►
From YouTube: IETF99-RADEXT-20170719-1330
Description
RADEXT meeting session at IETF99
2017/07/19 1330
https://datatracker.ietf.org/meeting/99/proceedings/
A
A
B
B
So
yeah
first,
it's
already
Wednesday.
So
ever
you've
probably
seen
this
there's
a
note.
Well,
please
note
well
the
note
well
and
then
we
have
a
very
light
agenda
being
a
volunteer
for
the
minute
taking
that's
you
know.
We
don't
need
a
Travis
cry,
because
there
is
nobody
on
me
taking
on
Java
I
think
they
don't
know
is
pretty
okay.
Unless
somebody
wants
to
add
something:
oh
okay,
then
we
can
take
a
short
look
about
what
we
did
in
the
last
couple
of
months
and
we
have
a
discussion
about
one.
B
So,
let's
first
start
with
the
overview
of
what
happened.
We
didn't
meet
that
ITF
Emma.
We
didn't
meet
at
ITF
98.
So
these
are
the
changes
from
ITF
97.
We
published
two
RFC's,
the
radius
extensions
for
IP
port
configuration
in
reporting,
RC
8045
and
data
types
and
radios
at-44
for
a
change.
We
have
nothing
in
the
editor
view,
nothing
in
ITF,
NASSCO,
nothing
is
processing.
We
have
one
document
currently
still
active
awaiting
the
Shepherd
right
up,
that
is
the
CEO
a
proxy
from
another
Kok
and
Hume
Cronyn.
B
B
There
is
one
working
Robottom,
which
is
in
principle
in
progress,
that's
mine
on
the
correct
use
of
the
response,
identity
field.
To
be
honest,
not
many
people
asked
me
about
it
somewhere,
even
opposed
to
trying
out
multiple
identities,
so
actually
I'm
prepared
to
drop
this
one.
The
benefits
are
small.
If
you
write
an
each
subsequent,
there
is
some
advice
in
there.
B
You
might
want
to
read
it,
but
it's
the
kind
of
document,
which
is
also
perfectly
fine
as
an
expired
draft
to
point
to
and
say
look
there
is
some
advice
in
the
air
and
that's
it
proposals
for
new
work.
We
had
the
idea
to
move
radios
TLS
from
the
current
experimental
status
to
step
it
back,
which
is
worked
by
Jim,
shot
I've
seen
him
around
here
earlier
today.
B
We
don't
have
a
draft
yet
we've
had
some
conceptual
discussion,
ok
and
that
discussion
was
I
think
about
a
year
ago,
so
I
see
this
work
is
also
pretty
much
stalled,
a
shame
to
drop
that
one,
because
registry
less
is
being
used.
People
are
asking
about
it,
they're
asking:
why
is
it
experimental?
So
it
would
really
be
good
to
get
this
one
done,
but
well
somebody
has
to
do
the
work
and
test
volunteer.
B
So,
on
the
other
two
things
which
are
currently
being
actively
discussed
in
the
working
group
are
the
radius
extended
identifiers
attribute
from
Anka
Chen
and
mr.
Chen.
We
have
an
a.1
draft,
it's
one
of
two
approaches
to
get
this
done,
so
we
will
discuss
this
today.
Actually
the
authors
cannot
make
it
to
the
meeting,
so
they
asked
me
to
run
the
slides
and
I
guess
I'll
do
that
and
then
we
can
discuss
about
it.
Despite
me,
not
very
much.
B
The
second
approach
to
the
same
problem
is
a
draft
from
another
who
has
also
submitted
to
revision,
so
these
two
will
be
treated
as
one
block.
Lastly,
Ellen
would
like
to
remind
about
the
proxy
considerations
document
he
is
intending
to
write.
So
this
is
just
a
reminder
in
case
someone
is
interested,
let's
see
if
we
find
lots
of
interesting
people
and
that's
basically,
all
we
have
on
our
wake
up.
Specifically
the
laura1
people
were
here
once
as
if
interested
it
was
pretty
much
straightforward
radius
related
work,
but
they're
not
coming
anymore.
B
Okay,
that's
it
for
document
overview.
You
see.
We
have
only
very
few
things
to
do
so.
The
one
thing
I
did
was
trying
to
do
the
shepherd
right
up
for
the
seer,
a
proxy
document.
