►
From YouTube: IETF104-LISP-20190329-0900
Description
LISP meeting session at IETF104
2019/03/29 0900
https://datatracker.ietf.org/meeting/104/proceedings/
A
B
B
C
D
E
G
Hello
good
morning,
everyone
so
welcome
to
the
Lisp
working
group.
I'm
gonna
start.
Is
there
not
well
as
usual,
which
basically
mean
everything
we
say
here
today
is
a
contribution
to
the
IETF
I
guess
as
well,
the
joke
we
were
making
a
couple
minutes
ago,
but
anyway,
the
usual
pointer
and
pointers,
I'm
Rick
lien
on
a
Joel,
our
partner.
We
are
the
two
co-chairs
Padma
our
Secretary
was
him
is
missing
the
other
secretary.
We
will
see
what
happened
to
him
and
I
would
like
to
go
just
as
usual
on
the
working
group
updates.
G
So
this
is
the
exact
same
slide
like
in
Bangkok,
which
doesn't
mean
we
didn't
make.
Any
progress
just
doesn't
show
up
yet
on
the
data
tracker.
So
there
are
a
bunch
of
documents
that
that
I
don't
whole,
but
the
main
documents
on
on
which
we
were
working
between
Bangkok
and
now
are
the
two
miss
documents
and
list
sack.
So
on
the
Miss
documents
we
had
a
thorough
security
review.
G
We
had
some
some
delays
going
back
and
forth
on
the
image
we
had
a
meeting
with
then
this
this
week
and
I.
Guess
everything
is
sorted
out.
All
that
will
give
a
summary
in
the
meantime,
lisp
sack.
We
put
it
on
working
good+
call,
because
one
of
the
comments
from
the
ASG
was,
if
it's
difficult
to
evaluate
the
level
of
security
of
the
these
documents
when
part
of
the
protection
security
is
in
the
Lisp
SEC
document
so
will
be
good
to
have
everything
together
to
to
review
okay.
G
So
we
put
in
what
a
working
group
last
call
actually
been
right
away,
make
a
good
review,
and
so
these
there
are
a
few
changes
that
will
be
down
to
the
document,
but
I'm
pretty
happy,
because
that
then,
basically
I
think
we
sorted
out
all
the
problems
and,
at
the
end
least
pieces
better
than
before.
Okay,
hopefully
we
will
progress
quickly
and
move
through
other
stuff
concerning
the
agenda
today,
so
we
have
Albert
with
the
updates
on
the
base
document
and
this
act
I
guess
today
as
well.
G
A
quick
update
on
the
young
model
is
are
the
two
working
group
documents
and
then
so
we
had
a
request
from
the
north
on
the
decent
release,
mapping
system
and
shadown
about
disability,
geospatial,
Lisp
blackboard
for
automotive.
It's
gonna
do
a
demo
technical
stuff,
I
have
to
say
so.
The
original
agenda
had
then
25
minutes
discussion
time.
Okay,
after
the
this,
we
will
open
the
mic,
but
if
there
are
no
big
discussion,
we
have
to
late
request
about
the
slot.
She's
least
uber
lay
about
to
ask
it,
and
least
anonymity
at
masks.
C
J
Okay,
so
I'm
going
to
present
the
typically
I
come
here
and
I
present
what
we
have
changed
into
document
now,
I'm
going
to
explain
what
we
plan
to
change,
which
is
the
plan
ahead.
This
was
what
we
was
discussed
with
the
reviewers,
particularly
with
the
security
reviewers,
and
the
title
is
wrong
because
it's
actually
1636
edited
this
plus
Lisp
second.
So
this
is
a
summary
of
the
list.
These
are
least
that
summarizes
pretty
much
all
the
comments
that
we
got
from
the
Security
Review
and
how
we
are
going
to
address
them.
J
J
The
two
three
four
and
five
are
security
of
data
plane
mechanisms
which
each
of
them
has
some
security
concerns,
which
was
right
by
the
reviewers,
so,
for
instance,
deeming
this
well
known
it
can
lead
to
spoofing
of
gleaming,
can
lead
to
resurrection
of
on
opposite
Achatz
same
with
LSB
for
the
currents,
then,
on
his
to
shorten
decent
allows
of
path.
Attackers
to
do
s
a
lisp
site,
a
map
versioning
which
enough
path
attacker
can
spoof
it
to
to
to
try
to
prevent
signaling
effects.
J
Then
the
the
six
is
the
ask
for
a
clarification
on
the
interior,
bring
protection
that
we
have
for
my
register,
then
seven
is
that
we
are
using
the
keys
directly.
The
the
private
share
key
directly
to
sign
the
map
register,
and
this
has
some
security
concerns
which
we
are
addressing
and
finally,
it's
rather
a
clarification,
which
is
how
we
are
protecting
against
and
to
replay
attacks,
map-request
map-reply,
so
the
first
one
now
I'm
going
to
go
through
all
of
them
and
I
know
to
spin,
which
is
how
we
plan
to
fix
em.
J
So
the
first
one
is.
This
is
a
scenario
right,
so
you
you
have
a
non-lisp
set
capability
are
and
that
needs
to,
and
then
you
have
an
idea
that
is
not
requesting
through
Lisp
SEC
nav.
Now,
how
can
this
ITR
understand
that
this
non-lisp
set
capability
are?
Is
non-leaf
second
discussed
to
be
done
in
a
secure
way?
Otherwise
we
are
open
to
a
done
great
attack
and
the
way
we
have
decided
to
address
it
is.
J
This
is
protected
by
Lisp
sack
and
now
it's
up
to
the
local
policy
of
the
ITR
to
decide
whether
to
send
a
new
map
request
without
Lisp
sack
or
decide
not
to
communicate.
But
the
interesting
thing
is
that
that's
a
way
to
the
idea
to
get
secure
information
that
the
ITR
is
none
Lisp
sack
and
prevent
downgrade
attacks.
J
J
If
you
want
to
read
them
so
then
issues
two
three
four
and
five,
which
are
security
of
discipline
mechanism
and
that
something
that
we
discussed
with
Ben
and
there
is
a
verbal
agreement
which
only
means
that
there
is
a
verbal
agreement.
We
need
to
write
that
into
the
document
that
they
need
to
review
it
again.
