.
I
think
it
was
already
mentioned.
In
the
last
time.
We
we
should
pay
attention
to
data
moving
it
forward.
I
think,
is
ready
or
or
really
close
to
it.
Okay
and
then
we
have
a
bunch
of
documents
for
which
we
need
to
to
to
focus
on
them.
Now
that
the
these
documents
are
over
or
almost
over.
A
Okay,
that's
the
next
next
task,
I
would
say
the
agenda
for
today
we
have
first,
the
least
reliable
transport,
okay
mark.
We
will
present
that
and
then
we
have
dino
with
the
satellite
networks
and
distinguished
name
encoding.
G
Okay,
so
we
just
wanted
to
provide
an
update
on
the
reliable
transfer
draft
now
that
it's
a
working
group
document,
I'm
mark
I'm
presenting
on
behalf
of
list
of
authors.
Next,
like
this.
G
So,
in
terms
of
updates,
what
we've
done
is
yeah
we've
been
integrating
all
the
changes
that
we've
been
discussing
during
the
last
meetings
in
this
document
in
particular,
what
we
are
doing,
what
we've
done
is
is
provide
further
details
on.
How
do
we
do
the
transition
from
udp
to
reliable
transport
then
also?
G
The
draft
now
incorporates
the
reliable
transport
bit
that
drives
the
understanding
between
the
etr
and
the
web,
server
that
when,
when
to
use
a
reliable
transport
and
now
the
document
includes
quick
as
an
alternative
protocol
to
to
use
for
reliable
transport
and
discusses
a
little
bit
how
to
select
the
protocol.
When
you
have
multiple
options-
and
I
reserved
the
last
part
of
the
last
slide
that
you'll
see
for
portal
locations
all
the
discussion
that
we
had
in
the
mailing
list.
G
Let's
see
now
next
slide,
please
so
from
udp
to
reliable
transfer
the
the
idea
of
that
the
draft
defense.
It's
always
you
must
always
first
register
with
you
lease
udp
once
successful,
udp
registration
has
happened
and
that's
the
only
time
that
you
can
switch
to
reliable
transfer.
G
So
what
we've
done
in
address
is
clarify
how
this
works
right
in
particular,
and
there
is
this
concept
of
as
a
security
measure.
The
the
map
server
should
never
accept
any
reliable
transport
session.
G
No
yeah
yeah,
the
idea
was
to
keep
things
a
bit
simpler,
just
to
always
go
back.
You
know
to
whenever
possible.
D
Right,
just
just
because
I
mean
you're
already
authenticated
and
well,
I
don't
know
no.
D
G
No,
no,
that's
a
good
point.
The
draft
now
doesn't
consider
it
but
yeah
it's
a
good
idea
to
tonight.
The
problem
is
that
you
don't
know
whether
you
have
lost
the
state
when,
when,
whenever
there
is
a
problem
with
a
reliable
transport,
so
the
the
simpler
way
that
we
found
was
to
just
go
back
right
and
restart
from
scratch.
E
D
So
the
question
was:
he
said
that
if
the
tc
position
goes
down,
then
you
have
to
go
back
and
authenticate
again
via
udp.
D
Okay,
so
okay.
G
Okay,
so
the
other
thing
that
I've
now
incorporated
is
the
reliable
transfer
bit
this
bit
is
just
used
so
that
the
etr
doesn't
need
doesn't
need
to
guess
whether
a
map
server
can
accept
a
reliable
transfer
establishment
or
not.
So
the
the
idea
is
in
the
map
register.
We
set
this
bit
whenever
an
etr,
so
that
the
atr
can
express
intent
to
establish
a
reliable
transport
next
slide
and
then
the
maps
the
map
server
in
the
notified
and
said
the
corresponding
leads
just
to
tell
back
the
edr
okay.
