►
From YouTube: CNCF Network Service Mesh Meeting - 2019-06-18
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
B
A
Cool
so
welcome
to
the
next
Network
circus
match,
meaning
so
before
we
get
started.
If
you
have
anything
to
put
on
to
the
agenda,
please
bring
it
up
or
add
it
on
and
and
we'll
see
about
about
getting
to.
It
also,
please
add
yourself
to
the
attendees
list,
so
the
attendees
list
is
really
useful,
because
when
we
go
and
do
outreach,
it
makes
it
very
easy
for
us
to
say
that
we
have
a
community
there
that
people
who
are
involved-
that's
diverse.
A
A
We
have
this
meeting
every
Tuesday
at
this
time
the
MSM
doc
meeting,
which
occurs
every
Wednesday
at
8
a.m.
Pacific
time
and
the
units
in
the
use
case,
which
occurs
every
second
fourth
and
fifth
at
8
a.m.
Pacific
time
as
well.
There
will
be
a
use
case
meeting
this
coming
this
coming,
Monday,
so
mark
your
calendars,
for
that
we
also
participate
in
the
CN
CF
Telecom
user
group,
which
occurs
every
first
and
third
Monday
at
8
a.m.
which
rotates
with
the
use
case.
A
We
have
Shanghai
cube
con
China
coming
up
in
Shanghai
and
Nicolai
and
I
will
be
giving
an
intro
maintainer
track.
There
is
a
CNC
F
answer
bar
that
will
also
be
on
I
I.
Suppose
somebody
will
be
manning
it
from
NSM
from
11:30
to
12.
That's
me
cool,
and
we
have
a
tug
birds
of
a
feather
kickoff,
which
is
going
to
occur
on
Tuesday
from
11
to
11:35,
actually
that
that
time
seems
short.
We
should
double-check
that,
because
last
one
was
an
hour
and
a
half,
and
this
one
showing
only
half
an
hour.
A
A
A
C
A
A
So
we
have
ons
Europe
coming
up
and
the
call
for
paper
for
that
is
closed
and
I,
believe
the
the
submissions
should
either
come
out.
I
think
the
end
of
this
month
or
early
next
month
that
terms
of
the
enters
of
the
selected
schedule
that
will
be
held
in
Antwerp.
We
have
ons
Europe
coming
up,
of
which
the
call
for
paper
closes
on
July,
1st
and
I.
Believe
decline
is
applying
for
a
talk
there.
B
B
A
A
Oh
so,
in
terms
of
we
have
cube
con
call
for
papers
coming
up
the
closes
on
July
12th,
so
we
need
to
get
a
shared
document
and
start
gathering
ideas,
so
the
cube,
calm
ones
tend
to
give
us
a
lot
of
mileage,
and
so,
if
you
could
only
attend
one,
that
would
be
the
Q
Khan
would
be
high
up
on
that
list.
It's
not
at
the
top.
A
We
have
also
will
be
co-located
events.
We
know
that
there's
a
void,
Khan
will
be,
will
be
there
as
well.
I
intend
to
put
something
forward
on
envoy
in
that
particular
talk,
we're
not
particular
co-located
event
and
I'm.
Pretty
sure
they'll
be
there'll,
be
a
range
of
others
that
we
can
participate
in
December,
10th
and
11th.
We
have
edge
computing
world
coming
up
and
Mountain
View.
The
website
is
still
to
be
to
be
announced
and.
E
C
Good
morning
great,
we
have
a
nice
post
from
VMware
open
source
yesterday,
so
I
link
to
that
there
and
in
the
last
week,
15
more
accounts
followed
and
service
mesh.
We
followed
61,
more
accounts
and
posted
23
times.
I
posted
about
the
Cisco
live
event,
the
cube
con
china
Save
date
and
some
CNCs
events
as
well.
So
we
already
talked
about
the
open
core
summit.
Dm
there's
also
a
mention
question
from
someone
and
I
believe
Frederick
may
have
responded
and
continued
on
the
conversation
about
that
No.
So
that's
there.
D
C
You're
welcome
so
that's
linked
here
in
the
notes
and
there
I
also
notice
a
call
for
demos
and
the
kubernetes
community
meetings.