But
what
we
agree
is
that
we
will
state,
in
the
definition
of
dreaming
LSB
occurrence,
I'm
a
burgeoning
that
those
mechanism
should
not
be
used
on
a
scenario
on
a
public
interest
in
area.
J
I
said
we
need
to
work
out
a
bit
more
and
on
the
writing
on
how
we
say
that
and
I
theory.
She
had
a
good
way
of
putting,
but
I
cannot
remember
exactly
your
wording,
something
like
I.
Don't
remember,
it
was
not
not
public
in
the
public
internet,
but
it's
what
we
will.
But
the
message
is
this:
one
was
something
like
yeah:
do
not
use
it.
If
it's
not
that
close
and
trusted
environment,
something
like
that.
Ok,
we
need
to
work
the
the
wording,
but
that's
pretty
much
the
message.
G
G
J
So
this
will
be
included
also
in
the
upcoming
versions
of
the
document.
Then
they
also
asked
for
a
clarification
on
how
we
are
protecting
against
until
pre
attacks
in
my
register,
which
is
something
that
it
is
already
on
the
text
and
I'm
just
refreshing
it
or
explaining
it
again.
So
the
in
this
case
the
attack
is
that
Anita
is
my
precious
Turing
with
a
particular
knowns.
Then
our
attacker
can
intercept
this
the
entire
message
and
replay
it
and
and
register
it
with
the
semi
ID
2
o'clock
mapping.
J
J
J
You
take
this
key
derivation
function,
which
I
understand
it's
pretty
much
like
an
H
map.
So
it's
not
very,
very
simple.
Then
you
take
the
nonce.
You
take
a
string
which
are
in
our
case.
We
have
decided
that
we,
it
will
be
Map,
Reduce,
authentication,
plus
the
private
Cherokee.
You
add
everything
together.
You
perform
this
function
over
all
these
information
and
then
you
get
the
key
that
you
will
actually
use
to
sign.
The
map
register.
This.
H
K
Ya
in
the
values
exchanges,
a
female
we
consider
interpreter
with
the
old
version
and
I
think
it's
your
next
slide
because
is
where
you
may
be
the
next
one.
Okay.
Now
then,
it's
not
in
this,
like
your
I,
mean
your
idea,
it's
not
in
its
like,
but
we
talked
about
it.
So
the
way
to
make
it
interval
with
existing
implementation
is
that
today
we
define
the
algorithm
ID.
That
is
there
in
the
packet
and
the
algorithm
lady
today
is
specifying
only
the
H
Mac
function
that
we
use
in
the
map
register.
K
So
basically,
in
the
map
register
message
message:
there
is
the
authentication
data
that
is
there
and
the
algorithm
is
telling
you
that
you
should
use
a
shark
one
or
56
whatever
we
can
define
a
new
algorithm
IDs
that
are
say
specifying
not
only
the
the
HVAC
function,
but
also
the
KDF
function.
So
then
we
define
I,
think
we
have
defined
right
now,
a
great
idea,
zero
one
and
two
will
define
a
great
IV
tree.
K
That
is
saying
this
will
use
H,
Mac,
sha-256
and
KDF
h,
KD,
f
from
AFC
I,
don't
remember
the
number
68
31
or
whatever
number
it
is,
and
that
is
basically
specifying
that
for
that
new
key
ID,
a
great
idea
mechanism,
the
ETR
will
have
to
do
not
only
the
H
mark
computation,
but
before
that
it
will
have
to
do
the
key
derivation
function,
and
this
is
extensible
because
we
can
define
future
algorithms
that
you
know
whenever
somebody
will
decide
to
charge
156
or
H
KDF.
So,
and
so
it's
not
good.
We
can
define
anyone.
H
This
is
Dino,
so
Fabio.
Can
we
do
that
with
like
a
high
order
bit,
because
if
we
set
the
higher
two
bit-
and
we
should
say,
let's
use
KDF
for
all
the
existing
hashes
that
are
identified
so
the
hired
if
it's
zero,
it
works
says
today
the
higher
a
bit
set
in
the
Lord
of
its
a
shot
256?
That
means
you
sha-256,
with
KDF.
K
Yeah
I
didn't
thought
about
that.
Yeah
I
think
it's
a
it's
a
once.
In
a
time
I
mean
we
have
defined
already
I,
think
H
Mac
96
in
H
Mac
on
the
28
that
are
1
&
2.
Maybe
we
can
define
rather
than
defining
3
we
define
you
know.
We
said
to
one
guy
out
of
it,
but
I
don't
think
that
will
become
a
I
mean
we
can
listen
other
than
assigning
3.
We
can
assign
rather.
H
K
I
don't
want
to
burn
two
entries
in
there
for
things
that
in
the
future
will
not
use.
I
mean
it
is
very
clear
today
that
we
will
define
H
moksha
under
28,
plus
HK
D
F
and,
let's
see
in
the
future,
all
the
other
will
will
be
will
not
be
so
we
can
do
that
I
mean
rather
than
defining
3
4
4
H
mark
plus
h,
KD
f,
since
H
max
Otto
Charlotte
28
was
we
define
1
2
extra
decimal
but
yeah.
It
might
be
a
small.
J
J
So
basically,
we
interoperate
with
legal
implementations
by
saying
H
Mac,
whatever
kvf
none
I'm
still
don't
get
the
high
bit
order,
but
sooner
or
later
we
understand.
Ok-
and
this
was
also.
This
is
the
last
item.
It's
Clara
fication
to
the
rebbe
words,
because
the
reason
some
overloaded
semantics
on
the
world
nuns
and
at
some
point
it
was
written
that
the
nuns
was
providing
anti
replay
protection
for
map-request
map-reply
and
that's
actual,
not
true
and
with
Lisp
check.
J
What
actually
it
is
providing
and
the
replay
protection
from
map-request
map-reply
is
the
fact
that
we
are
using
one-time
key,
but
this
music
is
used
only
once
and
that
after
you
receive
the
map
reply,
you
have
to
discard
the
nonce
and
the
one-time
fee
that
you
use
for
this
map
request
an
appropriate
change
and
then,
which
means
that
the
nonce
it's
a
random
number,
that
it
is
only
used
to
index
the
map
request.
So
when
you
send
a
map-request
the
end,
you
receive
a
mob
reply.
J
G
J
J
K
I
think
yeah
what
is
providing
an
T
reply.
Protection
is
the
fact
that
the
one
time
key
is
one
time
so
I
generated.