G
Okay,
so
what
the
working
group
document
now
incorporates
is
the
list
of
protocols
that
can
be
used
to
establish
the
reliable
transfer,
then,
as
part
of
this
list,
now
incorporate
quick
to
the
list,
as
we
discussed
in
the
mailing
list,
one
important
thing
that
we've
incorporated-
and
it
would
be
good
if
we
can
have
a
discussion
on
this.
But
first
thing
is
that
for
all
protocols,
what
the
draft
specifies
is
that
all
the
sessions
will
be
long
lived
right.
So
once
the
session
established.
G
This
is
what
there
is
this
understanding
that
the
state
is
synchronized
between
dtr
and
on
the
map
server,
and
this
is
what
simplifies
some
processes
with
reliable
transport.
But
the
second
part
is
the
one
that
I'm
interested
we
are
interested
in.
Hearing
back
is
as
of
now.
The
draft
specifies
that
quick
and
sad
will
use
a
single
stream
right.
G
We
know
that
the
protocol
offers
the
possibility
to
to
establish
multiple
streams
between
the
etr
and
the
map
server,
but
since
we
are
using
this
for
a
very
specific
interface
and
very
specific
messages
between
etrm
observer
for
now,
the
the
draft
just
reduces
this
to
single
stream.
G
G
This
this
part
of
the
draw
the
answers,
dino
had
a
comment
on
on
okay.
Now,
since
we
have
multiple
protocols,
how
do
we
choose
among
them?
And
and
what
do
we
do
about
the
potential
delay
that
can
come
from
the
fact
that
okay,
I'm
trying
one
protocol?
Maybe
this
protocol
fails
next
protocol
comes
so
the
draft
now
specifies
an
order,
but
just
as
a
suggestion,
let's
say
with
a
may.
G
So
if
the
etr
can
support
the
three
protocols,
this
would
be
the
suggested
order,
but
also
we'll
introduce
this
node
saying.
Okay,
since
we
don't
want
the
information
that
mapping
information
to
be
delayed.
While
we
try
to
establish
the
reliable
transport,
dtr
should
always
make
sure
that
it's
sending
complete
mapping
information
in
with
udp,
while
waiting
for
for
the
reliable
transport
to
be
established.
A
A
A
That's
a
good
one
about
the
order.
You
can
suggest
a
note
that
I
don't
think
there
is
the
need
to
to
mandate
an.
C
G
Another
good
one:
okay
yeah:
we
can
make
tcp
mandatory.
G
Good
and
next
one
yeah-
this
is
the
last
one,
and
this
is
what
we
discussed
in
the
mailing
list.
What
happens
with
the
ports
now
that
we
have
multiple
transports
for
dcv
is
easy.
We
can
reuse
for
342..
G
Then,
for
quick
and
sap
we
will
need
to
request
newport
just
one
minute
or
not,
since
we
added
the
ayanna
considerations
in
the
draft,
we
got
these
mail
automated.
I
guess
from
myanna
specifying
a
procedure
to
to
request
these
ports
and
they
also
say
that
the
closer
we
get
to
last
call
then
it's
when
when
they
will
act
on
it,
but
yeah
I
mean
I.
I
would
like
to
hear
based
on
past
experience,
if
you
think
we
need
to
do
anything
else
than
this
section.
E
H
G
G
E
A
A
G
D
G
I
E
Okay,
I
just
think
it'll
be
quicker.
E
Okay,
just
to
give
you
an
overview
of
what
it
is,
we
can
encode
what
we
call
distinguished
names
or
just
character
strings
that
can
be
put
in
an
eid
record
or
an
arlo
record
of
any
of
the
list
messages
we
use
the
afi
encoding
like
we
do
for
afi1
and
2
for
ipv4
and
ipv6.
E
So
it's
it's
just
basically
a
2
byte,
afi
17
value,
followed
by
an
ascii
string.
That's
null
terminated,
it's
pretty
straightforward
and
it's
useful
to
provide
self-documenting
the
mapping
database
records.
You
could
also
use
it
to
group
a
bunch
of
things
together.
