►
From YouTube: IETF101-MODERN-20180319-0930
Description
MODERN meeting session at IETF101
2018/03/19 0930
https://datatracker.ietf.org/meeting/101/proceedings/
A
D
C
A
G
A
B
B
J
A
Guess
I
still
need
a
note
keeper
if
the
blue
sheet
should
be
going
around
if
anybody
could
take
notes.
Otherwise,
I
guess
I'll
go
watch
a
video
after
this
and
capture
the
notes.
This
is
the
new
note.
Well,
so
it's
a
little
different
than
the
one
you
saw
this
morning,
which
was
the
old
note.
Well
so
maybe
some
interesting
reading.
B
J
I
There's
a
lot
of
brave
people
here,
you
guys
are
all
sure
you're
all
committed
now,
because
we
are
gonna
seal
the
doors
at
this
point
and
numbering
is
coming
for
you
all
and
it
isn't.
Just
me
me,
Chris
is
actually
gonna.
We
should
tag-team
a
bit
on
some
of
the
slides
we're
doing
here
for
a
Terry
trip,
but
first
we
have
a
great
announcement.
This
group
accomplished
something
it.
I
It
is
emitted,
a
document
that
has
been
approved
by
that
August
body,
the
internet,
Engineering
steering
group
and
that
I
will
work
with
Garcia
to
get
this
completely
finalized
at
the
last
minute.
Of
course,
the
is
she
had
some
actually
very
helpful
feedback
on
the
document
we
did
add
on
a
privacy
considerations
section
that
is
completely
new.
That
basically
talked
about
the
fact
that,
yes,
we
are
creating
in
modern
we're
proposing
to
have
these
these
databases,
these
services,
they're
gonna
house,
these
records
and
well.
There
are
some
privacy
implications
of
that.
I
These
things
may
have
more.
Who
is
like
data
I
guess,
then
we
typically
see
and
telephone
style
databases
today
and
we're
going
to
create
new,
centralized
places
where
that
could
potentially
be
accessed
by
adversaries,
and
so
we
put
in
some
some
language
that
was
very
clear
about
what
what
we
thought.
The
risks
were
anyway,
so
caveat
emptor.
If
this
ever
is
no
longer
a
science
project,
it
is
actually
something
people
want
to
do
in
the
world.
God
forbid,
and
there
are
also
numerous
small
editorial
fixes,
which
we
were
grateful.
I
I
think
we
got
we
got
most
of
them
in
I
know
been
near
comment
about
whether
or
not
I
PBX
is
had
already
had
this
capability
to,
like
you
know,
dynamically
allocate
numbers
to
phones,
I
think
I,
think
there's
something
slightly
different
about
this
than
that
so
but
they're
for
the
most
part,
you
say
I
think
we
put
in,
like
virtually
all
of
that,
so
one
thing
I
took
away
from
that
conversation.
There
were
a
lot
of
comments,
I
think
from
ops
tier
among
others
that
you're
kind
of
about.
Well,
what
is
the
security
story?
I
I
think
we
have
the
curve
that
the
chorister
Doc's
out
another
win,
but
will
clap
about
in
another
working
group,
but
maybe
because
of
that
I
think
we
now
have
an
opportunity
really
to
show
how
those
credentials
could
be
leveraged
to
do
a
new
style
of
like
completely
decentralized
numbering
administration.
That
would
be
like
crazily,
distinct
from
the
way
things
work
today.
That's
our
science
project,
that's
mostly
what
we're
going
to
be
talking
about,
but
just
if
you're,
if
you
care,
we
do
still
have
an
opportunity
to
make
changes.
I
I
Darienne
drip,
so
this
is
like
our
new
document
for
this
time
it
seems
like
every
time
we
come
to
this
there's
like
some
new
use
case.
We
we're
being
told
for
like
years
when
we
were
starting
this
out.
Oh
my
god.
This
is
like
total
science
fiction.
Why
is
in
the
RTF?
There's
like
no
practical
use
for
this
at
the
moment,
and
so
we've
produced
this
parade,
I,
guess
of
use
cases
that
have
been
progressively
more
interesting.
I
From
my
perspective,
they
kind
of
show
what
we
might
be
able
to
do
with
this
and
again
now
that
stir
is
a
bit
more
concrete.
