►
From YouTube: Network Service Mesh Meeting - 2019-06-11
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
we
have
recurring
this
particular
this
particular
event,
which
occurs
every
Tuesday.
At
this
time
we
have
the
NSF
document
call
which
occurs
every
Wednesday
at
8
a.m.
Pacific
time
and
the
NSM
use
case,
which
occurs
every
2nd,
4th
and
5th
Monday
at
8
a.m.
we
also
have
the
CNC.
If
telecom
user
group,
though
we
attend,
which
occurs
on
the
1st
and
3rd
and
Monday
this
Monday,
we
should
have
a
telecom
user
group
called
coming
up.
The
link
is
posted
on
the
agenda
notes.
A
A
B
A
B
B
We
have
there's
an
upcoming
event:
D
PDK
user
space,
that's
coming
up
in
Bordeaux,
France
I,
don't
know
that
we
have
anything
directly
out
of
some
related
going
on
there,
but
definitely
good
folks,
open
social
networking
summit.
Europe
is
coming
up
and
Antwerp
Belgium,
but
CFP
is
still
for
a
tiny
while
still
open,
and
so,
if
folks
would
like
to
go
and
submit
an
aside
really
to
talk
there,
we
would
strongly
encourage
that
so.
C
C
B
No
I
I'm
very
familiar
like
you.
A
tremendous
percentage
of
my
life
in
the
last
half
decade
refers
to
development.
So
that's
also
a
good
idea.
I
mean
the
good
news,
is
there's
a
very
high
permeability
of
acceptance,
I
think,
generally
speaking,
network
service,
fresh
foxes
and
some
of
the
most
highly
attended-
and
you
know
basically
most
look
forward
to
talks.
Open-Source
networking
summit
so
I
think
we
got
a
very
good
chance
of
being
accepted.
There
well
awesome,
and
then
you
know
no,
and
also
on
that
whole
work.
B
You
know,
Renaissance
been
doing
great
work
and
making
sure
we
got
proper
separation
between
things
so
that
you
know
the
promise
the
api's
for
the
forward
or
being
replaceable
or
actually
true,
I,
think
the
last
one
he
found
was
with
some
binding
in
the
integration
test,
which
means
that
he,
if
he's
down
to
that
point,
then
good
progress
is
being
made.
Yeah.
C
B
I
tend
to
view
it
this
way
right
when
I
think
about
bugs
I
think
about
design
bugs
that
implementation,
bugs
implementation
bugs
I'm,
not
super
concerned
about
those
just
get
fixed,
yeah,
there's
on
bugs
terrify
me,
because
design
bugs
typically
get
harder
and
harder
to
fix
with
time.
It's
the
fact
that
he's
not
hitting
any
serious
design
bugs
is
really
comforting.
C
B
So
we've
got
cute
cotton,
North,
America,
coming
up
in
San,
Diego
and
late
November,
and
that
cfp
closes
July
12.
We
really
probably
I,
know
in
the
past.
What
we've
done
for
cute
con
is.
We've
often
put
together
a
page
as
a
community
where
we
sort
of
brainstorm
together
on
talks
that
we
can
do
we
can.
Various
people
can
propose,
will
probably
gonna
want
to
start
doing
that
here
again
for
cute
con
North
America
yeah.
B
It's
it's.
It's
close.
It's
super
for
Elizabeth's.
We've
been
doing
this
a
long
time.
It's
super
hysterical
because
I
think
it
was
for
cute
con
Austin,
the
CFP
for
cute
con
Austin
was
open
for
something
like
14
months.
I
have
no
idea
why
it
was
super
weird
back
in
Cuba,
I
lost
in
2017,
and
so
it
was
just
it's
funny
that
they've
gotten
to
be
this
tight.
But
you
know
it's
a
great
conference.
I
mean
really
audience.
B
I
mean
quite
honestly,
I
see
the
wisdom
in
that
suggestion,
and-
and
you
know
that
one
we
got
a
little
bit
more
space
and
time
on,
because
that
comes
down
to
it's
by
virtue
of
being
an
everyday
CNCs
project
we
do
get
either
to
shorter
or
one
longer
slots
at
cube
con
for
use
by
the
project,
but
I'd
love
to
also
get
some
things
into
the
normal
CFP
process.
Oh.
C
B
Let's
go
ahead
and
get
that
out
of
here.
If
somebody
could
get
that
out
of
here,
that
would
be
great
I'm
open
source.
It
used
to
be
Linux
summit
and
now
it's
open
source
summit,
and
so
the
flavors
got
a
little
bit
broader.
The
the
few
times
I've
attended
I've
been
quite
pleased
with
open
source
summit.