So
maybe
you
want
to
associate
all
eids
on
a
train
with
the
name
of
the
train,
so
you
would
have
you
would
register
to
the
mapping
system
a
name
called
train
154,
and
then
you
would
have
a
list
of
eids
that
are
currently
in
the
train.
E
It
might
have
some
mobility
benefits
there
there's
a
lot
of
other
sorts
of
things
you
could
do
as
well.
The
idea
here
is
that
we
don't
want
to
break
the
semantics
of
this
string
with
power
of
two
in
addresses,
so
ddt
and
lisp
decent
those
two
mapping
database
systems
could
support
this.
So,
for
instance,
if
you
did
a
look
up
on
slash
root,
slash
dino,
slash,
slide,
slash
philly.
E
It
could
match
slash
fruit,
slash
dino
at
the
ddt
route,
which
would
be
the
delegation
authority
for
that
part
of
the
name,
and
then
the
children
could
then
support
slash,
slides
or
flashlights
billy.
So
it
looks
like
it's
more
specific.
Just
like
it
would
be
with
a
power
of
two
address
so
rather
than
it
wouldn't
have
bit
boundaries,
it
could
have
arbitrary
boundaries
just
based
on
so
so.
The
design
of
the
mapping
system
would
have
to
decide
if
the
slashes
were
the
boundaries
because
they
could
be
variable
length.
E
So
map
map
requests
lookups,
typically
in
the
use
cases
that
I've
implemented
and
that
are
currently
written
in
specs
do
exact
match.
But
you
could
also
do
partial
matches
as
well
like
if
I
looked
up
slash,
root,
slash,
dino,
slash,
philly
and
slash
philly
wasn't
registered,
but
slash
shoot,
slash.
Dino
was
you
could
get
that
returned
and
that
would
just
be
a
natural
thing
that
map
sir
would
do
with
no
changes
next
slide.
E
E
E
These
are
the
current
working
group
and
individual
submission
drafts
that
are
using
the
distinguished
name
stuff,
so
the
ecdsa
authentication
is
using
it
to
store
the
hash
value.
When
you
want
to
look
public
keys
up
in
the
mapping
system
to
verify
signatures,
we
use
it
for
grouping
type
stuff
in
predictive
rloks
and
we
use
it
for
encoding.
E
If
you,
if
you
use
the
lcaf
geocoordinates
as
an
eid,
and
you
want
to
map
it
to
a
name
like
say
you
have,
the
geo
coordinates
of
los
angeles,
you
may
want
it
to
map
to
a
name
which
is
called
character,
string
los
angeles.
You
could
do
that
sort
of
thing.
E
The
list
policy
distribution
guys
are
using
there's
a
lot
of
policy
stuff
that
they
want
to
use
that
are
going
to
depend
on
on
distinguished
names
and
in
all
these
cases,
they're
describing
how
the
structure
of
the
name
should
be
and
what
the
collision,
opportunities
or
or
avoidance
is
done
in
those.
That
was
a
comment
that
was
done
on
the
spec
overlay,
simple
nat
and
the
list
didn't
cast
stuff
refers
to
this
draft.
E
So
we
started
the
effort.
The
draft
was
built
in
2016..
There
was
a
lot
of
quiet
period
around
here
in
2020.
There
were
some
updates-
and
we
just
made
an
update
now
this
past
week
to
reflect,
make
it
more
clear
how
close
detection
is
done
if
there's
two
different
use
cases
using
names
and
how
they
don't
clash.
So
we
added
that
to
the
spec,
and
that
was
joel's
comment.
I
And
we.
E
So
we
addressed
the
collision
commentary
that
joel
had
in
slash
15.,
and
I
requested
this
past
weekend
to
request
this.
As
a
working
group
document,
I
sent
that
request
to
lisbeth
ietf.org
and
it's
really
it's
a
very
small
two-page
specs,
so
it
doesn't
take
very
much
to
read
through
it.
So
I
wanted
to
try
to
request
working
group
last
call
now,
but
it's
up
to
you.
B
E
D
F
B
H
J
J
G
C
B
A
A
Yeah
sharon
is
on
the
queue,
but
wait.
