►
From YouTube: CNCF Networking WG Meeting - 2019-01-15
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
A
So
what
I
thought
I
would
do
today,
I
didn't
know.
The
tenants
is
a
lot
loan.
I
thought
it
would
be,
but
I
would
at
least
talk
through
some
of
the
ideas
that
we
had
from
and
say
we.
The
last
time
we
had
like
a
real
meeting
was
probably
the
middle
of
dinner
September
right
before
a
lot
of
the
conferences
and
event
started
kind
of
taking
away
a
lot
more
time
from
people.
A
So
these
items
over
here
on
the
left
were
the
things
that
we
had
talked
about:
no
kind
of
network
extensions
for
no
policy
for
CNI,
maybe
looking
at
like
extensions
for
like
routes
within
cloud
native,
specific
capabilities
that
may
be
above
the
CNI
later
we
had
kind
of
talked
about.
You
know
yeah
the
interface
and
you
have
sort
of
things
above
that
interface
and
routes
seem
to
fit
that
category
and
other
things
might
fit
that
category
as
well.
A
We
talked
about
server
load,
balancing
ipv6,
DNS
service,
chaining,
slash
sharing,
you
know,
you
kind
of
you
discover
these
services.
How
do
you
sort
of
define
network
services
and
how
they
can
work
together,
especially
there's
like
a
dependency
need
there,
and
then
things
like
you
know:
service,
mesh
and
noak
observability
were
topics
that
we
we
touched
on
as
things
that
really
hadn't
been
defined
very
clearly
for
for
a
cloud
native
use
cases.
So
I
can't
thought
we'd
start
with
these
items,
and
you
know
we
can
add
other
items
to
this.
A
If
we
wanted
to
what
we
could
just
sort
of
start
with,
you
know
trying
to
you
know
at
the
next
meeting,
maybe
taking
a
couple
of
these
and
trying
to
define
what
what
we
want
to
do
with
then
within
our
environment.
You
know
within
this.
What
group
to
address
these
are
the
few
others.
You
know.
A
Yes,
when
the
server
list
works,
which
we
came
up
with
kind
of
a
landscape
for
service
and
I
didn't
know
if
within
the
network,
went
within
our
cloud
native
landscape.
Today
we
have
networking
as
a
category
but
I,
don't
know
where
that's
really
the
way
we
want
to
define
the
network
for
cloud
me
the
right
sort
of
a
bunch
of
network
vendors
right
now.
A
You
know
and
their
project
in
there,
so
this
is
like
a
question
I'm
not
saying
we
have
to
do
something
that
would
something
I
thought
about
was
sort
of
looking
at
network
from
different
areas,
kind
of
server
listed,
and
then
you
know
we
also
did
like
a
white
paper
in
service.
I
can
talk
about
what
that
means
in
terms
of
cloud
native
and
I
know
if
it
makes
sense
to
have
a
white
paper
or
a
set
of
white
papers
around
networking
and
cloud
native
on
what
that
means.
Yeah.
B
Nearly
all
of
these
to
help
brainstorm
a
bit
and
reinforce
the
network
landscape.
You
didn't
I,
just
popped,
a
link
over
and
zoom,
and
it's
it's
been,
maybe
18
months
or
so
about
a
year
and
a
half
ago,
or
more
that
the
CNCs
staff
had
asked
for
me
to
put
together
a
perspective
on
at
the
time.
It
was
a
container
networking
and
it
was
to
help
facilitate
my
panel
at
I
think
it
was
oh,
oh
and
I
have
a
panel
there
and
part.
B
I
actually
called
out
the
the
cloud
native
landscape
at
the
time
and
yeah
that
was
pre
serverless
lanced.
It
was
pre
sort
of
sub
landscape,
a
few
others,
and
to
make
it
simple
and
based
on
the
time
that
I
had
I,
tried
to
characterize
and
contrast
for
prominent
projects,
at
least
at
least
back
then,
and
it
was
yet
to
help
at
the
time
it
was
to
help
facilitate
a
discussion
and
a
panel
and
and
but
but
just
brainstorming
out
loud
about
the
possibility
of
a
you
know
like
a
networking
Wednesday
right
yeah.
B
Actually,
if
you
think
about
that
landscape,
actually,
if
the
landscape
was
defined,
it
might
actually
even
help
describe
some
of
the
other
cards
that
you
have
there
I
mean
because
you
can
imagine
in
a
networking
landscape.
If
we've
had
categories
here,
you
know
right
now.
