►
From YouTube: IETF108 HACKATHON 20200724 1400
Description
HACKATHON closing session at IETF108
2020/07/24 1400
A
A
Yep
great
okay,
let's
see
well
we're
gonna,
give
just
a
few
minutes
for
people
to
join
and
we
should
have
plenty
of
time.
I
don't
see
us
needing
the
full
two
hours.
A
My
guess
would
be
we'll
take
closer
to
to
one
hour,
maybe
a
little
bit
more
with
the
switching
from
one
presentation
to
the
other
but
anyways.
Let's
let
those
gather
for
a
little
bit
longer
for
those
of
you
who
are
presenting
in
order
for
you
to
present
you'll
need
to
be
able
to
you'll
need
to
be
able
to
get
into
the
queue
to
share
your
your
audio,
your
video.
A
If
you
like,
and
to
share
your
screen
and
then
you
should
be
able
to
drive
the
presentation
from
from
your
end,
if
you
prefer-
and
you
uploaded
your
presentation
to
github,
I
could
probably
download
it
and
drive
it
as
well,
but
if
you're
able
to
share
your
screen,
I
think
that's
probably
the
easiest
way
to
do
it.
A
What's
kind
of
seems
to
be
working
for
me?
Is
I,
rather
than
enter
into
presentation
mode
for
for
a
presentation,
though
I
kind
of
keep
it
in
not
in
presentation
mode.
That
way
it
doesn't
take
up
the
whole
screen,
and
while
the
viewers
can't
see
it
quite
as
big
at
least
I
can
still
see
my
muteco
screen
in
a
bit
of
what's
going
on
there,
so
you
can
see
what
works
best
for
you,
I'm
going
to
try
it
this
way
this
time
and
see
how
that
goes.
A
Great,
I
see
we
still
have
more
people
joining,
so
I'm
going
to
give
it.
I
think,
probably
a
7.05
we'll
get
started.
A
Just
to
give
you
an
idea
of
here's,
the
current
list
of
presentations
that
I
I
do
have,
I
believe
there
are
now
10
presentations
we're
going
to
stick
to
the
five
minutes
but,
like
I
said,
we
should
have
some
time
at
the
end,
both
for
general
discussion
and
if
we
need
more
time,
someone
wants
to
talk
more
q,
a
about
a
project
when
you're
presenting.
If
you
want
to
take
some
questions
on
your
project,
that's
fine.
Just
you
know
plan
on
that
being
part
of
your
your
presentation
time.
A
So
you
know
finish
your
presentation
a
bit
early
so
that
you
can
entertain
questions
near
the
for
the
last
portion
of
your
five
minutes,
because
we
are
going
to
try
to
keep
each
presentation
two
to
five
minutes,
just
to
keep
things
flowing
and
to
be
fair
to
everyone.
A
We
are
recording
this
session,
which
is
great
it'll,
be
available
afterwards,
because
there'll
be
a
lot
going
on
and
I'm
sure
I'm
going
to
miss
some
things.
I
want
to
be
able
to
review
it,
so
it'll
be
good
to
have
that.
A
A
Okay,
I
think
I'm
going
to
go
ahead
and
get
started
then
so
welcome
to
the
closing
session
of
the
the
hackathon
and
I
hope,
you've
had
a
good
week.
Certainly
things
have
been
different
with
you
know
doing
this
online,
and
hopefully
you
know,
while
some
things
are
not
as
good
as
in
person.
Hopefully
other
things
actually
worked
out
better,
especially
for
remote
participants.
I
imagine
when
everyone's
remote
it
might
be
a
little
bit
easier
to
work
with
each
other.
A
So
just
a
reminder-
hopefully
you
all
caught
the
the
opening,
but
for
for
those
of
you
who
didn't
a
couple
quick
reminders,
one
just
the
reason
we're
here
right
and
really
the
goal
here
is
to
move
all
the
valuable
work
that
we're
doing
in
the
ietf
forward,
at
a
faster
rate
by
bringing
people
together
working
on
not
just
the
drafts
that
we're
writing
but
the
to
try
some
code
that
verifies
what
these
things
are.
A
We're
standardizing
these
new
protocols,
new
fields,
new
things
for
for
the
internet
that
are
going
to
make
it
faster,
more
secure,
better
privacy
considerations.
You
know
everything
about
the
internet
really
and
be
able
to.
A
You
know
flush
things
out
early
and
not
only
that,
but
also
produce
some,
hopefully
some
handy
tools
and
utilities,
things
that
will
make
it
easier
for
other
people
when
they
go
and
try
to
add
support
for
these
new
standards
protocols
into
their
their
products
into
their
solutions,
so
that
we
really
get
not
only
just
a
larger
set
of
written
standards.
But
we
get
more
solid
deployments
out
in
the
market
quicker
of
everything
that.
B
A
Standardizing
in
the
idf
we're
operating
under
the
note
well,
as
all
ietf
meetings
do
so
the
presentation,
certainly
that
you're
doing
now
will
fall
under
the
rules
of
the
note.
Well,
hopefully,
you're
all
familiar
with
it.
If
not,
you
can
take
some
time
to
check
that
out,
as
as
we
as
we
get
started,.
A
Here
so
the
presentations,
as
I
mentioned,
we're
going
to
keep
them
to
five
minutes,
really
focusing
on
lessons
learned
the
feedback
that
hopefully
you'll
be
able
to
bring
into
your
working
group
assuming
they're
meeting
this
coming
week
and
or,
if
not
via
their
the
mailer
or
when
you
have
your
next
interim
meeting,
so
that
you
can
help
move
the
work
of
the
entire
working
group
forward
and
it's
nice
to
know
about
open
source
projects
that
perhaps
you
contributed
to
or
used
or
that
you
you
worked
with
if
you
crossed
sdos,
obviously
there's
things
important
etf
that
are
going
on
elsewhere,
being
open
source
communities
or
in
other
sdos.