There
was
on
shadow,
want
to
ask
something.
K
A
K
So
you
know,
I
think
this
is,
of
course,
a
great
idea
and
a
way
for
people
to
understand
how
to
solve
problems
using
a
routed
namespace,
which
lisp
is
uniquely
qualified
to
do.
K
But
it's
hard
to
explain
that
use
of
lisp
without
being
able
to
discuss
use
cases
as
part
of
the
charter,
so
at
least
select
use
cases.
Possibility
to
discuss
should
be
added
to
the
charter
for
us
to
really
show
the
power
of
these
router
name.
Space.
E
F
E
E
L
A
Question
yeah:
that's
a
different
question
in
the
sense
that
to
change
the
chart
that
we
have
to
recharge,
which
is
not
how
to
say,
I
mean
the
the
charter
rank
now,
has
a
a
strong
focus,
basically
on
the
beast
documents
so
that
to
work
on
the
proposed
standard
version
of
lisp,
which
we
almost
done
once
we
are
really
done
with
that.
We
can
even
open
the
discussion
for
our
chartering
to
refocus
the
work
of
the
of
the
working
group.
I
guess
so.
This
is
something
that
we
we
can
discuss,
but
not
today.
D
I'm
not
going
to
suggest
this
question
today,
but
what
I'm
going
to
say
is
that
if
everything
goes
to
plan,
we
should,
within
a
month
publish
the
the
revised
specs
right
so
within
a
month,
if
there's
a
charter,
you
know
we
can
start
the
process
right.
The
formal
process
of
approving
the
charter
and
the
iesg-
and
you
know
all
that
stuff.
A
D
A
D
E
D
Okay,
so
that
I'm
going
to
find
that
here
somewhere,
okay
and
the
other
thing
is
this
uses
ascii,
not
utf-8,
or
something
like
that.
D
E
F
D
E
E
L
G
B
E
B
B
B
Says
what
it
says
right,
but
the
question
will
come
up
and
whoever
is
shepherding.
It
will
have
to
be
prepared
to
deal
with
it.
A
E
E
A
E
Okay,
so
this
draft
is
a
first
stab
at
looking
at
using
the
satellite
networks
as
an
underlay
and
running
lisp
as
an
overlay.
On
top
of
that,
so
it's
kind
of
got
to
stand
on
your
head
a
little
bit,
but
I'll
show
you
in
a
diagram
in
a
second.
E
So
the
high
level
goals
for
this
draft
is:
we
know
that
a
lisp
overlay
can
run
over
any
ip
packet
delivery,
underlay,
be
it
ipv4
ipv6,
and
if
the
satellite
network
can
deliver
ip
packets
which
existing
the
the
satellite
networks,
I
know
that
are
in
operation
today
can
do
that.
E
We
can
have
lisp
run
as
an
overlay
on
top
of
that
satellite
network,
but
actually
without
touching
anything
in
the
satellite
network,
and
this
would
be
just
another
example
of
how
we
can
run
lisp
over
like
we
do
the
capital
I
internet,
where
the
rloks
are
on
the
ground
or
3gpp
network,
where
we
run
lisp
on
the
enode
b's
or
in
the
ico
network,
the
aeronautical
network,
where
we
run
lisp
on
the
ground
stations,
we
would
just
run
lisp
on
the
ground
station
here
that
have
satellite
links
up.
E
This
is
not
necessarily
how
it
works,
but
kind
of
what
are
the
benefits,
but
as
always,
any
lisp
underlay
or
any
underlay
network
doesn't
store
eid
state.
So
that
would
be
the
same
with
the
satellite
network.
The
satellite
network
is
unaware
that
lisp
is
running
over
it.
It's
just
just
like
it's
kind
of
unaware
of
any
other
applications
that
run
over
it
too.
So
it's
not.
It's
actually
ignoring
the
satellite
network
and
just
using
it
as
as
an
ip
delivery
system.