It's
application
and
application
definition
and
development
orchestrating
their
management
on
this
landscape,
but
pregnant
working
when
it
might
be
security,
observability
management,
routing,
server,
load,
balancing
rabbit.
You
know
it
might
be
a
PID
twins.
It
might
be
because
network
as
a
term
as
a
space
covers
it
in
some
respects.
B
B
A
I
think
kind
of
helps
right
because
it
shows
it
kind
of
shows
what's
in
scope,
for
what
group
this
is
what's
covered
by
other
areas.
You
know
thinking
about
this
from
the
standpoint
of
you
know,
with
with
kubernetes
having
SIG's
now
and
I.
Have
a
network
saying
I
was
kind
of
thinking
of
you
know
what
what's
our
purpose
in
this
ecosystem
right?
Oh,
we
doing
the
same
thing
as
a
community
sake.
Should
we
joined
forces
with
them
and
the
more
I
thought
the
more
I
thought
about.
A
B
The
challenge
from
my
perspective,
in
responding
to
that
question,
was
that
it
was
undeniable
that
the
communities
networking
Singh
was
a
greater
center
of
gravity
than
than
this
working
group
was
areas
I.
Think
with
organization
like
this
I
think
we
can
change
some
of
that
yeah
to
your
point
like
yeah.
A
B
There
are
thing
there
are
concerns
in
areas
to
address
that
are
well
beyond
kubernetes
and
things
that
are,
you
know,
certainly
of
concern
to
to
end
users
and
to
vendors
between
high
hybrid
things,
work
with
communities
on
crema
cobranet
he's
running
as
a
service
than
you,
then
that
working
between
those
there's,
a
number
of
but
like
ingress
controllers,
is
a
good
example
like
he's
something
that
certainly
that
focuses
on
in
terms
of
their
networking
city,
but
there's
a
number
of
projects
within
there.
How
do
those
relate
to
each
other?
B
B
Get
into
politics
there's
nothing,
but
the
point
being
like
highlighting
here's.
Here's
here's
why
this
here's,
where
the
cloud
native
aspect
of
this
project
has
helped
it
see
much
more
adoption
in
these
environments
and
I
think
a
white
paper
that
discusses
some
of
that
is
of
interest.
I'd
recently
recently
put
out
a
report
through
O'reilly
on
surface
mesh
and
some
of
what
I'm
talking
about
was
spoken
to
in
there
I
think.
B
These
of
these
topics-
ability
do
you
know
sir
Courtney
acts
as
an
example
I'm,
assuming
that
that
I
haven't
been
involved
in
that
group
I'm
sure
they've
got
some
other
community
where
they
meet
within,
but
I
mean.
Maybe
that's
right
now
we're
just
ripping
here,
but
maybe
that's.
What
a
car
to
put
on
here
is.
Is
the
attention
me
like
you
said,
going
over,
you
can
any
city
or
submitting
some
things
to
that
or
actually
going
over
to
those
other
projects
like
Cortinas
or
envoy
or
easier,
or
you
know
or
take
it.
A
B
A
Think
so
too,
and
I
think
once
once
the
I
think
that
the
end
user
communities
get
a
little
bit
more.
It's
getting
a
regular
meeting
cane,
it's
not
gonna
present
to
them
in
an
upcoming
event.
You
know,
meaning
they
can
say.
Hey.
We
here's
some
of
the
work
groups
we
had
today:
we'd
love
participation
from
the
user
community
and
things
that
you're
trying
to
solve
a
problems
you
fight
you're,
facing
in
the
network
area.
C
B
B
Reflect
for
a
moment
give
ourselves
some
credit
now
with
respect
to
you
to
try
to
help
industry
address
limping
boxing
dresses
seriously
pension
around
the
adoption
is
yet
seen
on
between
swarm
and
kubernetes
back
when
that
was,
you
know
back
when
that
war
was
raging
actually
yeah.
This
is
an
excellent
vendor,
neutral
forum,
for
you
know.
B
B
B
Right
so
in
a
white
paper,
you
maybe
suggest
either
letting
it
down
in
descriptions
or
qualifications
of
what
it,
what
it
is,
the
beads
the
the
concern
I'm
interested
in
I'm,
very
pleased
to
see
me
you're
doing
this
I'm
happy
we're
having
this
discussion,
because
I'm
concerned
that
isin
very
defined
as
we're
going
to
reform
working
which
to
define
settings
that
the
question
of
what
what
seeing
should
be
created
and
the
reflection
of
how
active
reacted.
This
group
has
been
like
people
might
suggest.