I
think
we
can
make
a
better
case
for
that
than
we
could
in
the
past
next
time.
So,
first
of
all,
we
just
have
like
acronym
soup
up
on
the
first
slide,
sir,
for
those
of
you
who
have
unwisely
against
my
repeated
entreaties
stayed
in
this
room
without
having
really
any
idea
what
it
is
we
talked
about
here.
I
We
have
two
documents
that
that
Chris
went
and
I've
been
working
on
what
one
is
called
Terry
Terry
is
it's
a
record
format.
We
kind
of
made
it
relatively
protocol
and
encoding
agnostic.
It's
really
more
of
an
information
model
than
anything
else.
That
kind
of
talks
about
what
we
think
Records
should
be
that
described
telephone
numbers.
I
We're
gonna
have
like
the
detail
slides
about
this
later,
but
think
about
it
like
one
of
the
things
that
makes
our
effort
here
and
modern
different
from
what
we
did
in
the
past,
with
enum
and
drinks
and
spearmint
and
so
on.
Is
we
kind
of
adopt
a
life
side,
life
cycle,
ecosystem
approach
to
thinking
about
numbering,
how
you
create
a
record
that
shows
the
acquisition
of
a
number?
How
you
can
then
manage
it
change
things
about
it,
change
the
administrative
and
service
data
associated
with
and,
finally,
how
you
can
resolve
that.
I
How
something
enum
like
can
go
kind
of
get
that
record
and
the
thing
that
is
different
from
all
the
previous
ways
we
did.
This
is
there's
now,
like
one
record
format,
that
all
these
things
use
the
various
the
various
other
ways
we
looked
at
this
in
the
past.
We
had
kind
of
well
here's
one
resolution
protocol.
I
It
has
this
syntax
and
semantics
to
it,
and
here's
this
provisioning
protocol
and
actually
they're,
like
not
aligned
so
having
like
one
record
thing,
was
kind
of
one
of
the
one
of
the
formative
principles
here
and
rip
is
work
that
Chris
did
Chris
you're
welcome
to
get
up
and
describe
drip
yourself
if
you'd
like
or
I
can
talk
about
it
a
little
bit
I.
Don't.
E
E
Conservative
in
terms
of
synchronization
of
data
and
making
it
complete,
it
contains
a
voting
mechanism
that
allows
all
the
nodes
in
the
network
to
validate
the
information.
It
has
a
voting
procedure
and
then
the
commit
procedure
as
well.
So
it's
meant
for
not
extremely
large
distributed
nodes,
but
you
know
reasonable
size
complimentary
to
what
we
would
typically
see
in
a
telephone
network.
I
Very
good,
so
we've
kind
of
taken
it
as
an
action
for
some
time
to
try
to
show
how
these
two
pieces
of
work
we're
doing
relatively
distinctly,
could
work
together
here
and
last
time
in
Singapore,
Chris
and
I
got
together
and
actively
some
slide
we're
about
this
understand.
We
have
a
draft
there's.
Actually
something
can
like
read
that
at
least
gives
use
cases
and
like
basic
ideas
behind
this
next.
K
I
You
know
a
lot
of
people
asked,
we
talked
about
Terry
and
its
operations
and
I
do
have
a
Terry
deck
after
this
all
at
least
run
through
a
bit
to
go
through
it.
But
there's
this
notion
in
Terry
that
there
are
simple
client-server
operations.
Like
you
see
here
between
the
client,
the
services
where
you
say
things
like
oh
I
need
a
new
telephone
number.
Can
you
allocate
b1
from
inventory
or
oh
I'm,
trying
to
place
a
call?
Can
I
talk
to
the
service
and
ask
if
there
are
any
service
records
associated
with
this
telephone
number?
I
What
we
definitely
talk
about
much
that
was
like
how
Terry
services
might
spread
information
to
each
other,
and
so
the
low-hanging
fruit
that
seemed
for
the
intersection
of
the
way
Chris
was
thinking
and
I
was
thinking
was.
We
would
use
this
nice
gossip
protocol
to
actually
is
kind
of
this
server
to
server
protocol
for
this,
and,
if
you
imagine
that
there
were,
there
were
large
community
service
providers,
people
like
a
Comcast
and
AT&T
and
so
on.
They
might
have
a
service
or
maybe
multiple
service
instances
like
this.