A
A
This
is
the
the
list
of
projects.
The
plan
is
to
go
through
them
in
order,
and
hopefully
the
person
I
have
is
the
the
right
person
who
will
be
presenting.
If
not,
you
can,
let
us
know,
via
via
the
chat
or
just
by
entering
the
queue
and
lorenzo
and
the
the
folks
from
etecco
are
going
to
help
me
a
bit
with
the
the
queue
management
so
that
we
can
make
sure
that
you
you're
able
to
speak
and
share
your
screen.
A
So
so
yeah
when
it's
time
for
you
to
present,
make
sure
you
get
into
the
queue
and
when
we
see
you
enter
the
queue
we'll
be
able
to
unmute
you
or
if
you
have
some
questions
at
near
the
end
of
our
presentation.
If
there's
time
for
q
a
make
sure
you
get
into
the
queue
and
we'll
be
able
to
acknowledge
you
and
and
then
give
you
the
floor.
So
you
can.
You
can
share
your
question.
A
Okay,
so
with
that,
I
think
we'll
start
with
walter
and
if
you
want
to
share
your
presentation.
B
B
This
actually
was
the
first
time
the
network
management
research
group
of
the
irtf
has
been
presented,
something
at
the
at
the
acaton
and,
being
this
based
on
activity
from
an
irtf
group
is,
let's
say,
less
focus
on
standard,
specific
implementation
of
any
itf
solution
and
it's
more
related
to
research
or
or
main
idea
and
architectures,
and
actually
this
is
what
we've
been
discussing.
Particularly
the
problem
we
were
working
on
is
about
intern-based
networking
for
those
of
you
who
are
not
familiar
with
this
approach.
B
Basically,
the
idea
is
to
define
this
intend
as
a
set
of
operational
goals,
that
our
network
should
meet
and
outcomes
that
a
network
is
supposed
to
deliver,
define
in
a
declarative
manner,
without
specifying
how
to
achieve
or
implement
it.
So,
basically,
the
idea
is
to
move
from
a
more
prescriptive
approach,
which
is
the
traditional
way
where
you
configure
and
manage
your
network
to
something
different,
something
that
is
more
declarative,
and
this
is
how
all
this
ideal
intent
based
networking
comes
into
play.
B
There
has
been
some
activity.
There
have
been
some
activities
within
the
nmrg
research
group
about
this,
and
this
is
brought
to
a
couple
of
drafts
that
are
currently
being
being
developed.
One
is
about
trying
to
define
what
is
actually
an
intent,
so
having
concept
and
definitions.
B
As
well
as
some
discussion
on
how
an
intent
is
different
from
policy-based
network
management
and
service
management
and
service
models,
and
then
also
there
is
another
draft
about
intent
classification.
So
those
are
the
references
for
those
of
you
who
want
to
get
more
about
about
this.
So
what
what?
In
specific
we
wanted
to
solve
in
this
academ,
we
wanted
to
understand
if
we
can
apply
the
intent
concept,
different
levels
that
basically
correspond
to
what
you
can
have
as
different
stakeholders
consider
the
case
where
you
have
infrastructure
providers.
B
You
have
service
providers
that
providers
that
wants
to
to
use
some
reserve
as
slice
on
top
of
that
infrastructure
and
then
customer
or
verticals
that
are
actually
using
that
that
slices.
So,
basically,
we
have
infrastructure,
provided
that
provides
some
sort
of
slice
intent
interface.
B
The
service
provider
is
supposed
to
consume
this
slice,
intent
interface
and
provide
on
on
eastern
service
intent,
interface
and
then
customers
of
vertical
who
are
actually
interested
in
running
the
services
to
consume
that
service
in
there.
So
what
we
actually
wanted
to
do
is
to
understand
if
this
is
something
that
is
viable,
that
is
possible
to
do
with
a
proof
of
concept
built
on
a
network
function.
Virtualization
framework
we've
been
using
basically
openstack
as
a
virtual
micro
infrastructure
platform
and
open
source
manual
as
a
nfv
management
and
orchestration
tool.
B
So
the
idea
was
to
actually
demonstrate
the
how
we
could
actually
set
up
defining
specific
intent
at
the
two
levels
that
that
I
was
mentioning
slice
intent
and
service
intent
and,
of
course
we
did
this
in
a
very
simple
scenario,
with
a
single
point
of
presence
for
the
for
the
virtual
infrastructure.
B
So
what
we
did
what
we
were
able
to
do?
Actually
we
successfully
proved
the
concept
we
successfully
completed
the
demonstration.
Actually,
the
work
on
this
demonstration
was
a
joint
work
developed
by
myself
and
my
and
one
of
my
phd
students
at
the
university
of
bologna
in
italy,
as
well
as
other
colleagues
at
cnit
in
italy.
Those
activities
were
actually
part
of
the.
B
I
would
say
the
work
that
we
made
in
the
framework
of
the
network
management
research
group
in
the
last
year,
more
or
less
after
they
did
the
this
kind
of
ends.
On
demonstration
of
the
of
the
concept,
we
were
able
to
have
a
truthful
discussion
on
the
next
steps
to
be
to
be
done
within
the
nmrg
group,
especially
integration
with
high-level
intent,
expression,
translation
techniques
and
also
to
apply
the
methodology
for
the
intent
taxonomic
classification,
which
is
the.