This
had
gone
through
working
with
basketball,
I
think
couple
of
times
I.
Don't
nobody
had
any
issues
anymore,
so
I
thought
no,
it's
the
time
to
actually
get
it
done.
It
actually
reads
nicely
so
I'm
quite
happy
with
the
document,
but
for
some
reason.
B
So
yeah,
but
as
I
read
the
document,
despite
several,
what
people
ask
calls
the
work
request.
Called
scrutiny
must
have
missed
that
there
was
a
section
which
had
one
word
as
a
content
which
is
stuff
and
for
some
reason,
I
thought
this
probably
doesn't
pass
for
a
clearance,
so
this
draft
has
to
go
through.
In
other
words,
the
thing
is:
if
a
graph
gets
that
far
with
this
word
stuff
as
content,
can
we
really
say
we
have
working
group
process
here?
B
I
mean
many
people
are
supposed
to
read
the
document,
and
at
least
one
of
them
should
realize
a
there's,
a
chapter
missing.
So
this
is
the
first
hint
that
this
is
not
a
very
healthy
working
group.
To
be
honest,
so,
let's
see
about
the
last
agenda
item
the
future
of
radix.
What
did
you
about
it.
B
Right
so
the
next
thing
to
actually
discuss,
then,
is
the
two
approaches
on
what
to
do
about
read
the
radius
limitation
that
you
can
only
have
256
packets
in
flight
between
the
client
and
the
server
simultaneously,
which
is
not
an
issue
for
some
other
departments,
which
is
an
issue
for
large
departments.
We
currently
have
two
drafts
talking
about
that.
So
let
me
first
bring
up
one
of
those.
B
B
This
had
some
allergies
in
the
people
on
the
list,
so
they
have
now
shrunken
their
proposal
to
define
one
new
attribute
for
the
extended
Aggie
all
changes
to
the
packet
header
on
so
on.
This
is
now
still
a
standard
radius
packet.
The
idea
is
that
there
is
an
extra
attribute,
the
packets.
So
despite
the
identifier,
you
also
have
a
new
attribute
which
gives
you
another
another
space
to
identify
packet.
Of
course,
the
issue
with
that
is
that's
both
the
sending
sending
site
has
to
edit
at
the
receiving
site
has
to
understand,
attacked
up
on
it.
B
So
this
is
something
to
signal
and
yeah
there's,
basically
some
signaling
messaging
in
the
state
to
serve
a
message
here.
So
the
attribute
itself
is
pretty
straightforward.
The
steadiest
server
can
be
used
to
negotiate
whether
or
not
a
given
server
actually
supports
and
understands
this
incoming
attribute,
and
that,
basically,
is
it.
B
Yeah,
so
there
is
a
patch
to
free
radios,
things
actually
work,
that's
pretty,
okay,
of
course.
The
capability
new
discovery
is
something
that
already
has
never
had
and
we
have
backporting
it's
every
single
time.
We
want
to
get
that,
get
something
new
done
in
radius
and
yeah
I.
Guess
it
kind
of
works.
I
I
see
that
it's
about
190
lines
of
code
in
C.
So
it's
not
the
big
change
that
rewriting
the
radius
packet
head
would
be
much
more
gentle.
So,
okay,
the
authors
say
it's
a
simple
proposal
and
simple
documentation.
B
Only
few
code
changes
at
a
small
place,
identifier
management
can
be
trivial,
can
simply
use
a
monologue,
monotonically
increase
encounter
and
there
is
a
large
benefit.
It
removes
a
long-standing
protocol
limitation,
I
think
that's
also
people
say
their
spec
is
stable.
It
is
backward
compatible
because
we
use
data
server
to
negotiate
this
no
changes
to
the
packet
header
and
no
change
to
overloading
of
existing
fields
and
the
office
requests
by
expecting
group
adoption.
C
Brian,
wise
Cisco,
so
I'm
a
little
bit,
I
guess
and
maybe
have
an
opinion,
doesn't
work
with
the
same
companies.
Authors
but
I
really
haven't
been
following
their
work.
I
will
just
save
having
watched
as
going
from
what
they
had
to
this
very
simple,
accurate
extension
seems
very
reasonable
and
it
solves
a
problem
that
you
should
do
it.