E
The
overlay
requires
the
underlay
to
deliver
packets
to
our
local
addresses
and
the
underlay
can
that
the
underlay
can
route
to.
So
the
assumption
is,
is
when
an
itr
pre-pens
a
packet
and
uses
the
outer
destination
address.
We
assume
that
the
underlay
could
forward
that
pack
and
that's
what
we
call
a
routing,
locator
or
outlook.
E
The
underlay
number
can
transport
ipv4
ipv6
packets
and
can
be
dual
stack,
just
like
it
can
on
the
terrestrial
network.
Okay
and
when
path
optimization
is
available
in
the
underlay
just
like
on
the
capital
internet.
Now,
if
you're
running
segment
routing-
and
you
want
the
xtr
to
influence
the
path
of
the
underlay,
it
could
prepend
source
routes
or
segment
routes
in
the
packet.
This
would
be
the
same
in
the
satellite
network
as
well.
If
we
wanted
to
use
any
kind
of
source
routing,
be
it
relative
to
the
satellite
network
or
using
itf
mechanisms.
E
E
So
here's
a
pictorial,
we
see
those
satellites
up
in
space.
We
call
that
the
underlay
and
we
see
those
links
between
satellites
called
isls
inner
satellite
links.
Those
are
typically
one
or
more
laser
links
that
go
between
the
satellites
in
low
earth
orbit
in
leo,
and
it
turns
out
that
this
is
what
amazon
is
about
to
to
deploy
as
well
as
spacex
doing
with
starlink
okay.
So
what
we
do
is
that
these,
what
we
have
on
the
ground
are
the
overlaid
nodes
and
the
mapping
system.
E
Okay,
so
lisp
an
xtr
will
run
on
the
ground
station
that
has
a
satellite
link
that
goes
up
today.
Most
communications
that
happen
is
you
go
up
in
one
satellite
and
then
down?
If
you
go,
if
you
can
go
like
the
quarter
of
the
earth,
you
can
do
it
with
one
satellite
because
of
where
the
the
satellites
are
positioned
in
leo
in
geosynchronous.
That
was
always
the
way
because
it
was
much
higher
up
added
delay,
but
it
can
capture
more
of
the
earth's
surface
here.
It
may
not
do
that.
E
E
E
The
id
mobility
draft
says
it's
no
different
than
if
the
eids
are
roaming
in
a
vm
on
a
top
of
rack
switch,
it's
the
same
sort
of
machinery,
that's
used,
so
eid
mobility
and
even
predictive
our
looks
can
be
used
even
in
this
scenario,
because
it's
these
are
list
features
and
we
don't
care
what
we
don't
depend
on
the
underlay
to
do
anything
here.
E
So
if
they
go
up
to
space-
and
you
don't
want
anybody
to
snoop
on
them,
even
though
the
russians
are
leaving
the
space
stand
center,
now
you
could
you
could
encrypt
the
packets
and
of
course
the
gs
xdrs
can
use
alternate
paths
when
maybe
the
satellite
network
is
not
totally
connected,
and
I
give
two
examples
here:
normally
you
would
go
on
an
xtr,
you
would
go
up
to
a
satellite
and
then
maybe
go
across
an
isl
to
another
satellite
and
then
down
to
another
xtr.
E
Maybe
the
isls
are
because
the
satellites
are
always
moving.
The
lasers
are
going
up
and
down
all
the
time.
You
may
not
have
that
connectivity,
but
of
course
we
can
use
list
traffic
engineering
and
actually
encapsulate
to
an
rti.
So
we
would
go
up
from
this
xtr
to
a
satellite
and
then
back
down
to
an
rtr.
F
E
To
satellite
and
down
so
those
those
sort
of
traffic
engineering
features
are
available
by
just
using
regular,
lisp,
okay,
and
we
could
use
the
telemetry
stuff
that
we
proposed
in
that
draft.
This
draft
here
to
actually
test
the
satellite
network,
so
we
know
which
paths
are
good
and
we
can
we
could.