B
B
What
we
learned
is
basically
ibm
is
still
a
green
field
in
terms
of
research,
so
there
is
still
a
lot
of
things
to
do,
and
this
is
basically
the
the
message.
The
main
message
that
we
would
like
to
share
with
our
research
group
that
we
really
need
to
bring
together
our
diverse
programmer
expertise,
because
we
need
to
integrate
people
their
works
on
language
on
for,
for
intent,
translation,
refinement,
people
that
works
on
validation
or
on
machine
learning
in
order
to
learn
and
understand
if
things
are
going.
B
So
all
the
the
points
that
you
see
here,
those
are
really
needed
in
order
to
build
a
real
ibm
system.
So
basically
that
that's
all
the
I
would
like
just
to
thank
the
the
guys
that
helped
me
in
running
the
this.
This
presentation
in
this
demonstration.
Again,
you
have
the
link
here
for
the
video
recording,
if
you,
if
you
are
interested,
if
you
want
to
know
more,
there
will
be
the
nmrg
meeting
next
week
on
wednesday,
and
this
is
the
link
to
the
agenda
and
I
guess
I'm
done.
B
Oh
yes,
the
the
group
chair
asked
me
to
to
report
on
the
on
the
result
of
the
of
the
akatos
session
that
we
had
on
wednesday.
So
we
will
be
discussing
and
I
will
bring
that
message
to
the
group.
A
C
Don't
hear
you
yet:
oh
okay,
okay,
okay,
there
we
go.
Okay,
let's
start
hello,
everyone.
My
name
is
a
phd
student
in
beijing,
university
of
post
and
telecommunications.
My
major
is
about
srv6
and
operators.
One
was
looking
for
some
solutions
about
the
the
loop
traffic
control.
I
noticed
f8
and
apn
6
and
had
some
interest
in
them.
So
we
read
some
documents
and
developed
both
a
demo
for
them.
C
C
C
Well,
some
days
ago,
somebody
asked
the
meaning
of
it
and
apn6
in
the
middle
east,
so
here
we
think
we
should
introduce
them
briefly,
as
just
mentioned,
I
fade
and
apn6
can
form
a
closed
loop
control
system.
I
fade
takes
charge
of
the
instead
of
telemetry.
It
supports
diverse
types
of
data
and
the
dynamic
configuration
for
indicating
them.
C
There
are
two
essential
functions
of
the
f8
demo.
The
one
is
the
encapsulation
of
ios
options
to
specify
desired
type
of
data,
and
the
other
is
the
obtaining
and
uploading
of
specified
data.
One
document
of
f8
mentions
that
there
are
two
modes
of
iom
telemetry,
namely
postcard
and
passport.
In
postcard,
the
iron
data
is
obtained
and
uploaded
by
20
nodes,
while
in
passport
uploading
is
responsible
by
endnote
and
the
item
data
is
transmitted
along
with
user's
payload.
C
The
next
is
apn6.
It
presents
some
hop
and
hop
options
to
indicate
the
requirements
and
can
determine
the
control.
According
to
these
options,
we
implemented
two
types
of
options:
to
identify
the
application,
information
and
four
types
of
sub
tov
to
indicate
the
performance
requirements,
namely
one
wireless
delay,
delivery
and
the
peculiar
ratio.
C
We
implemented
the
demo
based
on
p4
16
and
conducted
some
simulations
based
on
bmw.
2.
here
shows
some
realized
functions
in
effect
we
implement
both
postcard
and
the
passport
and
both
include
the
encapsulating
of
iom
options
and
the
obtaining
and
uploading
of
iom
data
in
api
6.
The
core
functions
are
the
encapsulating
of
the
mentioned
options
and
subtle
and
the
differentiated
srv6
seed
list
determination
in
api
6.
We
also
implemented
some
related
essential
services
processing.
C
Well,
the
next
period
we
plan
two
main
works.
One
is
transplanting
these
demos
to
the
barefoot,
tofino
switch
and
the
other
is
deploying
the
transplanted
demos
on
cni
sin
is
a
large
scale.
Clustering
experiment,
infrastructure
for
some
novel
network
schemes
is
still
on
construction,
but
at
present
is
some
notes
I
have
been
already
used
and
finally,
if
you
have
some
interests
or
puzzles
about
the
demo
or
any
others,
just
contact
us,
so
that's
all
and
thanks
for
listening,
hope
to
see
you
again.
A
Great
well,
thanks
for
for
sharing
that,
I
believe
I
saw
your
presentation
already
uploaded
to
to
github
as
well.
If
others
wanted
to
take
a
look
at
that,
so
yeah.
A
Let's
see
so
next
up
sylvester
on
quick,
quick.
A
D
D
So
I'm
presenting
for
the
quick
measurements
group.
We
are
working
with
a
tool
called
spin
dump
and
the
goal
is
to
provide
tools
and
ideas
for
traffic
measurement
in
the
area
of
trans
encrypted
transport
is
for
debugging
research,
network
house,
monitoring,
optimization
and
network
management
are
just
adjustment,
and
what
we
did
is
that
we
we
did
updates
to
the
spindam
tool
and
we
also
discussed
how
to
evolve
the
spindle
and
how
to
do
updates
to
it
in
the
future.
D
So
what
got
done
there?
There
were
many
many
small
updates.
Basically,
so
we
worked
on
traffic
classification
to
to
find
out
what
percentage
of
my
traffic
goes
to
the
to
to
facebook,
for
example.