I
That
would
have
records
in
them
and
they
would
spread
these
records
around
to
these
various
services
and
then
within
their
individual
networks,
it
would
kind
of
cache
all
of
the
valid
records
that
are
being
shared
in
the
gossip
network
at
all
of
the
nodes.
You
would
then
be
able
to
perform
that
the
tairy
operations,
acquisition
management,
retrieval
and
so
on,
to
take
care
of
that,
so
that
this
seems
to
be
like
the
the
simplest
way
to
describe
the
intersection
between
the
way.
Chris
and
I
have
been
thinking
about
these
things.
So
next
next
slide.
I
Now
that
the
end
goal
of
this
is
to
try
to
design
a
distributed
registry
for
those
of
you
who
again,
who
are
here
for
the
first
time
she's
some
years
ago
now,
three
years
ago,
maybe
there
was
a
meeting
at
the
FCC
and
in
Washington
United
States
to
talk
about
well,
what?
How
can
we
get
a
blue
sky
about
what
the
future
of
numbering
databases
could
be,
and
there
was?
There
was
great
deal
of
interest
there
and
doing
it
a
distributed
registry
as
an
alternative
to
the
centralized
TN
database
registries
exist
today.
I
These
are
things
like
the
log
and
the
impact
and
so
on,
but
but
to
be
clear,
although
that
is
one
of
the
motivating
problems
behind
modern,
this
is
not
proposing
to
eliminate
or
change
any
particular
existing
thing
that
is
being
deployed.
This
is
a
science
project.
There
are
a
bunch
of
ways
that
we
could
kind
of
kick
the
tires
on
this
that
we'll
talk
about.
This
is
not
like
ready
to
go
and
replace
anything
practical
at
this
moment
in
time.
So
there's
not
the
intention
of
this.
I
This
exercise
to
do
that,
we
want
to
try
to
define
something
to
be
suitable
for
experiments
in
actual
deployed
environments
without
actually
replacing
like
the
guts
of
anything
today
so
yeah
we
have.
This
is
drip,
providing
a
way
to
share
tarry
records
between
these
services.
The
most
interesting
use
cases
I
think
for
this.
You
know
we
talked
about
using
drip,
just
basically
as
a
a
transport
protocol
for
the
way
we
define
the
tairy
operations.
We.
I
A
mock-up
JSON
format
for
Terry
requests
and
responses
to
go
between
clients
and
servers
that
didn't
like
scene
I,
really
like
like
excite
me
in
any
way,
though
so
I
think
the
interesting
ones
are
all
ones
where
there
are
numbering
spaces
that
are
shared
by
multiple
authorities
and
that
we
assume
that
you
know
when
I
showed
that
picture
of
the
participants
and
the
drip
network
kind
of
at
the
middle
in
the
previous
slide,
and
that
they
all
share
some
authority
over
a
space
that
allows
them
access
to
a
naming
resource.
It's
in
them.
I
We're
going
to
talk
about
them
here
as
CSP
is
this
last
bullet
here
is
really
only
for
people
who
really
follow
Terry
closely,
there's
kind
of
a
distinction
between
a
CSP
and
a
registrar,
we're
gonna
gloss
over
the
purposes
of
this
discussion.
Think
about
the
CSP
s
like
they're
carriers,
people
that
would
own
numbering
resources
next
slide,
Oh
South
allocation
is,
is
the
one
that
I
get
excited
about
anyway,
when
I
think
about
how
this
could
work.
Now.
Imagine
a
world
where
you
have
multiple
CSPs,
who
all
have
some
kind
of
a
credential.
I
It
would
be
like
a
store,
credential,
I,
think
that
allows
them
to
sign
for
telephone
numbers
that
are
in
a
given
numbering
block,
but
initially
there
are
actually
no
numbers
in
that
range
that
are
allocated
to
any
of
them
all.
Those
csps
have
the
credential
that
allows
them
to
self
allocate
resources
within
that
block.
I
If
they
are
so
inclined,
it's
it's
perhaps
easiest
like
the
closest
thing
we
have
to
something
like
this
today
is
the
way
that,
like
free
phone
stuff
works
in
the
United
States,
where
for
800
numbers,
you
know
it's
different
from
ordinary
Geographic
numbers.