Those
are
held
every
Thursday
at
10
a.m.
Pacific
time
I
believe
they
have
things
on
their
calendar,
maybe
through
to
June,
maybe
mid
July.
So
if
there's,
if
you'd
like
to
do
a
demo,
there's
a
link
here
for
the
kubernetes
community,
so.
B
I
was
directed
here
also
because
I
reached
out
for
this.
What
was
the
name
of
the
event
that
we
wanted
to
do?
The
webinar
yeah
so
essentially
I
think
it
was
Chris.
Doing
answering
me
that,
because
we
are
sandbox
project,
we
are
not
supported
for
a
webinar
as
a
project.
We
can
do
it
from
the
company
perspective
like
VMware
or
Cisco,
and
some
of
the
companies
that
are
participating
there
and
they
were
say
essentially
saying.
B
C
B
A
B
Thank
you
so
and
on
the
webinar
just
to
share
my
experience
with
the
service
desk,
everything
went
really
smoothly.
Quick
I
mean
like
in
an
hour.
It
was
clear
that
yeah
when
they
work
a
couple
of
people,
involved,
sending
responses
back
and
forth
and
in
the
end
it
was
clear
what
what
we
should
do
so
just
to
share
that
if
we
need
it,
it
works
with
really
well
the
service.
That's
coming.
C
B
D
A
A
Especially
in
this
area,
so
we
we've
notified
Nasheed
that
we
were
they
were
considering
putting
it
on,
and
so
she
wanted
to
know
as
well.
If
there's
any
problems
with
with
the
wording
or
anything
like
that
as
well,
because
the
the
wording
can
be
changed
or
clarified
as
well.
So
if
there's
any,
if
there's
any
problems
with
the
code
of
conduct
that
we
flagged
they
are,
there
are
something
we
can
bring
up
with
them
or
bring
them
with
Missy.
A
A
D
And
everything
one
other
thing
I
will
suggest
to
folks
look
into
with
the
you
with
a
work
in
progress
tag.
There
is
actually
a
new
github
feature
called
draft
that
will
let
you
push
a
poll
released
in
the
draft
mode,
which
literally
precludes
it
being
merged
until
you
click
the
publish
button,
oh
and
and
that
I'm
finding
to
be
super
cool
because
I'm
working
in
other
places,
where
sometimes
people
are
super
happy
to
merge
the
thing
you
pushed
and
it's
like.
No
stop
stop
not.
A
B
Okay,
so
the
the
situation
here
in
the
on
the
release
hasn't
changed
that
much
since
last
week,
except
for
we
now
have
to
dr.
hope,
propose
where
we
can
push
different
images
and
I.
Guess
that
it's
going
to
be
a
manual
process
for
now
like
publishing
the
docker
images,
or
at
least
I,
can
really
just
talk
and.
D
Actually
use
the
network
CI
repo,
because
right
now
we
do
have
the
repo,
but
it's
not
being
used
and
I'm
chasing
down
a
few
bugs
on
the
ER.
There's,
actually
a
link
in
the
meeting
agenda
pointing
to
that
outstanding
action
item
so
once
I
got
that
so
it
will
actually
merge.
Then
we
can
merge
it
and
merge
it
also
to
the
branch
and
then
we'll
start
publishing
to
the
network
service
master,
CI,
repo
for
the
CI
artifacts
and
then
from
there.
We
can
look
at
actually
handling
proper
publishing.
D
The
way
we
discussed
for
the
main,
repo
and
I
think
the
things
we
discussed,
the
main
repo
was.
We
would
have
things
tagged
as
releases.
We
would
have
a
latest
tag
that
pointed
to
the
latest
release
artifact.
We
would
have
a
master
tag
that
points
to
the
latest
stuff
coming
off
master
and
then
we
would
have
for
release
throttle
branches.
You
know
something,
like
you
know:
branch
dash,
V
0.1,
that
points
to
the
latest
from
that
branch.
That
was
my
understanding
of
the
last
discussion.
That's
what
I'm
trying
to
do
in
the
PR.
F
D
D
B
B
B
B
Them
so
I
I
was
looking
at
some
other
project
which
one,
but
people
were
having
talks
like
near
LEDs.
Next
release
like
more
generic
tiles.
That
actually
can
somehow.