It's
a
friendly
Bunch.
A
Yeah
I
have
submitted
another
I
have
a
connection
again,
so
I
have
submitted
something
to
the
DB
DK
user
space
in
Bordeaux,
so
I
put
something
there
about
integrating
the
integration
story
between
MSM
and
BBK
in
the
same
light
as
what
I
did
at
the
Fido
make
some
focuses
on
DB
DK.
Instead,
so
I
will
let
you
know
what
happens
if,
when
I
receive
notice
on
that
I.
B
Cool
and
then
we've
also
somebody's
listed
at
edge
computing
world
in
Mountain
View
and
that's
still,
but
probably
a
good
place
to
talk
as
well
and,
as
always,
please
add
any
events
that
you've
got
an
actual
confirmed.
Msm
presence
at
please
push
a
patch
to
the
site,
so
we
can
get
them
up
there
because
that's
we're
trying
to
make
that
the
point
of
reference
for
folks
yeah.
A
So
but
they
asked
me
for
help
with
like,
with
their
chili
or
inaudible
inaugural
event
and
asked
me
for
help
to
get
to
get
them
to
focus,
what
switch
they
focus
on
and
what
should
be
what
should
be
accepted,
which
track
should
they
have
so
I'm
helping
out
helping
them
out
with
some
of
that
stuff?
So.
A
D
Hello,
yes,
this
last
week
has
been
pretty
productive
for
the
and
service
mesh
Twitter
accounts.
We
gained
six
followers,
followed
eighty
more
people
and
accounts
and
submitted
twenty-two
tweets
and
retweets.
There
were
some
interesting
things
that
popped
up.
There
was
a
request
for,
or
a
mention
rather
of
network
service
mesh
regarding
the
OVS
orbit.
Podcast
there's
also
a
lot
of
discussions
about
service
mesh
in
general
on
Twitter.
B
A
A
C
And
then
this
has
been
this
CNC
app
webinars
that
I
have
been
pointed
to,
and
we
had
I
mean
we
changed
some
messages
with
to
seem
about
it
and
she
was
kind
to
find
that
we
have
two
ways
to
actually
engage
there.
One
is
open
a
ticket
and
the
other
one
is
send
an
email
and
I
know.
Someone
told
me
that
death
schedules
pretty
tight
for
the
months
to
come,
but
maybe
we
can
do
something
like
September
October
or
something
like
that.
That's.
B
C
B
B
C
A
Yeah
I
think
you
know
always
delegated
to
to
somebody
if
someone
wants
to
take
over
and
do
the
actual
content,
so
smooth
wants
to
write
a
blog
post
or
something
similar
like
no
I
I
think
the
ticket
has
to
come
from
us,
but
I
think
that
we
could
hand
off
the
actual
task
to
to
someone
else
and
get
then
get
their
name
on
it.
Yeah.
B
And
quickly,
I
think
the
only
reason
the
request
needs
to
come
from
you
know:
Nicolai
Frederick,
myself
or
Andre
is
just
because
then
they
actually
know
we
are
representing
the
project.
It's
asking
a
little
bit
for
them
to
track
all
the
participants
in
a
project.
They
can
usually
just
track
the
the
maintainer
zhh.
A
A
A
A
C
C
C
Challenge
is
there
like
our
default.
I
mean
we
have
a
lot
of
assumptions
within
the
the
cost
like,
for
example,
not
go
detail,
but
maybe
the
surroundings
cryptic,
like
our
some
charts.
There
is
the
version,
but
I
think
that
by
default
they
just
default
to
latest.
So
you
either
need
to
specify
it
explicitly
the
version.
So
that's
not
probably
something
that
that
we
need
to
figure
out,
because
currently
the
situation
is
like
this.
C
Our
other
options
I
mean
like
we
can
have
some
automation
that
actually
tries
to
figure
out.
Are
you
on
a
crunch?
And
if
you
are
on
a
branch,
then
it
changes
things
I,
don't
know,
because
it
seems
like
there's
a
lot
of
manual
work
and
if
not
not
a
lot,
but
there's
so
many
work
any
days
manual
work.
Then
it's
error-prone.
You
know
people,
etcetera,
etcetera,
so.
D
B
D
B
It
strikes
me
that
that's
fortunate
historical
accident
would
be
really
really
nice
if
we
could
get
a
separate,
never
surfaced.
Ci
registry
clean
out
the
old
one
and
only
put
release
things
in
the
public
facing
registry.
Don't
pull
the
gravy,
though,
and
I
know
that
there
so
Frederic.
You
know
the
doctor
space
super
well.