One
thing
he
is
a
big
number
is
under
28
bits.
I'll,
do
the
map
request
for
a
exchanged?
Replies
come
back,
I
I,
take
you
know,
I
check
the
integrity,
protection
and
then
I
discard
the
one
thankee.
K
M
K
64-Bit,
it
still
has
to
be
randomly
generated
because
you
don't
want
to
use
the
same
nouns,
because
otherwise
you
will
point
to
the
same
one.
Turkey,
but
the
security
I
mean
the
the
the
the
property
of
anti
replay
is
not
in
the
nose
is
in
the
one-time.
Key
right
is:
let's,
let's
make
a
difference,
because
otherwise
the
nonce
would
have
had
to
be
much
bigger.
H
So
I
just
wanted
to
address
Luigi's
question,
so
the
nonce
otk
pair
in
both
the
implementations
of
lists
I've
done.
It's
also
used
as
a
data
structure.
First
rate
limiting
map
request
in
lieu
of
the
map.
Reply
coming
back,
you
want
a
rate
limit,
so
what
happens
is
if
you
send
a
map
request
and
you
don't
get
a
map
reply.
You're
gonna
at
some
point
later
send
another
map
request.
You
send
it
with
the
same
nonce
with
a
different
otk
or
you
can
be
a
new
to
tuple
pair.
G
K
J
K
H
The
other
hand
the
attacker
could
just
be
in
the
mat,
a
man
in
the
middle
and
just
dropping
the
map-request,
causing
you
to
read
transmitted
to
see
if
you're
using
the
same
one-time
key
I
mean
one
time
is
a
relative
term
right.
Is
it
one
time
for
this
destination
query
or
is
it
for
each
individual
map
request
should
be
the
latter?
It
I
thought
yeah
yeah,
yeah.
G
N
A
really
quick
update
on
on
the
young
model,
so
yeah.
Since
the
last
lesson
we
publish,
we
basically
took
care
of
dinners
comments
on
the
underworld.
Him,
as
you
can
see,
on
the
slide
we
also
took
care
of
the
colony,
was
coming
from
about
distance
at
a
zero.
That
is
the
one
which
is
for
the
internet
and
34:34
birth,
and
then
we
also
make
sure
to
align
the
the
key
with
what
is
defining
this,
the
nurses
to
lead
into
to
in
the
gem
model.
N
Now
the
what
we
would
like
to
clarify
with
the
chest
in
the
working
group,
we
believe
that
we
are
now
aligned
with
with
the
latest
version
of
the
of
the
piece
of
comments.
So
what
are
the
next
steps?
Sad?
We
start
the
working
of
Lascaux.
Also
the
document
needs
young
doctor
review
I'm,
not
completely
sure
of
the
process,
but
I
believe
that
that
it's
up
to
the
tears
to
us
for
our
for
our
doctor
review.
G
O
H
O
H
N
B
F
N
G
H
B
H
H
We
created
the
document
back
in
July
of
last
year
and
we
presented
it
I
presented
it
in
London
about
one
year
ago
and
then
Colin
presented
it
in
Montreal
a
one
draft,
the
202
draft.
We
just
did
an
update
last
November,
adding
a
pull
based
mechanism
to
the
already
push
based
mechanism
that
we
have
and
we'll
explain
that
in
a
second.
H
H
What
if
the
Lisp
XT
ours
could
be
their
own
mapping
system
and
half
didn't
have
to
depend
on
a
third
party,
so
we
can
basically
build
a
democratized,
decentralized
control,
plane
and
throughout
the
presentation
where
you
see
green,
we're
talking
about
e
IDs,
where
you
see
red
we're
talking
about
our
locus,
so
the
use
cases
are
basically
cryptocurrency
or
blockchain
applications.
Take
advantage
of
this
decentralized
approach.
H
Emergency
networking,
where
you
don't
have
access
to
the
capital
I
internet,
but
you
have
connectivity
among
devices,
can
use
this
sort
of
feature.
Plug-And-Play,
VPN
networking
when
you
don't
want
to
depend
on
anything
else.
You
come
up
with
a
container
system
with
micro
services
and
then
come
down.
It'd
be
nice
that
they
could
do
find
their
own
mappings
to
each
other
by
having
a
decentralized
face
network.
H
H
So
if
you
just
take
that
last
picture-
and
you
just
squeeze
these
sites
like
that-
because
they
have
this
backdoor
while
you're
listening
connectivity,
how
would
the
mapping
system
be
useful
if
the
links
broke?
So,
let's
say
you
know,
you're
in
New,
Orleans
and
Katrina
comes,
and
you
have
no
connection
to
the
internet.
How
are
people
going
to
be
able
to
find
each
other
note
that
these
are
all
wireless
devices
here
moving
around?
H
H
So
we
have
a
push
based
model
that
we
presented
last
year
and
we
actually
download
it
for
you,
we're
gonna,
repeat
that,
just
to
show
you
the
complete
design
and
how
we
fit
in
the
push
model
and
the
pull
model,
but
what
if
each
xtr
could
be
a
map
server
and
what?
If
each
xtr
could
map
register
to
each
other's
XT
hours
so
they're,
not
only
a
data
plane,
no
they're.
H
Also
part
of
the
mapping
system
and
the
mapping
system
would
always
be
synchronized
because
the
updates
are
being
pushed
to
everybody
who
needs
to
use
them.
Ok
and
then
the
xtr
could
be
a
resolver
for
itself
since
you're
getting
the
map,
you
don't
have
to
send
a
map-request
externally
since
you,
since
everybody's
synchronized,
you
send
a
local
map
request
to
yourself
or
just
do
a
memory
lookup
to
yourself
in
the
push
based
model.
H
What
we
want
is
maybe
a
subset
of
those
X
TRS
could
be
co-located
in
B
map
resolvers
and
map
servers,
not
all
of
them
because
of
scale.
This
is
very
much
similar
to
blockchain,
where
you
have
a
set
of
seed
nodes
that
bootstrap
all
the
peers
and
if
we
know
a
common
seed
node,
if
Padma
and
I
are
using
Luigi
as
a
seed
node,
we
can
be
attached
to
this
peer-to-peer
network
at
the
application
layer
and
because
of
Luigi
patent
and
I
could
discover
each
other.
H
So
that's
kind
of
the
idea
here
with
the
pull
based
model
as
well.