This
is
a
slide
house,
a
introduction
to
the
next
topic,
something
that,
like
you,
can
reflect
your
roadmap
in
your
issues
like.
If
you
see
what
I
mean
like
it's.
D
Actually
interesting,
it
has
the
benefit
that,
when
the
the
stuff
that
you
didn't
actually
get
to,
that
was
scheduled
to
next
release
if
it
doesn't
get
done
next
release,
it's
still
next
release,
yeah.
B
I
mean
like
we
have
a
couple
of
things
which
are
still
pending,
go
from
zero
point,
one
which
are
obviously
not
getting
there,
because
the
branches
they're
doing
active
development,
and
we
need
to
move
them
to
the
next
release
if
we
were
planning
for
it
or
we
need
to
can
kind
of
put
them
in
some
categories
saying:
okay,
this
is
near
future.
This
is
distant
future
and
blah
blah
blah.
B
B
D
Then
there's
the
spec
board,
because
I
don't
think
I
want
to
be
super
clear.
The
technology
tree
is
probably
not
capturing
everything
yet
yeah,
and
so,
if
you're
sitting
there
going
where's
my
spec,
the
answer
is:
please:
let's
get
it
worked
in
there
right
because
I
know,
we've
got
lots
of
people
doing
lots
of
interesting
things
in
the
in
the
spec
space
about
what
we
could
go
forward
and
not
everything.
That's
gotten
as
much
love
as
I
would
like,
because
the
world
is
full
of
shiny
objects.
D
Something
from
the
talk
proposal
looking
at
this
is
I've
glanced
at
SMI
and
I,
like
a
lot
of
what
they're
doing,
I'm,
not
sure
how
that
relates
to
network
service
mash,
but
I
agree
with
the
intuition
that
maybe
somehow-
and
so
you
know,
as
a
Vonnegut,
you
closer
to
sort
of
figuring
out
something
that
makes
sense
there.
I
would
love
to
see
a
spec
for
that
just
so
we
can
get
a
sense
of
how
we
might
you
know,
dealt
a
little
bigger
communities,
yeah.
A
B
G
They
don't
seem
very
open
for
interaction,
yet
they
have
one
general
select
channel
that
it's
not
pretty
active,
they
don't
have
any
meeting
say
so.
Issues
opened
from
external
people
with
no
comments
and
poor
requests
are
from
the
few
people
that
start
in
the
project,
so
I
think
I
get
some
ideas
where
to
start
from.
If
we
want
some
integration,
but
it's
this
moment,
I'm
still
not
sure
how
we
can
interact
well
with.
Yes,
my
community
I
think
we
might
need
to
wait
a
bit.
I
mean.
D
My
general
counsel
would
be
probably
patients,
I've
worked
with
tons
and
tons
and
tons
of
communities,
including
lots
of
communities
in
their
initial
stages
and
starting
a
community
is
way
harder
than
it
looks,
and
so
usually,
when
I
see
a
community,
that's
like
the
one
you
just
described
before
SMI,
it's
literally
less
than
a
month
out
of
the
gate,
and
it
seems
to
be
having
trouble
setting
up
for
interaction.
You
just
check
back
in
a
month.
They
usually
much
better.
A
Yeah,
even
the
the
initial
network
service,
much
meeting
started
off
like
that,
like
it
was
me
and
add
in
a
couple
people
hopping
on
as
like
man,
we
need
to
get
more
people
on
here,
so
yeah
I
think
it's
an
important
problem,
though
so,
and
the
people
driving
it
are
pretty
high.
They
do
have
some
pretty
high
profile
people
on
it,
so
I
think
they
will,
if
they,
if
they
keep
up
with
it,
I
think
they
will
build
a
community
around
it.
G
Yeah
I
ask
you
have
concerns,
it
seems
like
asthma
will
be
successful,
at
least
from
politics
point
of
view.
They
are
going
to
push
for
it,
but
yeah
I
think
it's
good
to
wait.
A
bit
I'm
continuing
with
the
research
I
may
start
trying
some
things,
but
let's
see
how
things
are
going
to
developing
the
direction.
I
find
that
East,
your
people
are
not
very
interested
yet,
but
it's
still
not
sure.