Is
there
any
reasonably
automatable
way
to
clean
out
old
entries
from
old
label
entries
from
a
docker
registry.
C
C
A
A
It'll
be
something
because
nobody,
others
were
the
same
and
I
think
it
comes
down
to
darker
it
darker
games
more
like
it's
sort
of
like
stars
on
github.
You
know
it's
like
how
many
stars
you
have
and
it's
like
yeah
it
is
a
metric
of
like
your
project
is
growing,
but
it's
like,
but
there's
other
metrics.
You
want
to
track
instead,
like
downloads
or
so
on,
depending
on
the
type
of
your
project
and
there's
the
same
with
with
docker
it's
like
okay,
they
can
post
like.
A
B
B
E
C
B
Let's
do
the
following:
I'll:
take
an
action
item
to
go,
create
the
registry
and
the
to
go,
create
the
registry
and
get
the
credentials
in
to
circle
CI
and
a
ticket,
or
a
little
bit
of
JIRA
issue
to
migrate
over
to
using
that
just
that's
unreasonable,
yep
yeah
and
that's
sort
of
a
good
first
step
towards
actually
getting
the
mechanics
in
place.
Yeah.
A
B
C
C
B
Okay,
no
that's
fair!
So
if
you're
saying
have
latest
that
reflects
it
reflects
the
elitist
faster
I
can
live
with
that.
That's
perfectly!
Okay
with
me.
We
just
needed
to
actually
come
to
agreement
about
that.
If
that
makes
sense,
yeah
so
I'm
perfectly
fine
with
saying
we'll
always
push
the
results
of
a
successful
master
of
this
latest,
and
then
we
can
just
tag
the
things
that
our
particular
release
is.
If
people
want
to
attach
particular
releases.
That
seems
okay
to
me.
Good
then.
C
One
more
question,
because
I
now
I
have
the
release
branch
and,
if
I
build
something
in
the
release
branch.
The
change
is
that
I
mean
I,
can
push
images
there
from
the
branch,
but
because
they
are
not
tacked
I'm
not
pushing
anything.
So
does
it
make
sense
that
we
have
the
the
same
way
that
we
have
playtest
reflecting
master?
Does
it
make
sense
to
give
a
release
branch
named
like
at
least
zero
point,
one
images
sitting
there
just
reflecting
the
status
of
the
latest?
C
B
D
B
B
We
were
talking
about
the
release
repo,
because
my
initial
motion
was
for
the
release
repo.
We
only
put
release
artifacts
in
the
release
repo,
and
then
we
have
a
CI
repo
where
we
keep
the
other
things
and
nibble
I
suggested
that
we
have
latest
in
the
release
repo
be
latest
on
master.
Is
there
actually
well
established
best
practice
here?
The
year
aware
of
Fredrik
yeah.
A
So
in
the
in
the
main
repo,
we
want
the
the
latest
one,
the
default
repo
used
to
be
the
production
quality
thing
that
we
want
people
to
ship
with,
because
what
some
people
will
do
is
they'll
just
tag.
They
won't
even
put
a
tag
in
they'll,
just
put
from
Network
Service
mesh
or
they'll
stick
network
service
mesh
without
without
an
actual
version
number.
So
we
need
to
make
sure
we
always
do
them.
A
The
latest
and
it's
very
common
to
see
version
numbers
actually
tagged
like
they'll
tag
it
like
version
one
they'll
use
semantic
versioning,
so
version
11.000
or
1.5
or
whatever
it
is,
and
then
and
then
we
can
tag.
One
point
five
point
one
one
point
five
point:
two
one
point,
five
point:
three
or
so
on
so
they'll
they'll
typically
tag
the
full
version
and
then
the
version
patch
and
version
minor
and
patch.
So
that
way
they
can
peg
to
their
to
their
comfort
zone
and
beyond
that,
it's
it's
very
similar
like.
A
B
C
A
Yeah,
but
what
I'm
trying
to
remember
is
whether
latest
implied
default
or
whether
there
was
a
default
tag
and
that's
what
I'm
trying
to
remember
latest
if
it's
from
master?
That's
that's
fine
as
long
as
it's
not
the
default
branch,
and
so
that's
the
one
semantics
that
we
need
to
to
double-check
on,
but
I
think
I
thinks
:
default
was
the
default
branch
and
:
latest
is
something
that
is
arbitrary,
but.
A
B
A
Yeah,
what's
what's
nice
about
about
the
version
in
numbers?
Is
that
suppose
that
we
release
version
two
right
and
we're
working
on
version?