B
Yeah
I
mean
the
group
found
very
harsh
words
in
the
first
approach,
so
this
one
is
actually
a
lot
better.
Definitely
it's
also
not
groundbreaking.
Personally.
I
know
that
the
ready
HR
people
have
this
it's
a
decade
or
something
so
they
have
an
extended
VSA
attributes
just
works
in
the
tracking
I.
Don't
have
the
state
of
self
negotiation,
but
even
then
things
seem
to
work
because
they
get
this
yeah.
Yes,
a
back
so
yeah,
it's
something
you
can
do.
B
D
B
D
B
Yeah,
okay,
so
since
there's
nobody
a
musical,
particularly
not
Ella
Dukakis,
who
wrote
the
next
slide
egg
I
guess
I'm
not
going
to
read
all
of
his
life
that
in
all
the
details,
because
it's
like
15
or
20
pages
16.
B
B
B
Yes,
so
on,
whenever
you
send
a
packet
in
radius,
yes
opposed,
or
you
should
have
a
request
of
educator
in
the
end
of
the
packet
meaning
you
can
verify
the
authenticity
of
of
this
end
up
with
the
Shedd
secret.
As
it
happens,
if
you're
sending
the
set
back
up
twice,
you
will
also
end
up
at
the
same
request
authenticator.
So
as
we
need
to
change
something
else,
you
get
a
different
professor.
B
We
can
stop
educator
if
you
check
these
two
sentences
in
one,
you
could
say
right
if
I
receive
a
packet
with
the
same
ID
by
the
different
request.
Authenticator.
This
is
not
a
retransmission.
Obviously
something
else
has
changed,
so
this
is
no
duplicate.
I
will
just
process
it
as
normal
and
then
I'm
done
so.
The
question
of
duplicates
just
doesn't
exist
anymore,
just
because
you
do
have
more
bits
to
identify
a
packet
at
the
end.
It's
an
Bakula
deal,
but
then
it
has
the
side
constraints
that
you
have
to
have
a
request.
B
Authenticated
the
original
radio.
Spec
didn't
work.
You
require
this,
so
you
might
get
an
access
request,
having
done
for
some
reason
back
in
the
day
two
decades
ago,
people
didn't
think
this
is
an
issue
these
days.
There
is
the
radius
issues
and
requested
or
suggested
fix
this
document,
and
it
says
you
should
still
do
that
also
in
access
requests.
If
you
do,
you
have
a
solution
here.
Of
course
we
do
have
a
large
deployed
legacy
and
we
don't
know
how
many
masses
actually
do
that,
even
though
we
say
they
should
do
that.
B
So
the
proposal
here
is
a
bit
rough
around
the
edges.
If
this
is
not
being
supported
by
the
NASS,
there
is,
however,
some
negotiation
here
as
well.
I
think
there's
also
stead
of
server
involved,
just
to
be
sure
what
is
coming
from
the
NASS
and
what
you
can
expect
from
it.
So,
in
the
end,
yeah
different
implementation
detail,
but
solving
the
same
problem,
so
I
guess:
there's
plenty
of
slides,
I
encourage
everybody
to
read,
but
without
the
author
himself
being
there,
because
it
doesn't
make
much
sense.
So
I
guess
the
same
applies
here.
D
C
Brian
Weiss
not
a
as
well
in
a
form
more
as
I'd
like,
but
I
would
just
say,
listening
to
the
presentation,
I
like
the
idea
of
it
Nick's
licit
way
of
describing
the
behavior
other
than
and
implicitly,
which
is
what
I
think
Karen's
draft
just
doing.
There's
a
beauty
to
not
have
me
to
make
a
protocol
change,
but
I
think
it's
it's
nice
to
have
it
explicit,
what's
happening.
B
Yeah
well,
my
personal
opinion
is:
it
is
semantically
changing
something
without
touching
the
syntax
and
that
usually
captures
people
by
surprise
and
that's
not
not
usually
so
nice
but
yeah
I,
guess
I
read
both
drafts
myself
again
and
think
a
bit
more
about
it.
Chairs.
E
D
B
Okay,
that
actually
bring
us
to
the
last
item
arm
which
and
the
would
present-
if
you
were
here
so
since
it's
just
suppose
do
you
reminder.
I
can
also
remind
you.