So
it's
thin
dump
is
from
from
being
a
spin
bit
measurements
to.
We
are
evolving
it
to
be
a
more
general
measurement
too,
but
still
the
main
focus
is
quick,
but
there
are
many
other
things
involved.
D
D
D
We
updated
the
explicit
packet
less
measurement
part.
There
will
be
a
slide
on
that
also,
and
there
is
a
maintenance
for
new
versions
for
all
the
all
the
new
os's,
which
was
involved
from
other
people.
We
had
to
do
some
updates
for
the
tool
to
be
working
and
also
working
on
maintaining
the
tool
to
to
keep
up
with
the
quick
improvements.
D
So
for
model
generation
we
we
added
the
possibility
of
connection
statistics
for
each
closed
connection
and
and
what
we
actually
implemented
for
that
there
is
a
part
of
the
connection
statistics
here.
You
can
see
a
logarithmic
histogram
here,
which
can
be
also
exported
to
json
and
plotted
in
outside
tools.
But
you
can
you
can
see
the
heat
map
of
the
left
audit.
You
know
right
rtt,
here
about
srt
srt
is
a
secure,
reliable
transport
protocol
there
are
inform.
D
You
can
see
more
information
about
the
in
the
method
itself
and
the
implementation
of
the
quickgo
is
also
available.
So
in
this,
the
client
generates
a
qubit
signal
and
reflects
the
received
qubit
signal
using
the
arbit
signal,
and
the
server
does
does
similar
things
in
the
opposite
direction
and
using
these
signals
in
the
midpoint
it's
possible
to
drive
like
it
was
happening
in
the
network.
A
Okay,
fantastic
well,
thanks
for
for
sharing
it's
good
to
see,
continued
progress
on
spin
dump
and
the
the
measurements
around
quick,
so
yeah
congrats,
and
thanks
for
sharing.
Thank
you.
A
A
A
Yep,
your
audio
is
a
little
faint,
but
but
I
can
hear
you,
okay,.
E
E
This
hackathon
was
in
a
way
rather
uneventful.
E
Nothing
much
has
changed,
but
the
draft
has
been
accepted.
It's
waiting
for
the
rfc
editors
to
become
a
proper
rfc.
There
have
been
some
minor
changes
to
draft
just
before
the
draft
was
accepted
and
basically
verified
interoperability
after
the
changes
and
also
the
hackathon
will
sort
of
see.
If
we
can
do
something
useful
on
some
other
front.
E
E
There
are
the
the
big
star
of
the
this
hackathon
was
miroslav
who's,
the
developer
of
crony.
He
built
a
tool
to
be
able
to
do
sort
of
compliance
testing
with
a
draft
of
the
mts
key
exchange
part.
So
he
identified
some
issues
and
some
implementations
also
even
managed
to
fix
dishes.
E
During
this
hackathon
and
the
results
we
have
sort
of
a
google
sheet
with
lots
of
a
big
compatibility,
matrix
just
results
from
miroslav's
tests,
and
you
can
see
the
link
down
there,
but
the
the
google
sheet
is
a
bit
too
big
to
show
here
on
the
slides.
So
just
trust
me
when
I
say
that
most
of
them
do
work
and,
as
I
said,
miroslav
wrote
a
mtsk
testing
tool.
It
can
be
found
on
github.
E
Most
of
the
fails
that
we
see
in
or
most
of
the
fails
that
his
two
reports
are
due
to
implementations
being
too
lenient.
They
accept
things
that
aren't
really
don't
have
to
be
accepted
by
the
specifications
so
by
the
draft.
So
we
ought
to
tighten
them
down,
but
it's
it
doesn't
sort
of
break
anything.
It's
just
that.
For
example,
my
implementation
still
supports
tls
version
1.2
because
we
have
older
clients
using
tls
version
1.2
and
the
draft
was
changed
to
say:
okay
tls
version
1.3
only
there's
al.
There
were
also
a
few
bugs.
E
E
And
what
we
learned
well
automatic
testing
tools
are
really
useful,
being
able
to
run
them
over
and
over
again
and
see.
Okay
had
things
improved,
have
they
called
him?
Worse,
that's
really
good
and
well,
as
I
said,
interoperability
is
still
good.
Everybody
is
strict
in
what
they
send
may
be
too
bit
too
liberal
in
what
they
accept,
and
we
did
still
have
problems
with
fire
with
firewalls.
It
seems
that
for
a
couple
years
ago
there
were
some
big
denial
service
attacks
using
mtp
as
an
amplifier
to
increase
the
traffic
towards
a.
E
E
A
A
A
A
F
F
First
of
all,
ns3
is
a
discrete
event
network
simulator
and
that's
the
project
that
I'm
mainly
associated
with
and
discrete
event
simulation
really
is
well
suited
to
evaluating
congestion
control,
architectures
such
as
l4s,
and
it
has
some
complementary
benefits
to
physical
test
beds,
including
much
easier
accessibility
to
just
run
a
model
of
a
network
on
a
single
machine
and
the
ability
to
introduce
various
wireless
models
which
are
going
to
be
relevance
to
the
performance
of
something
like
l4s
and,
in
addition
being
able
to
find
and
cable
together.
F
F
F
So
that's
the
bigger
picture.
The
detailed
things
we
were
working
on
were
that
we
have
a
number
of
summer
students
on
the
ns3
project
that
are
working
on
various
aspects
of
the
l4s
architecture,
so
we
wanted
to
bring
some
of
that
work
together
and
push
it
forward
as
well
as
this
testbed
configuration
effort-
and
I
mentioned
a
couple
of
the
drafts
here
that
are
in
the
transport
area.