Three
like
default
goes
to
two,
but
version
3
is
a
master,
can
be
the
same,
and
so
we
can
actually
have
master
just
points
to
two
three
tag
effectively.
Whatever
its
label.
A
B
Can
also
run
into
a
problem
like
a
lot
of
times
when
I've
done,
docker
I
will
go
and
I
will
look
for
the
latest
release.
Looking
tag
and
I
will
grab
that
and
if
we
put
a
release
looking
kind
on
something
that
is
not
actually
a
release
thing
people
grabbing
something
that
isn't
release
thing
thinking.
It
is
yeah.
B
Know
the
implications
are
nicolai.
How
would
you
feel
about
the
following,
which
is
we
have
the
CI
repo?
We
have
the
release
repo
and
the
release
repo
latest
points
of
latest
release.
Artifact,
we
time
release
things
when
they're
actual
releases
and
we
have
a
mr.
tag
to
follow
master
and
then
for
the
throttle
branches.
Do
you
mind
the
throttle
branches
getting
stuck
in
CI
unpublish
and
the
CI
repo
and
publish,
or
what's
your
feeling
in
Thakur.
E
C
Quick
question
I
want
to
be
able
to
publish
I,
think
that
I
already
have
put
some
code
in
place
to
publish
the
images
that
I
produce
in
the
examples
like
my
only
wish
is
to
have
just
the
latest
of
the
master
ones.
So
does
it
make
sense
to
put
them
in
the
main
rapport?
We
have
a
separate
sample,
I
mean
I,
don't
know
no.
B
C
B
B
C
B
C
Okay,
the
next
thing
is
roadmap
mm-hmm,
so
I
finally
had
a
chance
to
actually
watch
to
the
fool
TG
kickoff
meeting,
where
actually
there
was
the
one
asked
from
10
about
us
having
her
old
mop
dock
is
somewhere
in
the
repo
that
we
tracked
what
we
plan
to
do
in
the
next.
You
know
three
to
six
months,
so
maybe
it
would
make
sense
to
reflect
this
somehow
in
this
in
our
spec
and
then
probably
link
to
it
in
the
main
readme
just
for
the
boat.
B
So
what
are
the
things
I've
been
trying
to
do
visually
for
that
under
the
entry
mm-hmm
I've
actually
been
putting
that
up
there
and
it's
not
fully
complete
guy
I.
So
if
you,
if
you're
like
where's
my
thing
and
let's
get
your
thing
there,
but
I've
also
been
trying
to
go
and
put
into
them
that
go
to
the
issues
specs
around
them,
so
that
if
you're
sort
of
like
okay,
what
are
they
doing
around
that
you
can
actually
get
to
it?.
B
D
B
We
have
things
captured
here
that
don't
have
specs
yet
and
we
need
to
sort
of
figure
out
how
those
all
fit
in
together.
So
maybe
starting
by
linking
to
here,
would
be
a
good
start.
Also,
like
you
know,
I
tend
to
think
in
pictures
and
some
other
people
think
better
in
terms
of
text.
So
it
might
be
worth
bouncing
off
the
documentation
group
to
see
if
they
have
any
opinions
about
how
this
might
be
better
represented.
B
C
B
B
I'm
inai
I
would
definitely
say,
like
we
could
just
get
a
little
check,
check
box
image
and
attach
them
here
as
we
complete
things
because
I
know
there
is
some
work
underway
on
security,
I
think
there's
some
work
starting
to
get
underway
mdns,
you
know,
so
you
know
that
there's
there's
some
forward
progress.
I
was
just
talking
to
someone
yesterday
who
may
turn
up
in
the
community
to
start
working
on
the
sessions.
Payload,
Stas
and.
B
B
And
in
my
get
some
other
things
here
that
are
not
clearly
represented,
like
I
know
that
there
was
a
there's.
A
very
interesting
conversation
happening
around
possibly
more
sophisticated
selection
algorithms
than
round-robin
once
you've
got
a
class
of
candidates,
and
that
could
be
interesting
as
well.
C
B
I
definitely
think
for
a
lot
of
these
things.
You
know
security,
inter-domain,
probably
all
the
way
through
Hardware
next
I
think
it's
probably
stuff.
We're
gonna
want
to
do
other
next
three
to
six
months,
but
like
everything,
it
depends
on
who
turns
up
interested
in
working
on
what
in
the
community
I
mean.
This
is
always
this
is
the
age-old
question
of
open
source
projects
and
people
asking
them
for
roadmaps,
and
the
answer
is
sore
depends
on
who
shows
up
and
what
they're
interested
in
yeah.