There
are
kind
of
these
organizations
that
exist
specifically
to
to
kind
of
acquire
an
800-number,
because
they're
particularly
want
maybe
its
cluster
like
a
domain
name
model.
I
If
you
think
about
it,
you
want
one
eight
hundred
United
one
or
something
right,
and
so
you
kind
of
go
to
these
businesses
that
acquire
this
for
you
and
then
a
carrier
will
serve
that
particular
number
on
your
behalf
bit
of
an
oversimplification.
But
there
are
these
things
called
rest
words
that
do
that
you
can
kind
of
think
about
it,
like
the
rest
works
today.
Have
access
to
this
free
phone
number?
I
They
can
kind
of
come
in
almost
a
first-come,
first-serve
basis
and
be
like
actually
I'd
like
to
buy
that
number
in
this
space.
In
order
to
serve
this
particular
customer,
and
what
we're
looking
at
here
are
ways
we
could
automate
that
process
to
allow
something
similar
to
self
allocation
where,
if
you
just
say
okay,
this
resource
is
available
provided
I'm,
one
of
the
entities
that
has
been
vetted
and
has
the
proper
credentials.
I
can
just
create
a
tarry
record,
push
that
out
and
say
I'm
now
the
authority
for
that
that
number.
I
If
everyone
in
the
gossip
network
agrees,
then
you
are
I,
think
these
are
fascinating
to
the
cases.
So
it's
a
sort
of
next
slide.
Yeah
I
mean
in
all
these
cases
whether
we
assume
that
CSP
is
going
to
be
allocating
to
itself
a
block
or
a
single
number.
It
works
pretty
much
the
same
way.
This
would
be
something
where
the
CSP
has
gone
through
some
vetting
process,
and
this
is
the
stuff
we
do
in
Acme,
where
we
make
policy
decisions
to
decide
what
kind
of
credentials
you
can
get
once
you
have.
I
I
You
sign
that
then,
with
it
with
this
credential
and
provided
that
the
peer
nodes
in
your
gossip
network
trust
the
credential
the
views
for
this
and
the
trust
anchor
for
it,
then
it
just
becomes
valid
it
propagates
through
the
network
and
gets
cached
everywhere,
and
that
that
is
now
a
record
that
is
eligible
for
for
someone
to
to
retrieve
and
to
use
as
either
administrative
or
service
record.
We
do
have
this
concept,
and
we
have
a
picture
of
this
as
well
that
it
could
be
in
order
to
enforce
particular
policies
about
self
allocation.
I
We
would
have
these
these
policy
nodes.
We
even
have
a
take
a
picture
of
a
policy
node,
basically
something
whose
job
is
to
listen
to
what's
going
on,
and
even
if
it's
not
like,
oh
you're,
trying
to
allocate
a
number
it's
already
allocated
to
me
and
I,
don't
like
that,
instead
you're
just
allocating
a
lot
of
numbers,
maybe
yeah
I.
I
Would
the
policy
note
would
start
voting
no
right
because
it's
there
to
enforce
some
policy,
that's
in
the
distributed
registry,
but
the
idea
is
to
eliminate
this
kind
of
single
centralized
point
you
have
to
go
to
that
is
keeping
track
of
everything
and
to
gotta
get
its
permission
to
allocate
any
number
to
yourself,
and
instead
you
allow
something
that
is
more
more
anarchic,
I
guess
something
that
doesn't
have
that
requirement
for
a
centralized
authority
in
it
Chris
do
you
want
to
add
anything
to
this?
This
is
this
is
I'm.
E
You're
reading
my
mind,
yeah
no
I
think
the
other
two
benefits
that
we've
talked
about.
Is
it's
a
much
more
real-time
process,
so
everything
gets
distributed.
You
know-
and
you
know
hopefully
milliseconds
but
maybe
seconds
at
worst-case,
and
then
the
was
the
other
one.
I
was
thinking
of
I
just
point
on
it:
oh
yeah,
the
other
one
is
having
a
local
copy
of
the
registry.
You
know.
Obviously,
when
you
do
calls-
and
you
want
to
query
databases-
you
don't
want
to
go
across
that
works.