The
radius
protocol
leaves
many
things
undefined
in
terms
of
proxying.
How
to
behavior
very
down
stream
survey
is
not
responding.
Does
that
mean
that
next
server
is
not
there
should
I
actually
market
that
and
jump
over
or
maybe
is
it
just
a
few
ups
down
one
server
which
doesn't
relay
the
messages
and
actually
the
next
pop
is
perfectly
fine.
B
A
B
B
Usually,
if
somebody
has
something
to
push
this
three
or
four
people
sitting
there
coming
one
time
doing
a
presentation
and
going
away
again,
we
don't
have
much
new
work
to
work
on
the
proposals
we
get
every
now
and
then
suggest
drastic
changes,
which
would
basically
turn
radius
into
something
else
in
a
lot
radius,
anymore
and
I.
Don't
think
that's
the
kind
of
what
we're
going
to
do
here,
so
the
amount
of
in
scope
work.
This
is
just
not
very
much
as
I've
said
earlier
in
terms
of
reading
drafts
for
working
with
last
fall.
B
C
Prime
wise
again
so
the
one
I
think
there
is
it's
valuable
to
have
a
place
in
the
ITF
where,
when
somebody
says
we
have
this
new
piece
of
functionality,
we
need
the
radius
extension
or
attribute
extension
word.
You
know
we
need
to
get
the
document
out.
If,
if
this
start,
this
group
doesn't
exist,
where
could
we
do
that.
B
Yeah
sounds
like
a
good
option:
I
mean
people
have
a
go-to
place
to
start
the
discussions
and
then,
if
it's
important
enough,
somebody
will
pick
it
up.
So
as
an
extra
data
points
just
a
few
hours
ago,
I
think
was
yesterday
all
day
before
I
met
also
had
this
meeting
I
decided
to
do
basically
the
same
thing.
B
B
Okay,
yeah
what
I
really
find
a
bit
sad
about
this,
like
I,
said
earlier
today:
reduced
Els
is
still
experimental
people
out
there
use
it.
It
is
useful,
but
I
get
these
questions
about
sugar
really
jump
on
something
that
is
experimental,
so
just
getting
that
one
on
standards
faculty
very
good
to
have
Brian.
C
B
We
had
a
discussion
with
Jim
shot
actually
had
slides
in
like
IDF
95
or
so
it's
not
a
lot
it's.
The
thing
is
that
TLS
1.3
is
coming
up
with
these
crazy
one
round
trip
time
or
maybe
it's
the
hundredth
time
session
starts,
and
that
might
need
a
bit
of
a
closer
look,
because
if
you
use
happen,
if
you
have
an
accounting
package
that
I
send
in
one
shot,
does
that
really
work
as
it
used
to
no?
D
It's
an
option:
how
much
would
waiting
40?
Let's
one
point,
free
hold,
you
up,
I
know
it's
just
about
done,
but
it
was
just
about
done
yeah.
It
is
just
about
done
who's.
Your
a
round
trip
discussion
extended
it
from
the
last
meeting
into
this
meeting.
I
expected
to
get
published
before
the
next
meeting
or
be
in
the
RFC
I
mean
with
you.
B
D
E
D
B
B
So
yeah,
actually
that
was
my
last
slide,
so
I
got
back
to
all
of
this
on
the
my
name
is
using,
including
closing
down
the
many
closing
down
the
working
with.
B
E
Just
just
for
clarification,
so
we
we
are
saying
that
we
are
closing
the
working
room
so
that
we
are
closing
the
working
group
after
the
last
remaining
document.
For
instance,
we
we
are
about
to
decide
the
world
adoption
of
one
of
the
extended
is
solution,
so
the
guess
would
be
that
we
work
for
that
ish.
E
D
Prefer
to
finish
it
up
in
radix,
okay,
we
might
be
starting
in
this
well,
so
it's
a
little
money
to
because
I'm,
a
security
area
director
and
an
ops
working
group.
Where
do
you
go
to
get
your
stuff
done?
Who's
the
ad
sponsor
for
you
I
think
it's
easier
to
just
finish
this
up
in
the
working
group,
but
let's
get
it
done
quick,
so
the
faster
they
can
turn
that
around
I
mean
if
we
can
get
this
through
working,
a
blast
call
and
through
80
review
and
to
I
use
to
you
before.