B
Well,
hey,
no
one
cares
about
networking,
but
the
fact
of
the
matter
is:
there's
a
suggestion
that
there's
a
observability
snake.
It's
like
an
example
I'm
going
to
be
creating
that
won't
look
anymore
in
cloud
native
land
in
this
distributed
systems,
environment,
observability.
Half
of
that
is
about
the
network.
B
The
other
half
is
you
know
in
app
EPM,
so
I
think
there's
so
many
things
that
we
could
be
discussing
within
here
and
observability
being
you
know
or
or
a
portion
of
observability
certain
meaning
security
is
another
set
of
level
as
a
genre
of
topics
that
might
be
a
backlog.
Item
I,
don't
know
that
you
know
so
Tara
and
calico.
Some
of
the
things
that
wrap
around
it.
You
know
how
they
hooked
in
with
committees,
security
policy
and
I.
B
Don't
know
that
they've
ever
spoken
to
that
in
depth
or
some
of
the
other
projects
operetta
some
of
the
other
sins.
Yet
so
remembers,
they're
good
ones
to
also
go
to,
and
you
know
have
to
have
them
having
a
point
to
some
of
the
things
that
they're
doing
I
mention
them,
because
they're
operetta
security
right,
no.
A
Definitely
a
night
I
am
I
thought
about
adding
some
of
the
I.
Don't
put
no
security
and
I
thought
about
premiums.
Okay,
I'll.
Add
that
it's
you
know
these
days
are
so
tied
together.
Yeah,
it's
hard
to
kind
of
you
know
separate
them
out
and
they've
been
looking
the
he
said.
You
know
what
those
of
ability
point
they've
been
thinking
about
the
same
thing
with
the
security
or
the
safe
work
right
and
I.
A
Think
there's,
there's
some
overlap,
things
that
we
should
be
kind
of
careful
not
to
to
move
into,
because
it's
you
want
to
have
too
many
more
groups
trying
to
address
the
problem
from
different
angles
right.
You
know
I
kind
of
sort
of
identify
where
it
belongs
and
stick
it
in
the
right
place.
Yeah
I.
B
A
B
A
This
cloud
native
this
you
know
for
we
continue.
You
know,
brainstorm
a
move,
this
planning
and
landscape
over
to
our
current
current
sprint,
because
I
think
if
we
define
that
that
landscape
and
those
at
those
categories
upfront
that
will
help
with
sort
of
where
we
put
these
other
items
and
what
we
do
with
them.
So.
A
A
B
B
A
A
B
I'm
telling
you
these,
you
know
they
get
significantly
different
or
can
be
from
Enterprise
land.
You
know
much
of
yeah
much
into
focus.
It
seems
that
has
been
enterprise
and
user
in
lying
right.
It
is-
and
this
is
one
of
those
things.
Yes,
it's
a
network
service
mesh
that
IO
eyes
their
landing
page
to
discuss.
It
I
actually
thought
that
that
program
was
somewhat
represented
in
a
different
Linux
Foundation
group,
like
the
young.
B
B
A
You
remember
from
like,
when
you
add
Q,
Khan
I,
know
yeah,
you
wouldn't
catch
up
with
each
other.
They
little
like
work
stream
across
the
street
in
that
building,
right
where
they
were
doing
sort
of
a
bunch
of
sessions
on
network
service
mission
and
oh
and
I
have
a
specs
right.
So.
B
B
C
B
D
I
can
hear
you
hi
I'm,
actually,
I
work
for
a
company
called
Erickson,
we're
a
training
and
consulting
firm
and
I'm
a
little
bit
more
about
operational
person,
but
we
did
notice
that
you
know
we
were
trying
to
be
really
active
and
participate
in
a
lot
of
the
group's
I
was
just
lying
sitting
on
here,
so
no
more
technical
guys
are
doing
a
little
bit
working,
but
I
was
just
hoping
it's
good,
like
you
get
word
back
to
them.
D
Yeah,
but
I
I
would
be
happy
to
see
if
they
had
any
feedback
that
I
could
send
along
guys.
This
way,
as
well.
B
Part
what
Ken
was
describing
previously
in
and
around
the
white
paper
or
a
person
was
a
bit
of
best
practices
and
guidance
toward
board
operations
and
qualitative
networks
or
networks
running
club.
A
that
were
close.
It
may
be
that
maybe
it's
on
your
guys's
expertise
or
or
questions
might
be
well
placed
within
there.