E
I
You
tend
to
have
to
like
go
to
a
centralized
registry,
be
like
hey
I,
want
to
make
this
change,
and
that
centralized
registry
then
replicates
that
database
down
to
all
of
these
nodes
that
exist
in
all
the
CSP
networks
that
are
then
access
to
make
routing
decisions,
and
things
like
this
is
this
guy
short
circuits
that,
where,
instead
of
you
having
to
go
up
to
that
process,
and
then
have
it
come
back
down
to
everybody,
you
just
blurted
it
out
and
provided
that
it's
yet
everybody
goes
long.
Next
slide.
I
There's
just
this
concept!
There's
this
distinction
allocation
in
the
assignment,
that's
significant
in
the
telephone
network
and
again,
if
you're,
not
deep
in
the
weeds
of
this,
this
may
seem
kind
of
arbitrary.
But
this
is
just
the
difference
between
you
know:
a
carrier
having
like
a
number
in
inventory
like
acquiring
a
thousand
block
and
it
actually
some
individual
number
of
being
a
lucky
to
a
consumer,
and
today
a
lot
of
decisions
are
made
about
whether
or
not
to
grant
carriers
additional
inventory
based
on
the
actual
percent
of
allocation,
and
so
on.
I
It's
in
that
that
blocks
they've
already
been
issued,
and
so
we
created
this
mechanism.
It's
actually
in
a
draft
called
Terry
valid
that
lets.
You
see
not
only
that
any
given
number
in
the
numbering
space
is
a
real
number,
but
that
it's
been
allocated
and
or
assigned
down
to
some
individuals.
There's
kind
of
a
few
states
that
we
identified
in
that
draft
and
the
process
here
is
very
similar
like
if,
if
a
particular
CSP
like
say
Comcast
has
a
block
allocated
to
it
and
then
wants
to
move
a
particular
number
into
a
state
of
assignment.
I
That's
just
the
generation
of
a
new
administrative
Terry
record
that
gets
pushed
out
through
the
network.
Everybody
ends
up.
Caching,
it
again.
Maybe
somebody
like
a
policy
node
is
in
the
gossip
pepper,
just
to
listen
to
those
assignment
records
going
through
to
be
able
to
say:
oh
I
am
tracking
this
person's
inventory
they're
actually
using
this,
so
when
they
then
need
to
go
and
say,
I
need
a
new
thousand
block.
Next
week
you
have
some
notion
of
whether
or
not
they're
making
responsible
use
of
the
namespace
next
time.
I
The
most
interesting
of
these
are
porting
cases.
That
is
a
case
where
a
number
is
moving
from
the
control
of
one
CSP
to
another
CSP.
These
are
cases
we
think
that
can
be
resolved,
though,
and
a
network
like
this
pretty
simply
or
basically
that
the
new
network
just
issues
a
record
saying
actually
that
telephone
number
of
longs
to
me
now
and
if
the
old
network
agrees,
then
it
doesn't
vote
no
in
the
gossip
network,
very
simple,
you
know
they're
there.
There
are
a
lot
of
hard
edges
on
this.
I
I
I
mean
I,
guess
the
real
question
we
have
for
the
group
in
terms
of
things
we
could
do
at
the
science
project.
It
is
a
science
project.
You
know
I
wish
Henning
was
here.
He
really
kind
of.
He
was
the
quest
giver,
who
put
us
on
this
quest
in
the
first
place
back
in
the
day.
We
think
this
is
a
reasonable
kind
of
direction
to
push
forward.
Given
what
the
problem
space
is,
we
think
we
can
run
these
use
cases
down
and
do
something
practical
with
them.
We
think
it's
relatively
implementable
again.
K
I
Yeah,
it's
just
a
picture
of
the
policy.
Men
probably
should
just
have
that
earlier
so
anyway,
next
slide,
so
yeah
I
think
that's
just
a
bit
closer
to
an
architecture
of
how
Terry
and
Jim
could
work
together.
Last
time
we
showed
up
we
kind
of
had
some
pie
in
the
sky.
You
know
thinking
about
it.
This
shows
some
use
cases.
This
show
some
things
we
think
practically
could
try
to
do
with
the
building
blocks
that
we
have
in
place
here.
We'd,
like
some
feedback.
This
is
just
Chris
and
I
like
you're
gonna
surround
right.