F
If
you
want
to
learn
more
about
this,
this
has
been
the
subject
of
previous
ietf
hackathons,
but
with
different
team
members
who
were
more
involved
in
the
linux
kernel
code.
At
that
time,
the
specific
problems
that
we
were
working
on
a
lot
of
integration
of
these
different
pieces.
We
have
a
model
of
tcp
prague.
The
student
there
was
working
on
the
rtt
independence
feature
that
was
recently
discussed
at
the
tsvwg
interim
meeting.
F
We
had
a
student
that
has
introduced
l4s
features
into
f
the
fq
caudal
aqm,
so
we
put
that
into
our
integrated
codebase,
and
we
wanted
to
create
an
experiment
with
the
integration
of
all
of
these
different
ns3
models
that
have
been
developed
on
different
forks
of
the
repository
and
then
try
to
start
validating
this
simulation
against
the
linux
testbed
results.
F
So
the
main
thing
that
we
were
able
to
do
this
week
is
to
produce
integrated
git
branches,
with
these
different
components,
putting
them
together
for
the
first
time.
In
some
cases,
our
ultimate
goal
is
to
get
these
into
our
main
line.
Next
release
is
scheduled
for
the
september
time
frame,
and
so
I've
posted
a
couple
of
branches
here
that
are
where
all
of
the
code
from
this
week's
efforts
reside.
F
We
also
have
published
some
scripts
and
documentation
for
creating
this
l4s
testbed
on
on
cloud
lab,
and
the
link
here
in
this
from
this
presentation
will
take
you
there.
I
also
have
a
summary
wiki
page
that
provides
more
instruction
and
details
of
how
you
can
basically
run
all
of
these
types
of
simulations
at
home.
F
F
F
A
Great,
I
see
there
was
one
question
from
sylvester:
what
bottleneck
speeds?
Can
you
measure
in
ns3
and
or
cloud
lab?
Is
there
any
limitations.
F
So
the
bottleneck
speeds
that
we've
been
evaluating
so
far
have
been
similar
to
the
speeds
that
were
used
in
the
the
l4s
previous
lfrs
beds,
so
they're
on
the
order
of
50
to
100
megabits
per
second.
We
don't
really
have
a
much
of
a
limitation
in
simulation
of
scaling.
That
up.
I
personally
can't
answer
the
question
in
terms
of
how
far
you
could
push
the
cloud
lab
infrastructure,
so
I
I'd
have
to
defer
answering
that.
G
G
Okay,
can
you
hear
me
there?
We
go
okay,
perfect
thanks
a
lot
I'm
presenting
for
the
bgp
monitoring
protocol
group.
G
We've
been
working
together
since
itf
106,
improving
the
bnp
implementation
and
the
lab
setup
where
we
are
testing
this
time
we
were
focusing
on
the
correlation
between
bgp,
logrip
and
ipfix,
and
also
in
the
verification
and
interoperability
of
some
tlv
extensions
to
pmp,
focusing
on
on
path
marking
and
the
rare
policy
attribute
tracing
message
type.
Besides
that
also,
we
want
to
monitor
and
understand
the
impact
of
bmp
in
terms
of
cpr
memory,
consumption
on
the
route,
processors
and
also
in
case
in
case
of
pgp
route
propagation
delay.
G
G
This
was
the
lab
network
we
were
using.
This
is
an
mplsvpn
network
which
is
more
or
less
the
most
challenging
in
case
of
getting
the
the
data
correlation
right
between
ipfix
and
bmp
from
a
lab
perspective.
This
time
new
was
the
traffic
generation
and
also
that
we
had
with
young
push
insight
on
the
cpu
memory
on
consumption.
On
the
routers,
we
understood
that
we
need
to
further
automatize
the
the
test
setup.
G
On
the
pma60d
side,
we
achieved
the
bgp
low
clip
to
ib
fix
correlation
with
prefixes,
with
bgp
route
distinct
issues.
G
We
are
on
the
way
of
decoding
all
the
tlvs
on
the
route
policy
attribute
tracing
and
we
identified,
let's
say
one
optimization
possibility
on
the
path
marking
tf
when
multiple
paths
are
present,
that
we
could
index
them
and
basically
gain
the
possibility
that
we
don't
have
to
mark
all
the
all
the
parts
within
the
bpu.
When
we
don't
know
all
the
the
paths.
G
We
did
perform
stress
tests
and
saw
some
increase
on
memory
in
cpu,
but
we
recognize
that
the
way
how
we
measured
the
cpu
increase
seems
to
be
not
realistic,
and
we
need
to
redo
that
for
the
for
the
next
hackathon,
the
wireshark
desect,
which
we
did
for
roughly
analyzing
basic.
Basically,
the
new
tlvs
and
message
types
worked
very
well.
G
G
G
We
had
levio
from
the
eth
university
master
student
on
board,
who
was
working
on
the
visualization
of
the
bmp
data.
G
He
was
using
d3
on
the
on
the
on
the
front
inside
and
plywood
on
the
backhand
side.
This
is
the
whole
end
to
end
chain.
He
was
using
so
pmact
on
the
data
collection,
kafka
for
the
message
broker.
G
What
we
learned
is
we
had
quite
some
few
first
time
on
board.
They
were
quite
refreshing.
They
brought
in
new
ideas.
G
This
was
the
crew
working
together.
Thanks
for
everybody,
it
was
really
a
great
time,
a
great
opportunity.
Any
question.
A
I
think
you
already
answered
one
of
the
questions
I
had
you
mentioned.