Of
course,
the
registrations
would
be
shard
it
just
like
they
are
today,
and
the
Eid
would
be
used
as
an
input
to
a
hash
function,
to
select
where
map
requests
and
map
registers
go
and
they'd
have
to
go
to
the
same
place
or
a
wouldn't
work.
And
then
we
have
redundancy
through
the
DNS
level
of
an
indirection
and
I'll
show
that
that
it
sounds
like
a
couple
of.
B
Issues
and
they
may
be
addressed
on
later
slides,
but
one
if
you
want
stability,
something
gets
strange
when
you
suddenly
add
another
server,
because
he
can't
actually
take
responsibility
for
any
addresses,
because
anything
that
would
hash
to
him
now
must
have
hashed
to
somebody
else
before
so
you
there
seems
to
be
a
problem
with
adding
and
subtracting
things.
If
you
require
that
the
function
is
permanently,
we.
B
B
H
Let's
look
at
the
push
based
system.
First,
a
consolidated
mapping
system
is
defined
by
a
multicast
group
address,
so
everybody
that
wants
to
use
the
same
mapping
system.
They
had
this
to
get
be
able
to
get
access
to
the
same.
Mappings
are
identified
by
this
group
address
and
then
the
XT
hours
are
part
of
the
mapping
system
that
are
part
of
that
mapping
system
joining
that
multicast
group
and
then
map
registers
are
sent.
The
group
all
xt
RS
will
then
receive
all
mappings,
and
then
we
can
get
efficient
distribution.
H
If
the
underlay
supports
multicast,
then
really
the
each
of
these
guys.
Each
of
these
map
or
XT
hours
only
have
to
send
one
pack
and
it
gets
distributed
for
the
underlay
and
if
not,
we
can
use
head
end
replication,
that's
already
part
of
the
data
plane
and
less,
and
then
that
map
request
has
very
low
latency,
like
I
said,
because
the
lookups
are
done
locally,
because
you
have
a
synchronized
database
and,
like
I,
said
the
ex
tra
and
exactly
one
map
register
for
NXT
ours.
Now
what
that
means
is.
H
Is
that
the
yet
the
xt
are
that
or
the
control
plane
module?
That's
sending
a
map
request
only
builds
one
and
basically
sends
it
down
to
the
data
plane.
If
the
data
plane
knows
that
there's
under
like
multicast
and
one
actually
goes
out
on
the
network
and
gets
replicated
in
the
network,
otherwise
the
data
plane
headend
replicates,
but
it
creates
a
multicast
or
multi
participants.
H
So
here's
an
example:
we
had
these
a
set
of
XT
hours
on
the
left
hand,
side
that
are
part
of
one
mapping
system
and
a
set
of
XT
RS
on
the
right
hand,
side
and
I
just
used
ipv4
on
the
left
side
and
ipv6
on
the
right
side,
but
they
would
basically
be
joined
to
224
one
one
one
and
push
around
there,
these
mappings
and
likewise
on
the
right
hand
side.
So,
let's
look
at
a
pull
based
mapping
system
and
see
how
that
can
work.
H
B
H
B
H
B
H
H
J
H
So
the
modulus
index
then
selects
this
DNS
name
it
based
basically
on
what
I
just
said,
and
then
the
Dia's
DNS
name
has
multiple
a
records
for
each
map.
Server
and
map
registers
am
a
pre
cusco
they're
the
same
map,
server
set
and
what's
interesting
about
the
pole
based
system
is,
we
fix
the
bug
that
was
identified
in
previous
presentations,
where
we
had
a
problem
bootstrapping
the
push
based
system,
because
we
use
the
mapping
system
to
do
the
group
joining.
H
But
if
you
need
to
use
the
mapping
system
to
do
the
group
joining,
how
do
you
to
make
the
mapping
system
real?
You
have
this
kind
of
circular
dependency.
If
you
don't
statically
configure
it.
So
we
think
the
pull
based
system
could
actually
bootstrap
the
push
based
system
and
then
what's
interesting
about
the
pull
based
system.
Is
it
doesn't
have
to
connect
all
the
map
servers
together?
H
H
E
H
Get
those
two
a
records
back,
so
we
see
we
have
an
organization
like
this,
where
these
map
servers
are
not
synchronized
with
each
other.
They
don't
connect
each
other,
there's
no
hierarchy
on
top
that
connects
them
together.
So,
for
instance,
this
xtr
one
that
wanted
to
register
this
slash
24.
What
has
that
character
string
with
the
instance
ID
and
then
do
it
modulus
floor
and
get
a
value
of
zero,
and
so
we'll
do
a
DNS
lookup?
H
It
could
do
it
ahead
of
time
or
when
it
needs
it
of
0ms
that
Lister's
net,
the
mappings,
the
DNS
system
returns,
1
1,
1,
1
and
2
2,
2
2,
and
since
it's
a
map
register
gets
sent
to
both
places.
So
you
have
your
redundancy
similar
on
the
right
hand,
side
for
an
ipv6
address.
States
its
index
was
3.
You
would
register
2
7
&
8.
Now,
let's
say
a
map
request
was
being
said.
H
B
H
Prefix
length
is
the
so
we
had
we
added
something
in
the
latest
draft
called
a
hash
mask
and
asked
mass
are
high
order,
bits
that
are
common
between
the
lookup
and
the
registration.
So
here,
if
you
look
at
it
close
enough,
the
hash
mask
could
be
the
high
order,
16
bits,
which
would
be
the
same
now.
Some.
B
H
B
H
B
H
Okay,
so
so
that's
the
algorithms
and
we
have
an
implementation
of
both
of
them
and
I'm
just
going
to
give
you
a
prefab,
Reif
them
out.
A
demo
means
just
a
bunch
of
screenshots
it'll,
be
really
quick
and
then
I'll
I'll
just
get
off
the
stage.
So
this
in
this
push-based
demo,
we
showed
this
some
in
Montreal
or
I've
guessed,
maybe
London
I,
don't
think
Colin
did
it
but
I
think
but
anyways.
We
have
three
containers
running
the
lists:
Persaud
netcode
and
we
have
a
docker
bridge.
H
That's
not
doing
multicast,
because
we
wanted
to
show
headin
replication
that
the
data
plane,
but
it
we've
certainly
demoed
it
with
using
underlay
multicast.