I
So
if
other
people
are
interested
in
this,
you
know,
of
course,
all
depends
on
we
having
it
us
having
a
green
light
for
Terry
and
rip
themselves,
which
are
still
not
remember,
biosphere
still
just
kind
of
things
that
have
been
floating
here
as
we've
been
waiting
for,
stir
to
wrap
up,
and
we
need
to
get
the
modern
framework
through
the
system,
but
we're
now.
You
know
at
a
place
where
there's
probably
some
bad
method
to
do
this.
So
what
do
people
think
cuz
this
so
I'm
almost
interested
for
the
daring
people
that
stayed
here?
I
I
D
E
Yeah
so,
like
John
said
I
would
echo.
This
is
more
I,
think
it's
getting
fairly
mature
in
its
thoughts.
It
also
seems
to
align
very
well,
like
John
said,
with
the
sort
of
graduation
of
stir
into
something
with
a
framework
where
we
can
apply
credentials
as
well
as
some
other
industry
things
talking
about
new
networks,
potentially
all
IP
type
of
networks.
There
seems
to
be
a
lot
of.
E
I
Thank
you
yes,
yeah
I,
don't
have
it
on
me,
but
so
no
I
mean
I.
I
do
think
that
there
there
is
potentially
room
for
like
a
ledger
to
play
a
role
in
this
and
something
Chris
and
I
have
actually
talked
about,
but
again
the
way
that
we
did
these
these
these
records
are
being
issued.
They
you
know,
because
of
exactly
the
problems
I
was
describing
of
you
know.
How
do
you
each
out
the
old
records
and
things
like
that?
A
ledger
actually
makes
a
lot
of
sense
for
some.
The
aspects
of
this
yeah.
E
N
More
seriously,
the
question
of
whether
you
want
a
gossip
protocol
or
some
other
kind
of
I
think
really
depends
on
the
threat
model
you
have
available
to
me.
My
sense
of
the
threat
model
you
have
is
that
people
who
are
doing
this
we're
largely
trusting
each
other
as
actors.
If
not
then,
like
you,
only
need
something
much
more
aggressive
at
the
end
to.
E
I
Well,
you
know
I
mean
I
I
feel
obliged
when
we
talk
about
these
things,
like
you
know,
I
can
come
here
and
talk
about
this
like
time
after
time
at
the
IDF
I
I
would
I'm
looking
around
of,
like
usual
suspect
carrier
people.
That
I
would
hope
to
see
in
a
room
like
this.
We're
not
here
so
so
you
know
I'm
a
bit
on
the
fence.
Right
I
mean
I'm
happy
to
do
this
as
a
science
project.
I
I
Okay
good
to
know
so.
This
is
why
we
could
give
away
time
by
the
way,
because
you
happy
it's
just
like
me
talking
here
for
I've
like
15
slides,
so
we're
almost
done
now.
I
didn't
watch
a
bit
more
about
Terry,
though
in
the
sense
that
I
did
make
a
couple
changes
to
it
and
I'm
not
gonna
I
have
like
all
the
Terry
tutorial,
slides
here,
I'm,
actually
in
a
bother
running
through
them
today,
but
Nick
next
slide.
I
I
We
now
have
this
up
to
a
zero,
for
it.
I
think
there's
still
a
lot
in
it
that
we
would
need
to
flesh
out
if
we're
going
to
use
it
for
any
particular
application,
because
it
is
like
I
said
it's
mostly
an
information
model.
We've
mocked
up
some
encodings
for
it
and
things
like
that.
But
it's
it's
not
something
that
we
have.
You
know
tried
to
produce
what
I
would
describe
as
an
implement.
I
A
okay,
I'll
say
this
much
so
the
the
the
the
main
change
we
made
in
this
in
this
largely
resulted
from
the
work.
We're
looking
at
with
drip
is
to
kind
of
make
the
Terry
record
and
Siri
operations
slightly
more
to
couple.
That's
the
main
change.
That's
in
this.
It
used
to
be
we
kind
of
led
talking
about
the
Terry
operations,
about
the
ways
that
you
acquire
records.
I
The
way
that
you,
basically
the
client-server
operation,
part
of
that
and
I-
think
we're
now
gonna
the
architectural
influence
architectural
direction
of
the
draft
to
be
much
more
about.
Okay,
here's
the
records
and
we
could
even
push
the
operations
into
ultimately
a
separate
specification.