Maybe
they
will
become
interested
in
the
near
future.
D
I
mean
one
of
the
nice
things
with
all
these
layer.
7
service
meshes
you
dovetailing
with
them,
is
lovely,
but
but
do
the
wonders
of
layer,
separation
and
networking.
You
can
run
the
layer,
7
server
semester
on
top
of
network
service
fashion.
They
don't
have
to
know,
and
so
that
you
know-
and
we
don't
have
to
really
know
that
much
either
there
are
a
couple
places.
We
could
probably
do
some
helpful
things,
but
yeah
so
I
mean
I.
D
G
Even
think
that
the
SMI
people,
even
if
it's
your
computers,
are
not
interested
I,
think
that
I
see
my
contributors
would
make
some
integration
or
even
try
to
to
run
together
with
SEO
to
integrate
it
on
their
own.
In
order
to
to
be
able
to
provide
the
common
API
and
not
share
it
separated
having
called
small
meshes
and
exploding
system,
I
think
there.
A
If
the
two
areas
I
would
recommend,
looking
at
furnace,
am
I
integration
at
this
point
and
I'm
not
even
implementing,
but
even
just
like
respecting
out
and
thinking
about
it,
but
the
first
one
is
enablement
being
able
to
to
add
a
service
mesh
by
policy,
and
so
we
we
already
have
an
envoy
example
which
can
be
used
as
a
starting
place.
For
that
kind
of
that
employer
happen
to
be
configured
to
work
with
the
service
fish,
then
that
would
that
would
get
you
an
initial
interaction.
A
To
these
this
cluster
and
the
second
block
of
IP
addresses
to
these
clusters
that
way
they
know
them
and
that
way
they
don't
collide
and
then
will
try
to
turn
them
into
an
into
a
giant
network
with
the
east-west
traffic
between
them
and
rather
than
go
through
that,
perhaps
a
question
would
be:
how
do
you
connect
workloads
to
workloads
between
between
clusters
using
network
service
mesh
which
could
include
an
envoy
or
service
mesh
component
to
that?
So
those
would
be
the
two
areas
I
like
that.
A
B
But
do
we
want
to
put
on
the
roadmap,
maybe
in
the
distant
future,
but
still
kind
of
at
least
starting
to
discussion
with
our
fellow
CNCs
projects
like
lickity,
or
maybe
not
that
fellow
senior
projects
like
Easter
or
some
kind
of
underlying
integration
and
verification-
and
maybe
maybe
measure
is
I,
don't
know
they
all
seem
to
be
connected
to
all
these
technologies.
So
what
people
think
about
this
I.
D
F
D
F
A
Yeah
spiffy
is
really
spiffy.
I
think
I
think
that
that's
actually
a
really
good
example,
because,
when
part
of
part
of
what
we're
gonna
have
we're
gonna
have
to
solve
is
how
do
you
know
who
you're
connecting
to
is
actually
who
they
say
they
are,
and
not
just
some
intercepted
or
random
thing
and
inspire
give
us
a
path
to
that.
But.
B
I
mean
if
we
take
our
multi
cluster
I,
don't
remember
where
it
is
here,
maybe
on
the
on
the
roadmap
top
like
floating
hinder
domains
or
in
general
inter
domain.
If
we
had
this
feature-
and
we
like
enable
like
traditional
service
mention-
let's
say
layer,
7
application
service
measure
on
top
of
network
service,
merge
then
essentially
for
them,
it
would
be
pretty
transparent
that
there
is
kind
of
disinterred
to
main
thing
going
on,
because-
and
it's.
D
B
A
If
you,
if
you
step
out
of
the
service
mesh
domain
for
a
bit,
then
you'll
start
to
see
some
other
types
of
activity.
So
that's
that's
why
I
mentioned
like
like
the
joining
of
l2
or
l3
domains
is
is
I,
know,
I,
know
of
a
couple
efforts
where
people
are
groups
are
trying
to
to
perform
that
using
various
types
of
of
VPN
strategies,
and
so
on
and
I
can't
I
can
tell
you.
A
They've
worked
on
those
for
well
over
well
over
a
year
and
still
haven't
produced
anything
that
that's
usable
at
this
point
and
when
I
talk
with
people
in
those
areas,
they're
frustrated
and
so
that
inter-domain
problem
is
definitely
a
big
problem,
and
it's
definitely
something
that
that
when
we
get
around
to
solving
it,
we
can.
