►
From YouTube: 2020-06-25 Istio Community Meeting
Description
Istio in 5 mins, meet our new T/PM, comparing meshes in the CNCF SIG, and looking at creating multiple control planes using the Istio Operator.
A
I
have
just
started
recording
my
name
is
dan
cerulli.
I
you
may
have
seen
me
at
these
before
I'm
a
product
manager.
I
work
at
google
I'm
sitting
on
the
steel
steering
committee
and
I
we
we
didn't,
have
a
plan
for
running
this
meeting,
so
I
am
stepping
in
here
at
the
last
minute.
I
will
share
my
screen
in
just
a
second.
A
A
Maybe
it
was
in
the
last
two
weeks-
it's
possible.
It
was
a
little
before
that,
but
I
didn't
see
any
references
to
it
down
here.
The
wonderful
megan
o'keefe
ask
me
on
twitter.
It's
the
only
five
minute,
video
that
if
you
have,
if
you're
unclear
on
exactly
what
this
is,
or
you
have
co-workers
people
that
you'd
like
to
explain
istio
well,
this
is
a
really
really
good
primer
on
on
what
is
doa
is
how
it
works.
A
So
I
highly
encourage
everybody
to
check
that
out.
These
meeting
notes
are
the
agenda
is
attached
to
the
calendar
invitation.
This
is
the
one
that
I'm
displaying
so
yes
feel
feel
free
to
check
that
out.
It's
a
great
video
megan
is
the
person
who
often
does
feature
walkthroughs
of
istio
most
istio
releases.
A
She
will
do
a
5,
10
15
minute
video,
where
she
just
walks
through
the
new
features
which
are
invaluable.
This
is
just
the
primer.
What
does
this
do?
What
do
you
use
it
for
so
really
good
if
you're
just
new
to
the
service
mesh
space
or
just
kind
of
understanding?
What
what
istio
is
all
about?
It's
great
so
feel
free
to
to
share
that.
A
Yes,
as
I
said,
I
I
I
was
thinking,
maybe
I
would
put
schweita
on
the
spot.
Shweta
singh
has
joined
google
relatively
recently
and
we're
lucky
enough
to
have
her
not
only
working
on
our
commercial
stuff,
but
she's
really
really
involved
in
open
source,
and
before
I
put
you
on
the
spot,
schweitzer
to
talk
a
little
bit
about
the
107
planning.
I
would
like
to
talk
about
the
fact
that
people
think
a
lot
about
contributions
to
open
source
in
terms
of
developers,
and
we
have
a
lot
of
developers
who
work
in
open
source.
A
A
lot
of
the
stuff.
That's
in
github,
of
course,
is
written
in,
go
it's
written
in
c.
It's
it's!
It
is
code,
however,
there's
a
lot
of
stuff
that
goes
on
in
an
open
source
project.
That
is
not
just
writing
code.
For
example,
we've
got
a
lot
of
stuff
up
there.
That's
written
in
markdown
and
some
of
that
stuff
is
written
by
developers,
but
a
lot
of
it
is
written
by
other
people
as
well.
So
we
have
writers
who
work
on
istio.
There
are
marketers
from
several
companies
who
work
on
istio.
A
We
have
some
product
management
involved,
although
clearly
not
enough.
Schweda
is
a
tpm
and
at
google,
the
role
that
tpm
is
to
help
us
help
ensure
that
we
deliver
the
things
we're
trying
to
deliver
and
it's
something
that
is
invaluable
in
in
actually,
you
know
communicate
understanding
and
communicating
what
you're
going
to
deliver
and
so
shweta
has
joined
recently,
and
I
think
one
seven
is
the
first
release
where
we're
really
going
to
feel
her
the
the
positive
effect
of
of
having
the
the
coordination
so
schweitzer.
A
Would
you
like
to
talk
to
the
group
a
little
bit
about
you
know
what
we're
doing
differently
for
the
one
seven
release,
sorry
for
putting
you
on
the
spot,
like
this
feel
free
to
pretend
you
have
network
difficulties.
If
you
don't
want
me
to
put
you
on
the
spot,
but
do
would
you
like
to
talk
a
little
bit
about
how
things
are
gonna
be
different
and
and
the
positive
effects
we'll
see.