A
Thing
I
thought
about
what
you
were
saying
that
I
need
is
that
the
you
know
we
have
a
couple
of
like
kubernetes
certifications
that
the
Linux
Foundation
has
been
doing
and
I'm
sure
you
know
Cisco's
already
added
to
the
you
know,
CCE
in
PCT
ie,
some
some
cloud
things
right,
but
I'm
I'm,
you
know
I've
been
thinking
long,
a
term.
You
know
with
this
workgroup
what
sort
of
changes
we
want
to
add
to
the
training
to
training
areas
for
kind
of
with
the
network.
B
B
Well,
that
isn't
that
forced
you
to
come
on
you?
Let
me
more
questions
there.
Oh
we're
close,
you
guys
are
running
of
your
network
oriented
teams,
their
particular
unaddressed
challenges,
or
you
know,
open
open
areas
of
major
concern
that
are.
You
know
whether
it's
here
a
related,
your
server
monitoring
related
plague.
You
guys
that,
are
you
treating
the
network?
As
you
know,
infrastructure
has
gone.
You
know
what
it
was
it
something
fun
for
you
guys.
D
D
Guess
I
realize
I
just
used
that
term
I'm
sitting
on
a
technical
group,
so
I
should
I
thought
about
I'm,
more
of
like
an
operational
like
like
a
administrative
operational,
so
I'm
not
sure
yeah
I
have
like
a
big
where's
but
I'm
not
sure
that
I
could
speak
in
a
way.
But
that's
super
informative
about
the
kind
of
trouble
set.
B
It's
just
kinda
nice
or
we're
okay,
but
there's
my
email
address
there.
I
am
I'm
generally
interested
toward
trying
to
help
can
outline
what
it
is
that
he
thinks
that
we
should
be
focused
on
you,
whether
it's
a
description
of
like
the
environment.
You
guys
are
within
and
stuff
that's
causing
you
pain
or
things
that
you
think
that
you've
mastered,
or
maybe
you
guys
have
best.
B
B
A
B
But
it's
not
in
a
it's,
it's
at
a
premature
State
I
think
we
I
forgot
a
popover
link,
just
as
we
were
talking
about.
Have
you
know
free
tools
and
that's
the
intention
behind
it
is
help
educate
people
and
also
help
answer
like
the
most
popular
question,
like
the
question
that
I
have
for
sure
get
I've
gotten
I've,
given
like
five
servicemen
workshops
on
this
year,
Cuba
on
the
dr.
Kahn's,
the
velocities
every
time.
Somebody
raises
their
hand,
it's
like
hey.
So
that's
really
neat.
The
service
measures
provide
a
lot
of
value.
B
What
what's
the
cost?
Like
you
don't
get
it.
You
know,
get
all
this
for
free.
What's
the
overhead
of
running
something
on
the
mesh
and
envoy
and
it's
kind
of
I
think
the
questions
ask
that
two
layers,
the
control
plane
and
at
the
data
plane
I
think
most
people
are
thinking
about
the
data
plane
and
having
to
process
all
those
packets
and
envoy
being
the
most
prominent
proxy
there.
B
You
know
more
than
just
producing
reports,
maybe
help
them
load
test
their
particular
deployment
or
benchmark
when
they've
done
a
deployment.
If
it's,
if
it's
okay,
if
it's
good,
if
it's
matching
with
others,
you
know
get
some.
Potentially
people
want
to
send
in
their
reports
to
then
have
an
anonymized.
B
B
A
About
throwing
like
the
white
paper
over
there
I
think
Buster
was
just
the
landscape
to
kind
of
define
the
categories,
and
then
they
don't
make
sense.
What
we
want
to
do
a
white
paper
at
right,
so
gonna
have
that
drive.
The
next
steps
for
us
seems
like
the
right
foot
step.
So
I
would
document
this
up.
I'll
try
to
clear
like
a
blank
template
for
that
with
some
thoughts
in
it
and
then
send
that
out
today
or
tomorrow
to
the
work
group,
email
list.
B
A
A
C
B
A
Don't
know
why
I
keep
changing
that
D
father
doesn't
never
change
all
right.
It
would
thanks
for
joining
today.
So
I
really
want
more
of
us
joining,
but
I.
Guess
it's
not
this
in
email
today
and
announced
the
next
I
mean
I
mean
my
maybe
one
a
few
days
before
it,
and
we
try
to
take
this
to
maybe
github
we
kind
of
document
and
discuss
your
network
landscape
and
do
they
come
in
the
public
domain
that
way
it
has.
You
know
more
visibility
on
it
from
the
overall
C&C
f
community
yeah.