You
were
using
yang
push
for
some
of
the
measurements
and
that
you
had
some
problems
with
the
measurements,
but
that
was
more
in
your
your
implementation
or
how
you
were
looking
at
the
information
you
were
gathering.
It
wasn't
like
an
issue
with
yang
push
or
lack
of
support,
or
anything
like
that.
For
example,.
G
Exactly
it's
basically
regarding
how
the
the
cpu
is
being
used
and
there,
of
course,
there
are,
let's
say,
different
measurement
points
which
are
important
and
from
the
hackathon
we
learned.
Basically,
we
need
to
have
a
deeper
look
at
it.
A
So
next
up
we
should
have
logan
and
for
dns
sec
or
perhaps
newton
would
be.
A
Jake
shawn
is
hoping
to
present.
I
guess
we
have
an
issue
with
the
laptop
if
we
could
go
forward
and
come
back
to
you
later.
If
it's
something
we'll
give
you
a
minute
to
see,
if
you
can
get
it
straightened
out,
but
if
not,
we
can
we
can
pass
by
you
and
circle
back
at
the
end.
You
can,
let
me
know
in
the
chat
what
you
prefer
looks
like
things
are
getting.
A
If
you
see
that
that
should
enter
you
into
the
audio
queue
and
then
we
can
enable
your
audio,
can
you
hear
me
now
now
we
can
great.
A
A
A
Did
you
try
to
share?
You
may
need
to
allow
it
on
your
computer.
I
A
I
Okay
right
good
morning,
everyone
so
for
this
itf
hackathon,
our
main
targets
are
dns
sec,
tls
and
http
3..
I
am
nitin
member
of
service
dome
the
demu
currently
working
at
west
highland
support
services
and
deal
a
lot
in
the
market
data
industry
for
more
than
20
years
now.
So,
as
a
member
of
substance
mu
in
usa
and
together
with
a
team
in
mauritius,
we
focused
a
lot
on
the
rfc
8624
tls
1.0,
tls,
1.1,
deprecation
and
quick
http
implementation.
I
So
for
dns
musculo
we
have
been
duplicating
dnssec,
algorithm,
dot,
pr
sent
already
to
duplicate,
tls,
1.0
and
1.1,
most
interestingly
zx.
We
are
enabling
http
free
now
the
http
3
support
and
http
scanner.
We
are
now
working
for
the
improvement
of
the
http
free
support
and
map
myself,
I'm
working
on
that
to
add
auditing
feature
for
tls,
1.0
and
1.1
scoring.
I
I
I
Personally,
I
believe
that
we
have
been
able
to
made
it
on
this
hackathon
itself
with
http
free
and
it's
a
very
new
things,
and
but
yet
we
have
been
able
to
send
some
po
already,
which
is
I
I
believe
it's
it's
a
it
has
been
accomplished
for
the
next
itf
hackathon.
I
We
are
going
to
focus
a
lot
on
http
free
adoption
and
we
invite
all
the
developers
to
come
and
join
us,
and,
and
we
can,
we
can
bring
more
a
robust
and
secure
internet.
So
congratulations.
The
new
members
alex
bcso,
who
made
it
into
this
hackathon.
I
I
So
I
would
like
to
thanks
charles
the
mediocre
team
everyone
during
this
pandemic,
and
but
yet
we
are
going
strong
for
for
this
hackathon
and
keep
safe
guys.
Thank
you.
A
All
right,
great
yeah,
thank
you
very
much
for
the
the
presentation
sharing
it
for
the
with
for
the
whole
group
there
it's
good
to
see
your
continued
efforts
are
on
these
these
protocols.
So
I'm
glad
the
online
version
of
the
hackathon
worked
out
for
it
welcome.
A
Let
me
stop
sharing
and
let
me
see
who's
up
next
then
so.
Next
up,
we
should
have
brendan
morin
for
suit.
A
A
A
A
A
So
if
the
screen
shares
not
coming
up,
that
could
be
it
and
we
still
have
not
seen
you
in
the
audio
queue.
So
probably
where
you
see
the
little
microphone
with
the
hand
on
it,
you'll
have
to
press
that
and
that
will
put
you
in
the
queue
and
then
we
can
accept
that
and
you'll
be
able
to
speak,
we'll,
be
able
to
hear
you
so
kind
of
top
left
strength
of
the
screen
right
below
where
it
says
your
participant
name.
H
H
Yeah,
apparently
it
didn't
have
permission
previously,
so
I,
in
order
to
give
it
permission,
I
have
to
restart
okay
there
you
go
I'll,
be
back.
A
Okay,
the
next
two
presentations
are
from
paul
one
for
I
p
wave
and
one
two
ns
sorry
i2
nsf.
A
A
Okay,
we
don't
hear
you
yet,
but
we
can
see
your
screen.
Okay
and
now
we
hear
you.
J
Okay,
good
so
hello,
everyone.
This
is
paul,
john
from
sanging
one
university,
so
I
will
report
our
ip
wave
basic
protocols
project.
J
Basically,
this
hackathon
we
want
did
to
implement
the
context
aware
navigator
protocol
called
cmp
for
ip
based
stabico
networks.
J
This
figure
shows
context
where
navigator
protocol
over
ip
a
wave.
You
can
see
some
vehicle
free
to
collide
with
some
obstacle
in
roadway
and
the
v1
detected
this
accident
and
then
disseminate
emergency
contacts
message
to
neighbors
to
avoid
these
kinds
of
accident,
so
they
change
their
maneuver
safely
to
avoid
the
consecutive
accident.