Actually,
if,
if
the
satellite
network,
each
satellite
is
an
ip
hop,
we
could
actually
trace
route.
We
can
measure
one-way
latency,
both
forward
and
reverse
hop
counts,
all
that
kind
of
stuff.
E
Just
like
we
do
the
under
the
terrestrial
underlay
boy,
it's
hard
to
breathe
and
percent
present.
At
the
same
time.
I
haven't
done
this
yet.
E
Ground
stations
xtrx
can
offer
eid
multicast
service
by
doing
head-end
replication
using
any
underlay
multicast
service.
So
if
the
satellite
network's
going
to
provide
a
multicast
service,
we
can
use
it.
Otherwise
you
can
just
unicast
replicate
over
it.
We
know
that
works
as
well.
I'm
not
it's
not
clear
to
me
and
it
hasn't
been
specced
out.
If
there's
a
multicast
service,
that's
going
to
be
supported
in
the
in
the
satellite
networks
that
are
being
deployed
today
or
are
planning
to
be
deployed.
That
should
be
interesting.
E
Work
like
I
said
this
is
kind
of
a
first
stab
we're
learning
more
about
what
the
underlay
could
do
and
are
there
advantages
to
make
the
overlay
work
better.
If
we
can
make
take
advantage
of
the
underlay
it's
kind
of
a
two-sided
sword
where
you
sometimes
want
to
always
ignore
the
underlay,
because
you
just
want
it
to
deliver
simple
packets
and
then
you
just
want
all
the
features
to
be
on
the
edge
in
the
overlay
and,
of
course,
if
you
have
a
mix
of
lisp
and
non-list
sites,
all
the
interworking
features.
E
So
if
you
have
somebody
an
eid,
that's
sitting
behind
a
ground
station
xtr
that
wants
to
talk
to
google.com,
that's
not
a
lisp
site,
all
the
machinery,
the
list,
the
list
and
that
stuff
and
the
pxtr
machinery
can
work
as
well.
E
The
only
thing
we
didn't
talk
about
is:
do
you
want
to
put
any
list
nodes
in
on
the
satellite
nodes
and
make
them
routers
not
sure?
Yet?
So,
if
anybody
has
any
thoughts
of
on
that
and
if
it's
useful,
we
could
determine
that.
I.
I
F
E
So
I
was
just
question:
should
we
make
a
working
group
document?
I
don't
I
don't.
I
don't
have
any
preference,
I
mean
we
could
work
on
it
as
an
individual
submission.
I
think
the
work
we
have
left
to
do
is
to
look
closer
at
the
satellite
networks
and
see
what's
going
on
and
maybe
try
to
do
some
prototyping.
I
think
my
open
source
implementation
could
work,
as
is.
E
I
just
need
a
box
with
a
satellite
link,
so
I'm
gonna,
I
my
plans,
are,
is
to
call
up
spacex
and
say:
can
I
put
some
open
source
code
on
on
your
ground
station
and
let's
see
what
they,
what
they
would
do,
what
they
would
say?
I
don't
know,
but
I'm
going
to
try,
but
you
know
even
actually,
even
if
it's
one
hop
away
from
a
starling
box,
it
might
serve
the
same
purpose
because
the
server
could
you
know
what
I
mean.
So
we
could.
We
can
still
test
it,
though.
E
Now,
when
it
comes
down,
it
always
comes
to
a
google
data
center,
so
you're,
not
you're
you'll
have
to
deliver
it
to
a
non-eid
or
something
like
that,
but
we'll
you
know
if
we
could
put
an
xtr
in
a
vm
in
a
google
cloud
or
whatever
and
test
it
that
way.
So
I'm
planning
on
doing
some
experience
with
that.
You
guys
are
welcome.
H
E
E
So
the
the
devices
that
are
on
the
ground
that
are
moving
from
ground
station
to
ground
station
they'll
use
eid
mobility.
According
to
the
draft,
we
did
itself
so.