B
Absolutely
so
just
a
small
correction,
I
I
joined
in
the
middle
of
istio
1.5.
Okay,
so
help
support
that
and
then
it's
tier
1.6
when,
when
I
joined,
obviously
from
that
to
the
current
istio
1.7.
As
far
as
I
understand
there
has
been
lots
of
improvements
in
terms
of
how
the
planning
will
happen.
How
do
we
actually
present
it
to
the
toc?
B
One
good
part
which
we
have
done
is,
and
even
with
one
1.6,
is
right:
every
working
group
they
actually
present
what
they're
planning
as
a
delivery
with
their
prioritization
done
contributor
added
to
those
the
release
type,
whether
it
should
be
alpha
beta
or
ga.
There
are
various
aspects
of
deciding
what
the
release
types
will
be.
Those
are
being
presented
now,
mostly
it
gets
approved.
B
B
So
one
of
the
improvements
is
automating
most
of
the
documentation,
so
we
spend
the
less
time
on
the
testing
we
used
to
have
join
the
community
testing
days,
which
happens
two
to
three
weeks
before
the
release,
and
that
is
a
really
big
process
of
having
multiple
engineers
sitting
in
a
room
for
quite
a
long
time
for
six
to
four
to
six
hours
to
do
the
testing
so
automating.
That
is
very
helpful
for
us,
so
that
either
some
of
them,
which
is
not
get
automated,
can
be
done
asynchronously
or
if
the
automation
is
done.
B
It's
just
a
matter
of
pressing
a
button
and
looking
at
the
results,
so
our
is
tier.
1.7
is
slated
to
release
on
11th
august.
I
highly
encourage,
if
you
are,
if
you
can,
help
being
a
contributor
to
the
open
source,
help
us
with
testing
the
testing
days.
Right
now
are
planned
for
29
to
july
and
4th
of
august.
So
if
you
have
any
questions,
what
needs
to
get
tested
where
you
can
help
contribute,
please
reach
out
to
me
or
any
of
the
working
group
leads.
A
B
B
That's
when
I
start
working
with
all
the
groups
and
get
the
dates
out.
One
more
thing:
people
have
not
seen
the
newsletter.
All
those
actually
goes
in
the
newsletter
as
well.
I
do
publish
the
newsletter
in
discuss
dot
io
website
as
well
and
pin
it
for
a
week.
So
all
the
dates
are
there
as
well.
If
you
are
entrusted,
please
follow
that
newsletter.
It
has
all
the.
A
C
Yeah
we
do
yeah
well,
I
just
came
on
to
just
see
how
things
are
going
here
in
the
community
but
yeah.
We,
we
announced
a
couple
of
things.
I
think
would
be
interesting
for
this
group.
The
developer
portal
for
istio,
which
is
a
is
a
product
that
allows
you
to
kind
of
treat
the
seo
ingress
gateway
as
a
lightweight
api
gateway.
C
With
with
the
ability
to
expose
apis
through
a
a
customizable
portal.
It
comes
with
things
like
rate,
limiting
server,
an
external
auth
server
and
ability
to
document
your
apis,
expose
them
to
users
and
have
themselves
serve
and
sign
up
for
api
keys
and
that
kind
of
thing.
So
that
was
one
thing,
and
then
we
also
released
the
new
version
of
glue
yesterday
1.4
with
support
for
stu
1.6.
C
A
Cool
okay,
as
I
said
we
we
didn't.
We
on
the
steering
committee
usually
passed
this
past
the
baton
for
running
these
and
this
week
we
we
forgot
to
do
that.
So
no
one
was
holding
it.
There
was
no
pre-plan
agenda,
so
usually
we
have
a
little
bit
of
a
pre-planned
agenda.
We
might
go
through
a
few
points.
Then
we
open
up
the
floor.
So
if
anyone
has
any
questions
that
they'd
like
answered
about
istio
anything
specific,
sometimes
we
have
people
on
the
call
who
can
answer
it.
D
D
D
Two
items
that
I
think
may
be
of
interest
to
well,
hopefully
to
excuse
me
hopefully
to
everyone
on
the
call
one
is
that
there's
we're.
So
one
of
the
hats
that
I
wear
is
as
the
chair
for
the
cncfs
network
sig
and
we
formed
a
service
mesh
performance
working
group
in
there.