J
J
This
is
the
figure
network
structure
in
simulator,
so
you
can
see
multiple
rcu's
three
rsus
are
deployed
in
a
highway
and
we
have
mama
stands
for
mobile
anchor
for
mobility
management
and
the
beakers
can
communicate
with
b2b
using
ip
wave
ocb
mode
and
also
they
can
talk
to
rcu
for
multimanagement.
J
So
we
implemented
this
using
omnet
plus
plus
the
this
figure
shows
omnipress
plus
network
simulator.
So
we
have
a
two
rsus
and
also
we
have
multiple
curls
moving
through
a
lot
of
way
and
also
mobility
anchor.
J
So
what
we
got
done
so
we
implemented
our
contextual
navigator
protocol
over
ipl
web
protocol,
so
we
demonstrated
so
we
implemented
the
coordinate,
maneuver
controller
algorithm
to
avoid
the
collision.
So,
on
this
second
son,
we
implemented
the
same
people.
We
didn't
separate
the
bigger
mobility
information
such
as
ecmccm,
yet
so
we'll
separate
next
hackathon
project.
So
what
we
learned
during
this
hackathon
beaker
can
exchange
their
mobility
information
over
ip
wave
protocol.
J
J
So
my
tm
has
two
members,
my
phd
student,
the
pnma
and
the
event
and
chris,
and
also
we
had
a
participant
from
different
okay,
universal
universities,
okay
from
central
university
and
congo,
national
university
to
adopt-
and
also
we
had
alex
alex-
was
the
he
was
a
main
contributor
ip
wave
and
the
early
one
also
joined.
So
our
project
is
a
open
source
project
in
github
and
also
we
demonstrated
using
video
clips.
So
you
can
take
a
look.
Thank
you.
A
Great
thank
you
paul,
so
you're
welcome
to
just
proceed
on
to
the
second
question.
A
A
Excuse
me,
so
you
wanted
someone
else
to
present
the
i2
nsf.
First.
J
J
J
So
on
this
hackathon
the
plan
is
we
wanted
to
reflect
the
latest
revision
of
the
following
working
group
draft
based
on
openstack
based
the
i2nsf,
especially
capability
data
model
and
consumer
pacing,
nsf
phase
interface
and
authorization.
J
So
we
have
also
script
power
translator.
This
is
not
a
good
draft
yet,
but
this
is
very
important
for
the
i2
nsf,
so
this
figure
shows
itunes
a
framework
for
internet-based
networking
based
security
services.
So
basically
you
can
see.
Other
integrator
gives
us
some
high
level
policy
and
then
this
policy
going
to
the
cloud.
J
So
the
cloud
has
a
security
controller
has
translator
and
consumer
facing
interface
data,
young
model-based,
high-level
policy,
translated
into
low-level
policy
that
is
a
deliberate
over
nsf
facing
interface
and
that
each
policy
delivered
to
appropriate
nsf
natural
security
function,
firewall
ips
filter,
tdos
automaticator,
so
we
implemented
this
i2a
system
on
top
of
open-based
openstack-based
nfo
system,
so
we
published
I
therapy
communication
magazine
this
year,
so
we
can
refer
to
this
paper
for
further
discussion
and
explanation.
J
This
time
we
used
this
network
topology,
so
we
installed
the
itones
itunes
user
as
a
decibel.
The
system
and
the
security,
controller
and
dms
stands
for
develop
management
system,
providing
a
network
secure
function
as
a
vm
virtual
machine,
and
then
we
tested
the
time
based
firewall
and
url
filter.
J
J
What
we
got
done
so
we
restored
the
i2
and
slab
framework
on
top
of
openstack
again.
So
web-based
item
as
the
user,
and
also
conservation,
security,
control
and
development
system
dms,
and
also
we
use
the
script
policy
translator
for
ibm
that
took
internet-based
networking
and
also
we
reflected
the
latest
revision
of
two
young
data
model
consumer
facing
and
analyst
phase
interface
based
data
young
model.
J
J
We
want
to
reflect
latest
restoration
interface
and
also
we
will
implement
a
monitoring
young
data
model
so
that,
basically,
we
push
out
all
data
young
models
into
the
rfcc
this
year
and
also
we
want
to
enhance
skill
to
follow
translator
for
automatic
setup.
J
So
this
is
my
team
information,
so
I
have
three
the
two,
the
first,
the
patrick,
is
a
main
player
for
the
second
song.
He
is
my
phd
student
jae,
hong
and
yosok
they're,
also
my
college
student,
and
also
we
had
the
the
participant
from
social
university,
professor
young
kim
and
young
and
kang
national
university
gyeongsi
and
manson
kicha
kato.
So
this
also
offers
a
project.
A
A
And
I
was
just
I
was
curious
if
the
high
level
one
is
like,
is
that
the
intent
based
right.
B
J
J
They
don't
need
to
know
the
ip
address
of
employees
and
also
the
sns
website
that
kind
of
information
store
in
the
database,
the
security
and
controller.
J
So
basically,
our
translator
translate
dip
addresses
of
computer
and
devices
used
by
employees
and
also
we
translate
sns
a
website
into
specific
website
websites
such
as
google
youtube
facebook
instagram,
something
like
that.
So
those
kinds
of
map
performed
so
basically
the
first
automatic
mapping
case.
We
we
can
match
the
such
kind
employee
into
the
ip
addresses
of
employees
devices.
J
Okay,
second,
one
is
to
generate
low
level
device
level
security
policy
we
want.
We
need
to
know
the
the
grammar
structure
of
the
production
rule
to
generate
a
loadable
yang
tree
based
xml
file.