C
B
To
referred
well
in
the
US,
we
just
don't
take
vacation,
okay,
but
Europe
Europe
is
a
satyr
in
more
civilized
place,
and
so
my
only
real
complaint
with
general
European
Saturday
around
summer
vacation
is
it
would
be
convenient
for
the
rest
of
the
world.
If
you
would
all
take
the
same
month
off
in
the
summer,
you
said
them
during
it
across
the
continent.
No,
we.
A
B
E
C
We
like
next
three
three
demos.
It
means
like
by
September.
You
think
we
will
have
some
of
these
I
know
that
you
know
there
are
various
people
looking
come,
various
things
I
mean
I'm
I'm,
just
seeing
it
happens
with
the
word.
Go
with
group
Group
Co
attendants
like
it's
like
people,
I
guess
that
are
traveling
or
taking
time.
Also,
we
probably
will
get
some
low
on
attendance.
B
Here's
the
thing
I
would
suggest
like
this
actually
gets
to
a
higher
level
conversation
which
is:
do
we
want
to
do
time-based
releases
or
do
we
want
to
be
a
feature
based
releases
I
tend
to
in
open
source
communities?
I
tend
to
prefer
time
BIST
just
because
if
you
go
feature
based
and
all
kinds
of
weird
stuff
can
happen
because
of
varying
participation,
but
I
am
curious.
What
other
people's
opinions
are.
C
Well,
I
mean
we
had
an
attempt
to
to
align
the
releases
around
Keep
Calm.
These
are
kind
of
you
know.
Six
months,
yeah.
B
Please
no,
like
time-based
releases.
My
experiences
really
use
projects
grow,
they
become
sort
of
roughly
time-based.
Kubernetes
is
roughly
quarterly,
but
you
know,
eventually
you
get
better
at
having
your
time.
I
know,
for
example,
like
the
Eclipse
project
tests
like
from
15
years,
hitted
States
very
complicated,
show
of
things
that
go
into
it.
So,
let's
think
we
got
some
comments
in
the
chat
on
liking
time
based.
B
The
conference
driven
child
things
I.
D
C
Okay,
so
around
their
own
mob,
I
think
that
that
we
more
or
less
are
agreeing
on
what
should
be
there.
Maybe
there's
no
point
in
to
continuing
to
to
put
it
up
front
again
like
to
put
it
on
the
on
the
discussion
here.
Maybe
we
should
try
to
convert,
maybe
tomorrow's
work
documentation
call
on
how
we
want
to
expose
this
to
the
world.
You
know
a
repo,
etc
I.
D
B
C
C
C
B
B
So
that's
that's
two
really
good
groups!
Writing
that
right
there
off
the
top
and
I
suspect
you'll
get
some
personas
under
that
that
we
can
work
out
as
personas
and
then
just
sort
of
figure
out.
What's
the
minimum
viable
set
of
things
that
those
personas
need
in
order
to
be
able
to
do
deployments,
I
think
that's
probably
good
advice.
B
C
I
was
looking
at
the
at
the
CNCs
dashboard.
Think
not
the
guys
from
Vogue
shoot.
True,
probably
tell
us,
but
would
it
make
sense
if
we
want
to
run
something
there
I
mean?
What's
the
start
of
the
CN
I
saw
that
the
latest
kubernetes
is
one
point.
Thirteen.
Is
it
something
that
people
are
looking
into
me?
Does
it
make
sense
for
us
to
also
try
to
go
there
with
some
I,
don't
know
what
they
were.
G
My
project
thought
something
that's
saying
there,
but
it's
something
that
we've
had
kind
of
in
in
the
back
of
our
mind:
yeah
interoperability,
type
views
and
stuff
and
I.
Think
at
that
point
you
could
see
something
like
innocent
being
used
on
a
for
those
use
cases
similar
to
what
singing
the
scene
attest.
That
is
yes,.
E
G
C
G
Think,
as
far
as
something
that
would
be
Saints
in
or
would
be
on
the
CNF
test
bed
and
if
we
can
come
up
with
functional
type
use
cases
where
we
say
here's
different
parts,
how
they
fit
together
in
a
something
similar.
Maybe
to
do
what
the
own
@vc
use
case
is
it's
trying
to
show
a
a
specific
type
of
functionality
this
desired.
G
G
D
G
Understand
yeah
I.
Definitely
don't
want
to
rename
right
now.
I'm,
just
highlighting
I
agree
that
the
naming
doesn't
work
for
everybody,
probably
just
if
we
can
start
documents
in
the
conversations
with
who's
the
persona
and
the
viewpoint
that
we're
taking
right
now,
with
whatever
the
demo
commentaries
use
cases
are
then
it'll
be
helpful
for
everyone.