E
The
xtr
movie
well,
this
is
an
interesting
new
property
that
we've
never
seen
before,
because
even
though
the
xtr
is
stationary
normally
normally
when
a
lisp
when
eids
move
they
get
new
r
logs,
because
they're
landing
somewhere
new
right,
but
it
turns
out
what's
happening,
is
now
the
xtr
is
actually
stationary
right.
But
the
satellites
are
moving
in
and
out
and
the
question
is:
will
the
satellite
network
re
reassign
the
rlope
to
that
ground
station?
E
But
if
it
does
it's
just
like
the
eid
moves,
so
it
is
kind
of
like
a
mobility
event,
but
all
our
machinery
will
work
with
in
that
case,
as
well,
so
interesting
kind
of
cool,
okay,
we'll
see
we'll
see
if
it
introduces
new
problems,
but
I
I've
done
this
testing
before
where
I
would
run
lisp
at
my
house,
and
I
would
disconnect
my
cable
link
all
the
time
and
every
time
I
disconnect
well,
I
think
it
was.
It
was
a
few
years
ago
with
att
dsl.
E
C
Kinji,
china
mobile,
you
know
I'll,
have
an
arrest
through
your
your
draft,
but
from
your
description
and
talk
today,
I
and
also
you,
said
okay
so
far,
there
are
only
one
as
a
link,
but
mate
is
a
link,
since
you
did
not
mention
anything
about
the
dynamics
about
how
the
rotation,
orbital
movement
no
type
of
things.
C
E
Oh
we're
just
it's
just
another
underlay
that
we
can
run
over
and
if,
if
we
want
to
reach
rural
places,
which
is
what
starling's
use
case
is
we
may
want
to
run
it's
it's
not
the
only
way
you
can
do
things.
I
mean
if
you're
our
loc,
if
you're
on
a
car,
if
you're
driving
in
a
car
and
lisp
is
running
in
your
car
and
your
outlook
is
changing
because
it's
on
a
4g
or
5g
network,
so
be
it.
E
But
if,
if
that
car
needs
to
communicate
with
something,
that's
in
a
you
know
up
in
montana,
northern
montana
or
something
and
they
don't
have
internet
access
except
for
starlink,
then
you
run
lisp
on
that
side
and
then
and
then
they
can
move
between
ground
stations
in
their
local
area.
You
know
the
cars
moving
around
in
this
urban
area
and
then
maybe
the
cell
phone's
moving
around
in
this
rural
area.
You
know.
C
Of
the
technology,
I
I
still
cannot
see
the
the
technical
difference
like
yes,
I
understand
more
in
order
to
talk
some
people
in
mahana,
not
some
some
place
nowhere.
You
have
your
satellite,
but
still
in
terms
of
technology,
but
to
me
just
like
in
the
moving
car.
It's
it's
have
no
difference
from
just
like
a
satellite
movement
are.
C
No,
no!
No,
because
tomorrow,
I'm
going
to
have
another,
I
have
a
draft
on
my
satellite,
but
I
really
try
to
understand
why
you
know
this
is
satellite
suppose
you
try
to
use
some
satellite
technology
that
is
significant
for
the
list
or
for
your
draft.
Is
that
the
not
having
the
restroom?
So
in
your
draft?
Have
you
talked
anything
about
like
satellite
technologies
that
will
make
your
draft
more
significant?
E
We
we
have
a
lisp
draft
called
list
mobile
network
that
talks
about
how
you
support
layer,
3
mobility
inside
of
3gpp
network.
In
that
use
case,
we
say
list
can
run
on
the
enode
b's.
So
when
cell
phones
move
move
from
one
enob
to
another
e
node
b,
they
don't
have
to
change
their
ip
address
and
their
connection
stay
up.
Okay,
so
that's
the
case
because
cell
phones
connect
to
towers
okay.
Now
we
have
these
other
things
called
ground
station
xtrs
that
have
rf
links
up
to
aircraft.