So
currently
we
manually
stored
this
production
rule
into
database,
but
next
step
we
want
to
automatically
store
into
database
for
this
kind
of
production,
rule
yeah
setup.
A
Okay.
Now
we'll
we'll
switch
back
to
brendan
again
to
get
your
presentation,
hopefully
this
time
with
screen
sharing
enable
okay.
Thank
you.
Yeah
charles
yeah,
thanks.
A
A
H
Okay,
good,
it's
working
this
time
all
right,
so
I
guess
I'll
start
by
saying
it's
a
little
bit
less
a
participation
that
I'd
hoped
for,
but
that
probably
has
something
to
do
with
us
having
a
different
hackathon
about
11
days
ago,
we
had
it
on
the
on
july.
The
13th
we've
already
done
the
some
some
wrap
up
for
that.
So
I
won't
go
through
it
here.
So
what
came
out
of
that?
H
One
was
that
it
was
clear
that
we
needed
to
explore
a
few
corner
cases
in
the
suit
manifest,
and
specifically
there
were
a
few
examples
that
really
do
need
to
be
added
to
the
to
the
draft.
So
the
idea
here
was
going
to
be
to
to
move
that
forward.
H
So
I
did
manage
to
add
dependency
support,
which
was
good
and
got
very
a
very
long
way
into
figuring
out
how
to
optimize
the
generation
of
manifests
and
discovered
a
possible
missing
bit
of
handling
multiple
components,
and
so
I
had
a
discussion
with
some
other
implementers
about
that
and
came
up
with
a
new
way
to
handle
it.
So
that's
all
good
too.
H
Yes,
that's!
I
must
have
jumped
ahead
of
myself
all
right,
so
we
have
a
new
implementation
that
we
are
are
proposing
to
take
to
the
suit
working
group
next
week,
so
that
should
move
things
along
and
there
were
a
few
more
people
who
were
interested
initially,
but
in
the
end
it
was
really
just
me
and
david
who,
who
kept
coming
and
david
was
actually
working
on
something
a
little
bit
different.
He
was
working
on
the
the
implementation
of
suit
manifests
for
mcu
boot.
H
A
Okay,
well
great
any
questions
for
for
brendan.
A
That
would
be
a
good
time.
Okay!
Well,
thank
you
for
the
presentation,
you're,
the
other
hackathon.
What
was
that
done
in
conjunction
with.
H
Oh
no,
it
was.
It
was
one
of
these
things
where
it
was
set
up
before
this
hackathon
was
announced,
and
and
so
we
figured
we'd,
go
ahead
with
it
and
hey.
If
we
get
two
out
of
the
deal
so
much
the
better,
but
as
it
happens,
it
didn't
work
that
way.
So
there
we
are
okay.
A
All
right:
well,
thanks
for
participating
in
for
sharing.
A
I'm
just
going
to
jump
back
to
my
presentation
here
so
so
we
went
through
the
projects
and
thanks
to
all
of
you
for
for
presenting,
I
really
appreciate
you
taking
the
time
not
only
to
participate
in
the
hackathon,
of
course,
but
then
to
to
share
your
results
with
everyone
and
all
of
the
results
presentations.
A
I
believe
all
of
them
are
already
here
last
I
checked
it
looked
like
all
of
them
were,
if
you
had
any
more
updates
of
that,
you're
welcome
to
continue
to
update
these
after
the
hackathon's
over
it's
just
a
great
to
have
that
resource
right.
So
when
people
look
at
the
watch
the
videos,
they
can
go
back
and
see
your
presentation
as
well
and
get
all
the
information
there.
A
A
A
A
A
So
if
you
go
to
the
wiki,
which
has
been
our
source
of
information
this
whole
week,
there
is
a
new
link
added
here,
a
hackathon
survey.
It's
a
it's
relatively
short.
I
think
it's
about
10
questions,
especially
since
this
was
our
first
time
doing.
An
online
version.
We'd
really
appreciate
it.
If
you
could
just
fill
this
out,
I
can
see
each
question
has,
like
you
know,
kind
of
a
multiple
choice,
thing
that
lets
us
know
how
well
or
not.
A
Well,
something
worked,
and
you
know
really
appreciate
any
comments
you
have
in
addition
to
that,
you
know
hopefully
we'll
be
able
to
hold
the
hackathon
in
person
again
in
the
future.
But
you
know
odds
are,
are
good.
That
won't
happen
right
away.
In
any
case,
we're
learning
a
lot
about
how
to
better.
You
know
new
tools.
A
We
can
use
things
we
can
do
to
improve
when
people
are
remote,
especially
like
the
integration
between
combination
of
remote
and
in-person
teams,
and
so
you
know
that's
going
to
be
valuable
things
that
we
find
that
work
well,
we
want
to
continue
to
use,
even
once.
We
do
hopefully
go
back
to
having
our
more
usual
format
and
the
of
the
in-person
hackathon.
A
So,
thanks
to
all
of
you
for
joining
me,
techo
folks
appreciate
all
your
help
with
managing
the
cues.
That
was
great.
Let
me
check
and
see
if
we
have
any
any
other
questions
nope.
I
think
we're
good
with
that
I'll,
we'll
we'll
call
this
quits,
give
you
all
back
the
rest
of
your
day
have
a
great
evening
rest
of
your
day,
wherever
you
are
and
yeah
thanks
for
participating
in
the
hackathon,
I'm
pretty
happy
with
with
how
things
went
overall.
A
Hopefully
it
was
a
good
and
useful
for
you
and
good
luck
with
the
rest
of
the
ietf
meeting
next
week.