We
can
get
a
lot
of
mileage
out
of
out
of
the
inter
domain
because
well
will
effectively
enable
the
federated
multi
cluster
and
not
just
a
federal
Moltar
cluster
but
across
other
other
devices,
domains
as
well.
A
B
I
think
that
we
mentioned
the
last
time.
That
is
actually
one
of
the
steps
that
we
need
to
take
into
the
direction
of
pops.
A
maturing
as
a
project
and
moving
into
from
sandbox
into
incubation
is
to
really
have
some
okay,
its
production,
as
possible
deployments
I
mean
even
if
someone
deploys
saying
same
in
a
CI
that
puts
tube
may
be
considered
as
some
kind
of
production
type
of
thing
for
some
people,
but
I
mean
I'm.
D
F
H
I
would
say
even
on
the
service
provider
space
right,
like
I
host,
like
hundreds
and
probably
some
day,
thousands
of
clusters,
and
sometimes
it's
workloads
that
span
these
clusters.
So
even
in
the
MSP
space
inter-domain
is
everybody
wants
to
do
hybrid
cloud
right,
multi
cloud
like
it
doesn't
matter
if
you're,
a
to
location,
pizza
shop
that
uses
AWS
or
like
you
just
want
all
your
crap
to
talk.
F
H
F
D
B
D
D
During
it,
some
connection
requests
clothes,
as
the
security
will
be
on
there's
inter-domain
yeah,
but
be
that's
why
I
spent
so
much
time
talking
to
the
spiffy
inspire
guys,
because
it
turns
out
the
heart.
The
the
good
news
is
the
really
hard
problem
for
us
for
security.
The
spiffy
spire
guys
are
solving,
which
is
how
to
handle
authentic,
a
double
identity
and
then
the
from
there.
We
have
a
little
bit
of
problem
around
provenance
that
turns
out
to
be
relatively
easily
solvable
and
it's
in
the
security
spec
and
then
you
know
from
there.
D
There
will
eventually
be
an
authorization
policy
problem,
but
I
think
we
can
slow
that
off
on
the
open
policy
agent,
guys
for
authorization,
because
people
are
gonna
have
all
kinds
of
opinions
about
who's
authorized
for
what.
But
if
you
can
secure
identity,
which
is
what
spiffy
inspired
us
for
us
and
you
can
get
the
information
about
provenance,
in
other
words,
who's
been
involved
in
getting
this
for
me
to
you
and
me
then
you're
in
pretty
good
shape.
D
The
more
eyeballs
for
security
stuff
is
always
better,
and
the
other
thing
I
wanted
to
mention
briefly
as
I
know
that
there
was
a
gateway
spec
that
Matthew
that
opened
and
I
need
to
actually
go
and
put
some
comments,
because
I
think
the
recent
update
to
the
inter-domain
spec
actually
will
cover
the
Gateway
thing,
but
I
would
want
to
make
sure
that
Matthew
actually
concurs
there
cuz.
He
often
thinks
very
deep
thoughts.
A
Yeah
I
revised
my
comment
about
limited
versus
big
business.
Huge
I
think
the
use
case
itself
is
massive.
I
think
the
number
of
people
who
are
going
to
be
willing
to
implement
it
is
going
to
be
limited,
which
is
good
for
us,
because
that
means
we
can
pretty
much
work
with
them
once
we
demonstrate
something
working
and
we'll
get,
and
that
should
be.
That
should
be
an
easy
driver
for
us
to
to
grow
the
enterprise
community.
Quite
saying
that
we
can
link.
A
Actually,
one
of
the
things
that's
I
really
love
about
the
technology
tree
is
it
reminds
me
of
playing
a
video
game
and
it's
like
you
can
choose
where
to
where
you
spend
your
skill
points
and
which
tree
do
you
want
to
get
pursed?
And
you
know,
and
it's
the
type
of
game
you
can
get
all
of
them,
but
you
have
to
but
bad
early
game.
You
know
and
your
how
fast
you
can
level
up
depends
on
where
you
spend
those
skill
points.