The
x-ers
are
doing
the
headed
replication
and
the
xkr-s
register,
an
ipv4,
Eid
prefix
and
a
name
a
ID,
that's
the
what
the
demo
does
and
if
you,
if
we
just
go
here,
those
are
the
three
XT.
H
H
So
each
of
these
XT
hours
have
joined
that,
and
so
the
data
plane
knows
to
replicate
to
each
one
of
these
arlok's.
So
they
can
get
the
information
okay.
So
let
me
show
you
the
po
base
demo.
This
is
one
map-resolver
one
M
RMS
container
simulating
those
eight
map
servers
that
I
showed
in
the
previous
example
and
then
I
have
one
ITR
container.
That's
map
registering
for
a
e
IDs
and
then
one
ITR
container.
That's
sending
a
map
request.
H
Of
course
the
DNS
prefix
is
MS
the
DNS
suffix
okay,
there
you
got
me
Joel,
here's
this
that
lispers
dotnet
in
the
modulus
is
for
so
this
is
an
example.
If
you
did
a
get
host
by
name
on
the
zero
dot,
MS
lispers,
not
net,
you
would
get
our
looks
and
they
have
to
be
routable
by
the
underlying
internet.
Saying
with
1.2.3.4
alaria
is.
B
H
H
B
Q
H
You
know
in
the
DDT
model
was
great
because
we
understood
bgp
peering
and
how
agreements
are
between
different
organizations
and
but
that
still
complicates
things
now.
These
things
could
be
run
loosely
coupled,
but
you
have
to
you
know
you,
as
a
mapping
service
provider,
have
to
know
that.
Sorry,
that's
not
my
hash
index.
You
have
to
go
to
my
competitor
to
figure
out
the
problem.
Yeah.
O
H
H
H
G
H
You
you're
relying
on
a
protocol,
not
a
global
infrastructure.
Today,
when
you
bring
up
containers
a
container
system,
that's
completely
isolated.
You
can
use
DNS
names
because
it
implements
its
own
version
of
DNS
right.
It's
only
talking
about
the
devices,
not
containers,
I'm
talking
about
the
same
thing
same
doesn't
matter.
J
J
So
what
you
do
is
so
when
you
come
from
app
requests,
your
hash,
div
and
then
out
of
the
hash,
you
know
the
name,
the
DNS
name
over
which
to
look
up,
and
then
you
have
the
IP
of
them
observable
to
query
right
right,
okay,
so
the
so
you
are
trusting
the
DNS.
You
are
basically
are
trusting
the
DNS
for
authentication
of
uavs
right
you're.
H
Using
another
level
of
indirection
by
using
the
DNS
naming
system
to
give
you
these
a
records
which
are
IP
addresses
of
map
servers.
Yes,
yeah.
Okay,
these
could
be
Etsy
host
entries.
They
can
be
statically
configured
in
your
configuration,
but
if
they're
statically
configured,
then
you
don't
have
that
that
dynamic
resource
discovery
thing.
R
B
Either
we
need
to
say
that
or
we
need
to
not
be
dependent
on
it
in
the
draft
I
mean
yeah,
of
course
not
well.
You
can
make
it
work
with
a
ledger.
We've
we've
done
this.
Before
of
we
can
make
it
work
with
this
or
that
or
the
other
thing
not
criticizing.
You
I'm
not
objecting
to
use
a
ledger,
but
we
need
to
be
clear
in
the
draft
yeah
about
what
we're
requiring
Dino
didn't
say.
One
word
in
his
presentation
about
a
distributed.
R
Look
I
agree
on
this.
This
specific
draft
was
more
so
getting
the
infrastructural
components
of
how
the
distributed
Magnus
system
work.
All
these
other
little
intricacies
as
far
as
like
DNS
or
a
blotch,
and
everything
else
are
kind
of
I
guess
secondary
to
that.
But
if
we
I
guess
we
could
maybe
do
some
modifications,
so
it
includes
some
distributed
ledger
in
there.
If
you.
J
S
S
Be
able
to
see
it
maybe
later
my
computer
is
a
tiling
of
every
meter
of
every
road
making
it
every
meter
in
every
road
addressable
using
a
standard
called
h3
using
hexagons.
It's
a
standard,
if
you
zoom
in
you,
go
from
big
hexagons
once
for
kilometres,
very
big
hexagons
to
one
square
meter
or
sixty
centimeters,
so
having
these
addressable
entities
as
part
of
the
road
is
going,
is
why
we're
going
to
use
lists
and
and
I'm
going
to
explain
why
in
vehicle-to-vehicle
networks,
why
do
we
want
vehicles
to
talk
to
polygons
instead
of
other
vehicles?
S
Okay?
Why
do
we
want
this
indirection
and
why
this
indirection
involves
this
geo
spatial
partition
and
why
this
partition
is
in
the
ID
space?
What
is
it
good
for
so
alright,
alright,
so
vehicle
to
vehicle
vehicle
to
infrastructure?
Is
there
to
improve
safety
on
the
road
on
the
shared
Road
so
that
a
traffic
lights
cat
eyes
on
the
roads,
other
cars
will
respond
between
what
we
do.
Some
avoid
accidents.
Basically,
however,
the
adoption
of
vehicle
to
vehicle
networks
in
the
last
decade
has
been
very,
very
slow,
not
good.
Why?
S
First
of
all,
it's
the
the
start
was
simple.
Just
share
GPS
vectors.
Okay
between
cars,
the
result
is
all
those
who
did
play
with
it
is
that
every
car
gets
a
bunch
of
this
about
every
other
car
around
it.
It's
it's
used
and
if
you
want
to
share
better
annotations
its
the
interoperability
of
what
are
you
shared?
You
see
some
obstacles,
you
see
a
jaywalker.
Do
you
see
a
blocked
car?
This
is
very
hard
to
do
interoperability
wise,
so
the
industry
is
kind
of
stuck
now
in
a
war
of
layer,
two
who's
gonna
operate.
S
S
The
reason
the
reason
we
confronted
with
it
earlier
than
the
OEM
car
manufacturers
is
because
we
actually
have,
if
not
the
largest
one
of
the
largest
vehicle-to-vehicle
network
out
there,
simply
because
of
the
way
we
went
to
market.
So
what
we
did
is
we
paired
the
iPhone
of
a
driver
with
a
very
cheap
HD
camera
like
a
$50
camera
like
this
small
thing
and
it
can
pair
with
the
built-in
cameras
of
the
car.