There's
a
few
objectives
within
that
group
that
I'll
provide
a
link
to
so
that
folks
might
check
it
out
if
I'm
interested,
this
might
be.
D
D
A
general
call
for
participation
there
if
people
are
interested,
we're
hopeful
to
help
empower
those
that
are
adopting
istio
to
well
better
understand
and
manage
their
the
sort
of
the
the
quote-unquote
overhead
of
a
mesh.
But
but
do
it
in
context
of
the
value
that's
being
provided
by
the
mesh
and
sort
of
help.
People
recognize
all
of
the
value,
all
the
things
that
the
mesh
is
doing
and
then
weigh
that
against
any
additional
latency
or
memory
overhead
or
cpu
or
something.
You
know.
A
D
A
D
D
That
yeah,
even
in
a
agnostic
fashion,
is
intended
to
characterize
the
configuration
of
a
mesh
configuration
of
your
workload,
end
of
your
performance
test
to
sort
of
capture
all
of
those
variables
into
a
common
format,
and
then
yeah.
Let
people
measure
their
environment,
tune
their
environment
and
then,
and
hopefully
squeeze
as
much
out
of
a
mesh
as
they
can
out
of
the
skill
as
they
so
can.
So,
there's
a
call
for
there's
a
couple
of
other
initiatives
actually
within
that
group,
but
that's
that
specification
is
one
of
them.
D
D
D
Just
anyway,
the
the
one
of
them
is
service,
mesh
patterns
and
so
sort
of
in
the
same
vein
around
what
we
were
just
talking
about
about
performance,
tuning
and
squeak.
You
know
getting
the
most
value
out
of
your
mesh
that
this
book
service
mesh
patterns
is
intended
to
help
people
get
the
most
out
of
their
mesh
in
in
other
ways,
not
in
in
terms
of
capabilities
or
to
provide
them
common
examples
of.
D
D
A
Have
people
who
are
looking
at
that
internally
as
well?
How
do
you,
how
do
you
describe
there's,
there's
you
steal
the
technology
and
our
site
maybe
does
a
decent
job
of
describing
that
technology,
but
those
patterns
for
implementation,
both
from
an
architectural
perspective
like
hey
I've
got.
You
know
three
clusters
in
different
regions,
what's
the
best
architecture
for
me
to
install
a
service
mesh
as
well
as
patterns
for
adoption,
and
I
and
are
you
looking
at
both
adoption
patterns,
as
well
as
deployment
patterns.
D
Yes
and
actually
dan,
your
some
of
your
work
in
this
area
has
been
well
or
I.
I've
well
noted
that
work
and
I
think
that
it's
been
well
noted
and
well
received
just
kind
of
the
analogy
that
you've
made
in
the
past
about
the
ability
to
to
consume
or
deploy
istio's
capabilities,
a
la
carte
and
do
so
in
a
piecemeal
fashion
and
sort
of
get
small
one
small
success
after
the
next
small
success
that
that'll
that
adoption
there
will
be
a
couple
of
adoption
patterns,
some
of
the
patterns.
D
That
said,
some
meshes
don't
do
as
much
as
others
or
or
they
have
a
different
architecture,
and
so
that'll
be
noted
in
there.
I
think
part
of
why
I'm
bringing
this
up
on
today's
call
is
because
there
wasn't
anything
else
to
talk
about
two
I
hadn't
really
articulated.
I
hadn't
told
you
mentioned
this
book
anywhere
else
and
I
guess
on
three.
It
certainly
occurs
to
me
that
there
are
any
number
of
folks
on
the
call
that
will
hopefully
benefit
from
the
book,
but
but
potentially
there's
a
number
of
folks
on
the
call.
A
Well,
cool,
I
think
I
think
that's
great.
I
think
that
anything
that
that
helps
people
along
along
that
journey
of
understanding
how
to
take
this
technology
and
apply
it
to
them
again
whether
it's
a
deployment,
pattern
and
understanding,
because
it's
complex
or
importantly,
those
adoption
patterns
and
what
I
did.
I
used
to
talk
about.
Istio
wallace,
cart.
A
I
often
now
refer
to
it
as
kind
of
incremental
I
should
I
should
do
a
talk
on
that
incremental
adoption
is
really
important
because
it
helps
I
was
talking
to
a
customer
just
today,
who's
really
interested
in
all
of
istio
and
is
thinking
about
it.