E
That's
what
the
account
standard
is
doing,
and
so
planes
are
moving
around
and
and
they're
eid's
up
there
and
they're
connecting
to
ground
stations,
and
so
that's
just
another
sub
network
using
that
wireless
network.
So
now
we
have
satellite,
which
is
yet
another
one.
And
of
course
we
have
the
terrestrial
internet
right.
That's
all
okay!
Does
that
make
sense.
I
Oh
sorry,
eric
klein
eleria,
forgive
me
for
I'm
not
a
terribly
well-versed
lisp
person.
Is
there
a
presumption
that
the
gs
xdrs
have
some
terrestrial
control
plane
connectivity
all
the
time
or
is
it
possible
that
they
may
only
have
control
plane
connectivity
over
the
orbital
domain?.
I
And,
if
so,
does
that
have
implications
for
the
control
plane,
reliability.
E
Yeah
great,
really,
good
question:
can
you
go
back
to
the
diagram,
so
the
mapping
system,
the
assumption
we're
making
in
the
first
version
of
the
giraffe,
is
that
the
mapping
system
is
on
the
ground
and
that's
the
control
plane
that
we
use
for
this
guy
to
find
that
load.
But
it
turns
out
that
the
mapping
system
ip
addresses
to
access
it.
The
map,
servers
and
resolvers
are
all
in
our
lock
space.
E
F
E
A
I
have
one
one
last
question:
I
I
mean
if
you
go
to
this
slide,
yeah,
you
say
across
in
the
second
bullet
or
across
our
locks
in
space.
If
in
space
you
have
our
locks,
it
means
that
the
satellites
have
to
run
lisp.
They
have
to
be
xtrs
right
because.
A
H
You
go
to
the
to
the
diagram.
F
H
E
L
Let's
go
mike
cisco,
hey
mike,
I
was
wondering
if
you,
if
you
gave
into
consideration
that
there's
any
mtu
limitations
between
the
gs,
xdr
and
the
satellite,
because
I
know
this
like
the
us
military
and
these
guys
they
have
crypto
devices
in
front
they're,
always
struggling
to
get
like
gre
and
mpls.
In
this
whole,
alpha.
E
A
Okay,
so
if
there
are
no
more
questions
or
comments,
there
is
jody.
J
Well,
just
this
out
of
curiosity:
well,
you
know
the
isl
iesl
links
they
usually
change
between
the
satellites.
As
far
as
I
know,
do
you
think
it
would
make
sense
converting
satellite
into
a
mobile
node,
at
least
mobile
node.
E
From
a
list
perspective
yeah
like
give
it
an
eid
and
let
it
let's
see
yeah.
J
H
E
J
E
And
then
40
minutes
later
it's
going
to
come
around
and
you're
going
to
see
it
again
right.
But
of
course
this
guy
doesn't
know
because
he's
homing
to
a
satellite
and
that
satellites
come
in
and
out
he
just
has
connectivity.
He
doesn't
even
really
care
what
he
is.
Remember
the
r
look
that
he's
sending
a
packet
to
is
on
the
ground
and
the
and
the
satellite
is
just
a
transit
network.
Yeah.
E
But
I
mean
I
haven't
considered
what,
if
you
want
a
satellite
system
to
be
an
eid
node,
that's
part
of
the
overlay
and
and
because
it
will
go
out
of
phase.
Do
you
want
to
be
able
to
reach
it
on
the
overlay?
It's
a
little
more
complex
use
case
and
we'd
have
to
see
a
really
strong
requirement.
Why
we
would
need
that,
but
right
now
I
just
want
to
use
the
satellite
as
it
exists
and
see
how
well
it
delivers
ip
packets
right.
A
This
is
something
we
can
discuss
more
on.
The
mailing
list
we
are
time
is
up,
basically
can
see
if
you
can
also
send
your
comment
on
the
mailing
list,
so
that
can
be
discussed
more
than
clarified.
That
would
be
helpful
other
than
that.
Welcome.
Welcome,
thank
you
for
being
here.
Thank
you,
dino
and
see
you
again
in
london.