I
think
what
we
want
to
say
is
the
core
of
Terry
is
the
records,
because
that's
a
component
that
would
be
used
by
trip
in
a
very
different
way
that
when
we
be
used
in
the
kind
of
client-server
cases
that
we're
describing
same
records,
that's
that's
our
defining
principle
here
in
modeling.
I
We're
gonna
use
this
single
record
format,
they're,
always
going
to
be
signed
by
the
same
kind
of
credentials,
but
that's
the
only
real
change
that
we
made
to
the
Terry
model
since
last
time
and
I
mean
I
can
give
a
Terry
tutorial
and
advertisement
of.
Why
I
think
it's
interesting
for
like
the
57th
time
here,
but
I
I
honestly,
don't
want
to
bother
so
flip
it
back
all
the
way
just
go.
Go!
I
Go
quick,
I'll
pretend
this
is
like
when
those
Peka
Peka
creature,
things
yeah
except
I'm,
not
gonna,
say
anything
and
just
it's
gonna
go.
Why
keep
going?
It's
got
acquisition
management,
retrieval,
there's
queries.
There's
responses,
there's
great
stuff
in
it.
We
had
done
in
this
lot
like
the
same
questions.
You
know
if,
if
people
think
that
some
of
the
use
cases
we've
defined
here,
things
like
nationwide
number
portability,
there
are
people
at
Addis
talking
about
that.
That
potentially
would
use
this
or
the
publication
of
the
list
of
valid
numbers.
I
That's
an
interesting
enough
use
case
to
warrant
just
going
the
next
step.
Just
saying:
okay,
we've
messed
around
with
this
for
a
bit.
We
think
we
understand
why.
This
is
useful
and
important.
Should
we
make
this
a
working
group
item
and
try
to
flesh
it
out
a
bit
more?
That's
that's
the
only
everyone
notice,
guys
and
I
think
the
answer
is
no.
Like
look
man,
we
got
other
stuff
to
do.
There's
like
a
ton
of
other
things.
I
We
need
to
do
in
some
sense,
getting
the
framework
document
out
that
we
talked
about
how
it
had
just
passed,
being,
invest
that
forms
a
bit
of
a
overall
architecture
for
how
we
think
things
like
stir
and
Acme
fit
together
and
there's
there's
some
value
in
having
that
framework
independently
of
ever
doing
any
subsequent
work
towards
this.
But
this
is
a
way
to
approach
registries
that
again,
I
think
provides
a
much
richer
set
of
services
than
anything
we
have
for
dealing
with
cellphone
numbers
in
the
internet.
I
N
N
A
good
addition
like
look
like
but
I'm,
not
like
a
carrier,
and
you
know
like
I'm,
consuming
this
analogy,
so
you
know
if
like,
if
you
tell
me-
or
someone
tells
me
that
likely
porterson
doing
this,
you
know
I'm
your
sin
helping,
but
you
know
I'm
not
interested
in
pushing
you.
Then
nobody
wants.
So
no,
no.
We
will
deploy
whether
that's
what
the
customers
want.
It
or
not.
N
I
O
Jeff
Hodges
PayPal,
so
we
are
coming
to
the
conclusion
that
doing
security
sensitive
operations
via,
for
example,
SMS,
is
not
going
to
go
away.
People
are
going
to
keep
doing
that.
In
fact,
it's
probably
going
to
get
even
more
down
the
road
in
the
future,
and
we
would
really
like
to
see
those
be
able
to
be
done
to
use
SMS
in
a
secure
fashion.
I
I
This
is
much
more
about
things
like
how
do
you
get
a
telomere,
the
first
place,
assuming
a
world
where,
like
databases,
are
really
difficult
or
where
people
are
transitioning
away
from
traditional
PSTN
structures,
and
so
you've
got
to
be
able
to
do
like
something
to
manage
what
the
routes
are
associated
with
numbers?
And
things
like
that,
so
this
is
the
problem
is
that
those
things
are
actually
I.
Think
we
got.
We
got
that
down.
This
is
like
the
way
back
office,
okay,
yeah.
O
O
I
O
O
De
New
York
and
in
a
previous
number
of
years
ago,
I
worked
in
an
environment
where
this
was.