But
you
know
really.
In
the
end
we
said
you
know
you
have
a
critical
need
for
mtls.
You
have
a.
They
have
a
some
regulatory
reasons
that
they
need
to
put
mtls
in.
A
A
We've
done
one
thing
after
you've
finished
that
then
you
can
take
on
on
the
next
one,
and
so
I
think
that
that
those
adoption
patterns
are
really
critical
for
for
every
technology,
understanding
how
you're
going
to
expose
the
people
and
the
infrastructure
in
your
organization
to
this
technology.
So
that's
great
glad,
you're
glad
you're.
Looking
at
those
thank
you
lee
and
I
see-
sudeep
vatra
has
put
an
issue
in
here
about
running
two
control
planes.
Is
that
what
is
that?
What
we're
looking
at
here
sadiq?
A
A
So
yes,
I
one
of
the
at
some
point
zoom
introduced.
I
have
to
admit
people
into
the
meeting
and
my
apologies,
if
I'm
not
doing
that
actively
enough
thanks
for
the
ping,
if
anyone
else
keeps
an
eye
on
that
participant
window
and
sees
cease
people
in
the
waiting
room,
please
let
me
know
so.
I
see
in
the
agenda
sudhir
batra
put
in
an
issue
which
is
unable
to
create
two
control
planes
using
the
istio
operator
and
sadiq.
Are
you?
Are
you
in
the
meeting
now?
Can
you
hear.
A
F
F
Yes,
yes,
yes,
so
so
now,
so
there
are
two
use
cases
now.
I
suppose
it
is
possible
to
use
multiple
iop
cr
to
create
multiple
instances
of
let's
say
the
english
gateway.
That
should
be
possible,
and
I
see
that
we
have
documentation
for
that.
Also
in
history
like
how
to
create
multiple
gateway,
but
my
request
was,
to
you
know,
use
this
iop
cr
to
create
multiple
control
plane.
A
A
F
E
F
F
So
his
pr
is
addressing
that,
like
you
know,
we
can
have
two
or
more
instance
of
iop
and
they
can
be
in
different
name
space.
I
think
that's
what
is
being
addressed
in
his
pr
which
got
merged.
However,
that
still
does
not
address.
You
know
a
creation
of
two
completely
separate
instances
of
the
control
plane
and
then,
as
you
said,
like
it
has
to
be
two
different
service
mesh.
In
that
case,
we
have
to
define
like
which
control
plane
should
be
able
to.
F
E
So
you
would
need
to
do
that
each
control
plane
would
need
to
live
in
its
own
name
space
right.
Yes,
that's
right.
Yes,
so
it
sounds
like
martin's.
Pr
should
allow
running
two
completely
different
control
planes
in
two
separate
name
spaces.
I
I
think
if
I'm
reading
this
right,
have
you
been
able
to
pull
it
down
and
try
it
out,
or
I
know
it's
just
a
test
build.
F
F
But
I
can
give
it
a
try
again,
but
I
think
my
issue
should
not
be
closed
because
till
the
time
we
are
able
to
you
know
completely,
do
a
two
control
plane
and
be
able
to
define
you
know.
Each
control
plane
has
a
management
over
a
particular
data
plane.
Namespace
till
that
thing
is
completed.
I
think
we
should
not
close.
This
issue
is
my
humble
request.
A
A
It's
not
clear
to
me,
which
aspect
is
not
right.
The
other
thing
I
I'm
curious
about
backing
up
a
little
bit
is
what
the
use
case
is
for
running
two
different
meshes
in
the
cluster,
because
my
real
concern
is:
is
this
a
a
pattern
that
we
do
want
to
recommend
and
if
we
do
then,
then
have
have
we
as
a
project
thought
through
all
of
the
implications
and
everything
that
will
have
to
happen
in
order
to
make
sure
that
happens
like
will
they
be
upgraded
independently
with?
F
Yes,
yes,
surely
surely,
so
I'm
coming
from
a
telecommunication,
you
know
a
background,
and
here
what
happens
is
we
have
different
vendors
and
each
vendor
is
coming
with
its
own
implementation
of
the
istio
control
plane?
And
you
know
we,
for
example,
it
could
be
like
you
know,
ericsson
and
nokia
right.