So
it's
very
easy
to
give
away
a
lot
of
cameras
and
have
eyes
on
every
meter
of
every
Road
all
the
time.
S
So
we
did
that
and
we
put
a
few
tens
of
thousands
is
every
market
we
went
into
and
we
got
to
like
a
new
york
bay
area,
israel,
chicago,
and
then
we
get
to
experiment.
Okay.
Now,
let's
we
have
these
eyes
on
the
road.
We
have
not
just
a
camera.
We
have
AI
annotations,
half
the
company's
AI,
a
machine
vision.
We
see
things.
S
How
can
we
communicate
and
part
of
this
giveaway
of
all
these
cameras,
these
drivers,
which
host
them
very
nicely
by
the
way,
because
it
serves
them
well
for
deductibles
and
insurance
and
and
one-click
things
like
that,
so
they
keep
it
well
tune
focused
on
on
the
road
pair
to
their
iPhone,
with
point-to-point
Wi-Fi.
So
the
insurance
companies
wanted
to
see
some
ADA's
assisted
driving
safety
features
and
was
very
easy
to
come
up
with
many
okay
and
one
first
one
was
highways
in
New,
York
slowdowns.
S
You
know
the
you
ride
on
the
highway
and
it's
suddenly
so
and
then
it
moved
afterwards
I
mean,
but
the
slowdown
is
there
for
a
while
and
it's
a
very
dangerous
situation,
because,
if
you're
not
focused,
then
you
get
these
pilots,
especially
if
it's
icy,
you
see
it
in
the
last
minute
you
hit
the
brakes,
you
start
to
twirl
it's
a
major
source
of
pretty
bad
accidents,
it's
very
easy
to
annotate
you
drive
and
suddenly
the
brake
lights
in
front
of
you
lit
like
Christmas
trees.
Now
you
want
to
tell
people
behind
you.
S
How
do
you
do
that
with
vehicle-to-vehicle?
You
start
to
play
lacrosse
with
other
drivers,
and
you
need
to
find
drivers
you're,
interoperable
wheel,
and
it's
it's
not.
It's
not
a
very
good
way
to
start
warning
people
about
a
very,
very
clear
hazard,
which
you
clearly
identify,
and
you
can
clearly
give
people
twenty
thirty
second
heads-up,
so
they
can
start
to
slow
down,
and
there
are
many
examples
like
it.
S
So
we
went
to
an
indirection
model,
which
is
the
car
that
Caesar
slowed
down
the
car
that
sees
a
jaywalker
the
car
that
sees
a
blockage,
somebody
unloading
goods
to
a
shop
and
will
block
your
lane,
will
not
tell
it
to
any
other
car,
but
will
tell
it
to
the
virtual
tile
of
the
road
and
that
has
a
unique,
add
annotation
as
a
unique
address.
Ie
ID.
S
So
this
annotation,
which
is
part
of
a
BDD
enumeration
standard
in
this
hexagon
ID
I,
can
snap
to
a
hexagon,
ID
or
meter,
not
some
jittery,
GPS
and
therefore
I'm
sending
a
message
to
a
tile.
So
what
better
approach
to
communicate
between
busy
ideas?
And
here
it
is
an
overlay
map,
assisted
overlay
list.
So
this
is
how
we
want
to
use
this
and
we're
actually
using
and
we're
converting
the
v3
vehicle-to-vehicle
architecture,
which
was
like
gaming
architecture
peer-to-peer
to
a
list
based
in
direction.
Alright,
so
so
we're
not
sharing
peer-to-peer.
S
There
is
no
point
interoperability
and
we
want
it.
Open
is
too
difficult.
Instead,
we
communicate
three
network
state.
The
state
is
in
EAD
space
represented
by
h3,
IVs,
64-bit,
hierarchical
codes,
and
it's
SP,
XT,
ours,
I
will
shut
already.
No
I
will
show
it
my
back
and
these
published
scars
to
hed
north.
So
how
does
it
work
before.
B
You
go
on
I
just
want
to
make
sure
I've
understood
what
you
just
said,
because
if
I'm
seeing
something
probably
other
people
are
wondering
what
this
looks
like
not
Lisp
terms
to
the
moment
case,
created
a
database
indexed
by
an
ID
per
tile
and
you're
storing
in
the
database.
The
reports
from
every
car
that
has
a
meaningful
report
about
this
tile,
presumably
with
some
currency.
So
you
don't
store
relevant
data
and
then
anybody
who's
interested
in
this
set
of
tiles
subscribes
to
it.
B
D
B
So,
there's
a
day
that
there
is
a
conceptually
a
it's,
not
really
a
server
per
tile,
but
that's
conceptually
a
server
per
tile.
So
when
you
do
the
Eid
lookup
you
get
the
address
of
the
exact
you
get
the
name,
the
address
of
the
server
which
is
responsible
for
that
data
or
the
database
key
for
that.
Okay,.
S
Q
S
So
that's
the
good
question.
So
in
order
to
publish
an
annotation
I
saw
this
I
have
to
be
1
meter.
Accurate
GPS
is
not
correct,
and
so,
but
the
way
a
auditor
is
through
machine
vision
which
can
be
used
localization
to
better
snap
to
set
and
give
me
the
1
meter
so
to
publish
I
need
very
good
technology
to
subscribe.
That's
not
the
case.
S
I
can
be
just
a
normal
navigation
app
and
when
I
go
into
a
cell,
I
need
to
get
a
dump
of
the
next
20
seconds,
so
I
need
to
get
a
heads
up
and
then
I
can
locate
myself.
But
then
what
were
the
hazard
is
on
the
map.
So
I
will
let
the
navigation
or
whatever
correct
the
GPS
as
best
it
can.
But
the
hazard
position
on
the
map
is
correct.
Well,.
B
So
that
would
seem
to
have
the
risk
that,
if
my,
if
I'm
the
driver
who's
you
using
the
data,
if
I'm
on
the
service
road
for
the
highway
and
the
highway
is
congested,
I
may
get
the
report
that
there
is
congestion
in
front
of
me
when
there
isn't
now.
If
it's
only
viewed
as
informative,
that's
probably
not
fatal.
But
if
I
get
told
it's
clear
in
front
of
me
because
it
thinks
I'm
on
the
side
road
when
I'm
actually
on
the
highway.