This
kind
of
thing
was
somewhat
critical
in
a
space,
but
I
guess
my
question
for
you
to
answer
what
you're
asking
of
this
group
is
sort
of
echoes.
What
echo
is
saying:
are
there
people
looking
to
deploy
this
now
or
are
we
building
something
for
the
time
when
we
think
people
might
want
to
deploy
it.
I
So
this
was
this
so
remember,
this
started
at
the
SCC
as
a
please
blue
sky
or
the
IP
transition
is
coming.
We
know
that
the
traditional
esta
seven
structures
are
going
to
go
away
at
some
point.
We,
what
would
it
be
like
if
we
just
had
to
design
this
from
the
ground
up
today,
so
that
it
would
work
well
with
the
internet?
Then
obviously,
stur
is
a
part
of
that
actually
heard
of
that,
but
this
is
like
we're
all
the
framework
II
like
back
osseb
back
in
this.
I
N
I
I
There
are
things
that
are
that
are
easy
to
do
with
Terry,
like
that,
like
the
valid
number,
let's
write
to
say:
okay,
here's,
here's,
here's
all
the
the
you
know
any
number,
that's
in
the
North
American
Numbering
plan
that
is
valid,
that's
going
to
fit
within
the
following
constraints,
and
so
any
call
you
see
that
it's
from
a
number
other
than
this.
Probably
you
don't
need
to
about
that.
O
I
Much
more
like
the
Terry
drip
stuff,
they
were
just
talking
about
yeah,
where
they
can
get
very
dynamic,
a
real-time
in
their
allocation.
You
wanna
be
able
to
do
that
too.
But
the
point
is:
that's
not
the
thing
that
I
think
you
know
agencies
energized
about
right
like
which
is
why
Martin
is
somewhere
else.
This
working
all
right
thanks.
L
Mary
boys
so
I
mean
I
think
this
is
it
a
good
place
now
I'm
starting
to
get
into
your
space
and
all
the
numbering
stuff,
and
it's
it's
pretty
crazy.
What
has
to
go
on
behind
the
scenes
to
manage
this
stuff
right
now,
I
mean
it's
been
saying
so
I
mean
in
the
long
term.
I
think
this
is
a
good
idea,
but
it
is
ways
ways:
okay,
yeah.
E
So
I
think
this
is
a
ripe
area
for
innovation
to
get
to
the
point
where,
if
you
believe
in
the
future
of
telephone
numbers
and
telephone
number
of
management
and
how
they
are
managed
and
the
ability
to
dynamically
allocate
things
which
I
think
you
know,
assuming
that
they're
relevant
going
forward,
that's
a
very
important
area
to
to
so
just
sort
of
trying
to
give
some
insight
into
why
this
work
is
important.
Going
forward.
I
Well-
and
you
know
there,
there
are
a
bunch
of
people-
are
trying
to
figure
out
things
like
how
you
do
and
then
P
divided
number
portability
or
the
number
lists,
and
you
know
if
we
rethink
that
we
have
the
good
document
formats
for
that
right
in
Terry.
Those
are
much
more
low-hanging.
Fruit
than
doing
dynamic.
Number
relegation
like
in
Jerry
drift,
and
things
like
that,
so
I
mean
I.
I
P
P
I
Applause.
Thank
you.
A
I
I
I
Well,
the
hum
I
would
like
is
people
you
know
come
if
you
would
like
to
you
know,
adopt
Terry
is
a
baseline
to
work
on.
You
know,
there's
some
more
practical
problems,
some
more
science
fictiony
long
term
problems,
but
hum
if
you
would
like
to
adopt
Harry
as
a
baseline
for
the
record
for
math
of
the
group
and
then
take
home
against,
and
if
there's
any
noise
in
the
room
on
either
of
those,
we
can
take
some
indication
of
what
people
think
right.
M
I
I
L
I
Mean
so
so
should
should
I
issue
a
version
of
this
that
says
draft
ITF,
honor
I
will
take
it
to
the
list,
and
then
we
can
do
so
yeah.
Okay.
That
at
least
would
be
a
step
forward
on
this
to
a
place
and
yes,
help
from
anybody
who
wants
to
actually
help
flesh
this
out,
or
it
would
be
great
because
it
is
still
is
it's
pretty
much
just
an
information
model
at
this
point,
so,
okay,
you
have
five
minutes.