So
there
are
two
different
instances
of
the
applications
which
are
provided
by
two
different
vendors
and
each
come
with
their
own
implementation
of
the
istio
control
plane
and
the
data
plane.
F
A
A
A
You
know
completely
completely
separate
service
meshes
within
a
cluster,
with
a
complete
different
set
of
configuration,
control,
plane,
data,
plane,
upgrade
life
cycle,
etc,
and-
and
my
guess
is
that,
as
as
astro
mart
marked
that
complete,
he
didn't
understand
from
the
from
the
context
of
the
issue,
what
the
what
the
full
request
was.
So
I
think
we
have
a
couple
things
we
can
do.
Yeah
and-
and-
and
like
I
said-
I
think
this
is.
This
is
the
kind
of
thing
that
it's
a
very
interesting
request.
A
It
might
be
that
the
toc
wants
to
weigh
in
on
whether
or
not
we
can
take
take
on
the
the
work
necessary
to
complete
that
use
case.
You
know
that
when,
when
we
start
talking
about
that,
not
only
are
there
features
that
need
to
be
implemented,
but
there
may
be
an
entire
suite
of
tests
that
need
to
be
created
that
don't
exist
in
order
to
validate
yes,
we
can.
A
We
can
completely
run
two
different
service
meshes
with
two
different
life
cycles
in
in
a
cluster,
so
that
might
be
the
kind
of
thing
that
we
do
need
to
bring
to
the
toc
to
get
validation
and
get
buy-in
and
their
lifecycle
team.
Our
testing
team,
a
bunch
of
different
teams,
are
going
to
have
to
kind
of
agree.
Yes,
we
can.
We
can
take
on
the
work
to
fully
support
the
use
case.
It's.
It
is
a
really
interesting
use
case.
I
agree.
A
E
F
Yes,
yes,
absolutely
absolutely
I
mean
yeah.
This
is
I
really
appreciate
the
time
taken
and
hopefully
now
it
is
clear
and
apologies
for
not
being
able
to
articulate
it
earlier
very
well,
but
this
is
really
you
know
like.
I
am
working
for
ericsson
for
atnt
project
and
this
thing
is
needed
for
for
many
vendors.
You
know
ericsson
have
formed
other
vendors
who
are
involved
for
a
tnt
environment,
so
this
is
like
coming
as
a
critical
as
a
critical
path
for
us.
You
know.
A
E
A
E
F
A
Okay,
okay,
so
I
I
I
will
put
you
on
the
agenda
right
now
for
that
july,
1st
meeting
just
to
make
sure
you've
got
a
spot.
If
something
comes
up
and
you
can't
make
it
I'm
sure
that
will
be
fine,
but
that
does
seem
like
the
right
course
of
action
here
is
take
it
to
the
environment's
working
group
meeting.
They
can
think
about
the
implications
for
the
product
overall
and
then
come
up
with
a
proposal
to
to
take
to
the
toc
so
cool.
Thank
you
very
much
for
bringing
that.
F
Yeah,
thank
you
so
much.
Thank
you
so
much
and
just
to
clarify
one
thing
further.
So
today
using
the
istio
operator
the
I
of
cr,
we
should
be
able
to
create
like
only
a
single
control
plane
and
but
it
will
allow
me
to
create
multiple
ingress
in
different
name
spaces
right
that
should
be
possible
today.
A
A
A
If
not,
then
thanks
for
your
time
today,
everyone,
as
always,
let's
see
our
next
meeting,
is
going
to
be
looks
like
july
9th
and
it
looks
like
I
won't
be
able
to
be
there,
but
I'm
sure
some
other
some
other
people
from
the
project
will.
So
thanks
for
your
time
today.
Hope
you
found
this
interesting
and
we
will
talk
to
you
in
a
couple
of
weeks.
Everyone
thanks.
A
A
If
people
did
not
see
it,
we
did
add
a
new
member
to
the
the
technical
oversight
committee.
Marriage
potter
from
aspen
mesh
who's
been
very
active
in
the
project
for
a
couple
of
years
now
was
running.
A
working
group
is
now
on
the
toc,
which
is
really
exciting,
so
he's
not
here
today,
but
congratulations
to
everybody
who
knows
new
age
all
right
thanks.
Everyone
see
you
in
a
couple
weeks.