That
could
be
a
serious
problem,
because
I
won't
react
in
advance.
S
Q
Other
than
the
thought,
so
we
always
had
the
problem
of
save
Elizabeth
free
roaming,
so
predicting
where
I'm
heading.
For
so
it's
there,
some
idealist
if
I
have
all
their
hexagons
and
I
know.
I'm
going
on
the
street
and
I
may
be,
like
me,
I'm
from
Germany
I'm
going
280
kilometers
power.
So
it's
their
way
to
pre
Rome.
That
I
can
already
encapsulate
to
the
next.
S
I
P
I
I
T
B
Strikes
me
I
probably
contributed
to
this
we're
diving
into
a
lot
of
details,
navigation
systems,
information
collection
systems
which
are
all
related
to
either
the
underlying
database
or
to
the
application
on
the
car.
Navigating
car
I.
Think
we,
because
we're
good
on
what
we
wanted
to
give
some
other
people
time.
We
should
focus
on
it.
This
is
being
used
for
this.
It's
really
tempting
we're
all
engineers.
We
want
to
go
refine
all
the
rest
of
it
and
Sharon
would
appreciate
the
feedback,
but
we
have
a
focus.
S
S
M
M
S
S
R
R
S
N
S
On
your
cars,
okay,
they're
roaming
around,
they
are
connected
to
the
edge
of
that
metro
area
based
on
low
density,
pretty
random,
which
one
but
low
latency,
and
then
these
cars
see
something
he
needs
to
tell
tire.
So
this
thing,
they're
64-bit,
we're
64-bit
watts,
so
it
sends
encapsulated
to
its
home
RTR,
which
then
looks
up
the
airlock
of
this
tile.
Okay,
so
you
can
look
up
the
mapping
system
in
not
there's
about
10
million
times
in
every
metro
and
discovering
which
one
is
our
row.
S
There's
also
part
of
the
network
on
volume
which
one
not
just
fields,
and
do
it,
but
they're
registered
nose
of
taro
that
registered
here
and
using
the
mapping
system
you
get
auto
in
get
to
xt
r3,
which
is
the
base
which
being
where
the
process,
the
kind
of
functions
and
the
state
of
the
a
tree.
So
this
is
how
I
talk
to
the
tire
by
talking
to
the
tile
I'm,
also
letting
everybody
know
that
I'm
behind
our
TR
one
and
I
also
get
sick
even
are
consistent.
S
Okay,
why
do
they
need
to
know
where
I
am
because
nobody's
gonna
talk
to
me?
But
the
tide
is
gonna
talk
to
me
and
here
the
idea
is
that
we
update
every
second
and
in
five
G
in
Navy
podcast,
if
you're
not
the
ten
times
a
second
without
at
all,
RF
and
but
for
now,
every
second
and
the
states
like
which
is
one
like
about
fifteen
sixteen
hundred
bytes
or
the
next
20
seconds.
So
the
multicast
goes.
The
channel
is
hid,
one
preferably
signal
gets
to.
S
S
The
latency
impact
is
the
next
leg.
Ons
are
very
micro
partition.
Therefore,
the
blocking
factor
is
very
low
that
doesn't
contribute
to
latency.
It's
very
well
partition.
What
does
is
typically
the
access
network.
So
today,
if
I'm
running
on
Amazon
edge,
then
it's
300
seconds
if
I
push
it
to
Equinix
Metro
it's
330
minutes
ago,
the.
B
Point
is
I,
think
let
me
put
it
if
we
add,
as
far
as
I
can
tell
you're,
not
trying
to
solve
car
in
front
of
me
has
suddenly
stopped
other
sensors
other
mechanisms
you're
trying
to
solve
the
problem
of
when
there's
a
problem.
That's
already
occurred
that
I
can
tell
you
about
slowly.
Give
you
enough
warning
to
react.
Different
class
of
problem,
different
latency
issues,
different
correct,
yes,
simplify,
a
lot
of
problems
by
focusing
on
a
problem
that
we
can
solve.
B
U
J
S
J
This
is
why
a
server
is
its
babysitter,
so
not
limited
to,
but
I
understand
that
what
but
but
unitive
Wow.
So
you
send
that
the
position
of
the
eggs
egg
on
the
smallest
one
and
you
need
the
mapping
system
to
be
able
to
tell
you.
Okay,
this
is
specificity.
Second,
there's
the
smaller
variety
it
is
located
in
that
server.
Yes,
exactly,
and
there
is
a
question
here,
we
want
double-click
on.
S
S
Okay,
this
is
a
stuff.
We
sleep
on
all
traffic
light
meters,
never
mind
that
this
is
jet,
you
can
click.
This
is
something
yeah
we
can
show
it.
It
doesn't
matter.
Okay,
use
case
sharing
in
town
annotations
through
beyond
line
of
sight
across
garbage
trucks,
all
the
stuff
that
hurts
commute
double
Park
Jay.
S
Yeah,
the
idea
is
that
we
propagate
we
have
an
ability
to
go
into
production
really
quickly
because
they
have
a
large
footprint
right.
Now,
it's
in
tens
of
thousands
per
market,
it's
gonna
be
hundreds
of
thousands
per
market
and
we
can
be
ahead
of
the
market
in
understanding
things
that
OEM
are
just
don't
know
what
to
do
about
it.
Yet,
okay-
and
this
is
a
test
that
we
did
in
an
Air
Force
Base.
S
This
is
actually
the
blockage
that
picked
up
by
this
traffic
and
merge
and
beyond
line
of
sight-
and
it's
very
clear
you
go,
you
see
that
the
map
sees
everything,
but
when
you
go
into
a
car
this
high,
you
don't
see
anything
and
then
you
get
a
heads
up
and
it's
really
maintenance.
Your
scale
is
just
for
people
to
finance
these
cameras
and
distribute
them
really
help
incoming.
T
T
So
how
do
we
achieve?
This
is
really
hiding
in
a
crowd,
or
rather
here
it's
in
pool
of
aggresses.
So
basically,
what
will
happen
is
that
we
use
a
little
bit
like
the
credit
cards
you
use
it
once.
Then
you
throw
it
so
the
next
time,
somebody
even
if
extraord
it
I
tried
to
get
back,
he
said,
got
no
guarantee.
That
is
actually
you
and
so,
and
then
we
recycle
it.
What
the
graph
is
very
simple.
T
Drought
is
actually
specifying
that
range
of
ID's
that
we
can
do,
or
you
know,
and
actually
giving
other
suggestions
of
how
we
could
actually
create
those
have
to
have
those
are
all
energies
in
cycle
through
it.
So
one
thing
that
is
important
and
that
we're
going
to
make
in
a
new
correction
of
the
graph
that's
going
to
go
out.
T
You
know
maybe
tomorrow
something
which
normally
we
want
to
say
that
we
can
have
several
of
these
FMOD
IDs
on
a
per
interface,
but
also
you
might
be
thinking
of
cases
where
you
may
be
using
one
specific
one
for
each
one
of
your
peers.
But
we
know
that
the
IDS
are
meant
to
have
limited
time.
Usage
or
else
is
just
become
it
degrades
to
their
normal
case,
where
you
have
a
long
living
identifier.
T
T
So
there's
no
big
changes
in
now
the
list
protocol,
the
graph,
is
actually
really
relying
on
existing
mechanisms.
The
graph
is
really
to
show
the
reservation
of
the
pool
and
the
current
evidence
recommendations,
and
so
we
would
like
to
ask
for
this
document
to
be
in
working
group.
Last
call:
it's
been
dead
for
quite
some
time
he's
been
pretty
stable,
it's
pretty
simple!
So
there
any
caution
comments
any.
G
K
You
know
review
process
now
the
SEC
is
being
added
and
really
that
is
sucking
out
all
of
the
air
from
the
room
and
I
mean
I
honestly
didn't
have
time
to
do.
The
proper
review
of
this
document
and
I
will
not
have
time
for
the
next
three
months.
I
suspect.
So
my
suggestion
is
really.
Can
we
wait
one
cycle
so
that
you
know
we
can
focus
all
the
energy
of
the
group
into
you
know
trying
to
push
this
thing
through
and
then
going
forward.
K
Let
me
add
one
more
thing,
so
one
nice
effect
of
this
is
pain
will
be
going
through.
Is
that
now
there
are
at
least
a
couple
of
guys
in
the
second
year
Ben
and
Eric
that
are
knowledgeable
of
list,
so
I
think
this
in
time
will
come
back
because
they
now
understand.
Well,
that
is
protocol
and
they
have
clearly
a
very
security
ever
strong
security
background.
So
I
think
that
if
we
take
a
little
more
time,
we
we
can,
you
know,
focus
more
energy
on
this
one.
That's
my
so
also.
T
I
hear
you
guys
and
actually
I
want
to
say
thanks
for
bulldozing
the
way
for
us
later,
but
you
have.
We
actually
was
going
to
ask
for
this
for
last
call
in
Bangkok
and
that
I
hold
it
one
cycle
by
myself.
So
this
is
the
second
cycle
we're
holding
it.
So
I
want
to
get
this
priority
when
you
come
back,
though,
that
would
be
the
only
thing
I
would
ask.
I
know.
G
We
proceeded
in
this
way,
so
we
wait
for
the
this
document
and
Lisp
sack
to
go
through,
and
then
we
start
to
move
forward.
The
other
documents
doesn't
mean
we
have
to
wait
until
more
real,
hopefully,
hopefully
up
at
the
same
time.
If
we,
if
we
stand
by
a
little
bit,
we
hold
on
a
little
bit,
it
means
as
well
documents
like
the
young
model
we
will
hold
on
as
well.
We
can
work
on
the
young
doctor.
We
will
ask,
but
I
mean
it
depends
on
the
energy
of
the
work.
H
Is
you
know
so
Fabio
there's
no
protocol
changes
to
this
at
all,
and
basically
a
host
today
can
choose
any
idea
wants
to
and
the
xtr
learns
about
it.
The
xtr
has
no
idea
if
it's
a
random
number,
that's
being
generated
an
allocated
address
from
I
ana
or
the
registries
or
whatever.
So
I
mean
it
does
specify
that
this
Eid
could
be
a
crypto
yi
ID,
a
hash
of
a
public
key
and
that
stuff's
that
more
complicated
stuff
is
put
in
another
document.
H
That
has
runs
its
own
course,
but
this
is
just
simply
saying
that
a
host
could
use
any
Eid
at
once
and
change
it
as
much
as
it
wants
all
the
machinery
lists.
Pastrini
doesn't
know
this
is
going
on.
It's
just
a
configuration
sort
of
thing,
so
it's
a
really
trivial
to
review
right
now.
The
the
security
area
will
probably
say
is
frequently
changing
Eid
secured
enough
or
do
they
need
to
be.
You
know.
G
But
those
are
my
work
in
there
correction.
We
have
two
minutes
what
I
propose?
What,
if
you
want
to
gain
time?
What
I
propose
is
what,
if
we
ask
for
a
security
review
right
away
of
the
document,
so
that
we
are
sure
that
once
we
go
over
the
last
call,
we
don't
have
a
any
issue
afterwards.
So
life
is.
G
G
To
wait,
anyway,
what
you
are
saying
we
have
still
to
wait
anyway,
because
there
is
an
interdependence
between.
M
H
D
H
H
G
Most
reasonable
way
to
move
forward,
in
my
opinion,
because
of
dicks
comments
is
we
asked
for
a
security
review
so
that
we're
sure
that
once
we
move
it
to
the
year,
the
is
she
we
don't
get
stuck
in
in
endless
their
discussion
about
a
security
flaw.
We
did
with
the
beast
documents
and
we
really
have
one
minute.
Please
go
ahead
so
I
understand
everyone's.
J
D
J
K
T
G
The
time
is
over,
so
thank
you
for
being
here
and
that
about
the
these
last
three
minutes.
What
I
gather
is
we
will
move
the
document
further
forward
in
the
sense
that
I
will
ask
for
that
security
review
that
we
will
not
get
stuck
afterwards
at
the
same
time.
At
the
same
time,
at
the
same
time,
we
don't
go
for
working
with
plus
call
right
away.
Okay,.
K
S
G
X
G
K
G
A
little
bit
and
wouldn't
be
a
bad
idea
if
we
give
time
to
other
people
to
get
accustomed
it
with
lists
so
that
they
can
review
the
security
part
I
mean
we
cannot
rely
only
on
Ben
and
Eric.
Otherwise,
we
will
be
always
stuck
under
on
the
pipeline
of
these
two
guys.
Has
this
been
presented
to
security.