►
From YouTube: Istio Community Meeting 4.4.19
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
Hopefully
the
month
is
off
to
a
good
start,
for
you
all
can't
believe
it's
already
April,
it's
kind
of
crazy
as
I
mentioned
agenda,
is
currently
clear.
So
please
add
some
stuff
to
it.
Does
anybody
have
any
announcements
or
anything
they
want
to
share
before
we
jump
into
questions
and
things
like
that.
A
B
B
The
topics
that
we'll
be
going
through
and
I'll
be
I'm,
mentioning
because
I'll
be
facilitating
to
our
session,
presenting
a
few
things
and
want
to
call
it
out
here
to
the
extent
that
there
are
people
either
looking
to
augment
that
agenda
as
it's
being
formed
or
people
that
might
be
there
to
each
other.
So.
A
A
D
A
D
A
E
You
can
ever
find
I
think
if
you
never
trouble
finding
interesting
on.
There
is
not
a
specific
Meetup
and
right
now
Archie,
although
as
always
about
three
days
before
a
competition,
would
that
be
a
really
good
idea.
Let
me
put
my
head
together
with
some
smarter
heads
than
mine
and
see
if
there's
something
we
can
do
one
of
the
nights.
It's
a
really
good
idea.
As
always,
there
are
I
just
put
together
a
slide,
something
like
ten
different
sessions
for
people
who
are
interested.
E
You
will
start
to
see
some
glimpses
for
those
people
who
are
interested
in
in
Google's
kind
of
strategy
around
around
commercialization,
but
for
just
general
sto
talk
I
think
there
will
be
plenty
of
people
around
who
would
be
interested
in
in
that
as
well.
So
let's
put
our
heads
together
and
we
will.
E
E
A
A
A
A
E
F
F
G
E
G
E
G
A
A
B
B
There
were
more
I
think
about
250
registrants
guessing
about
180,
you
showed
so
what's
the
right
word,
not
pinochle
and
not
epitome,
but
he's
you
know,
folks
that
were
very
clearly
interested
in
that
sympathetic.
People
have
both
vested
interests,
those
that
have
projects
or
commercial
operators
around
service
missions
and
service
mentions,
most
notably
being
a
CEO
and
then
envoy
and
it's
you
know,
and
it's
very
used.
Those
were
certainly
two
that
were
a
focus
at
the
conference.
B
B
There
were
some
things
around
pretty
interesting
things
around
the
next-generation
access
control.
So
for
all
of
us
who
are
familiar
with
role
based
access
control
and
some
of
us
who
are
familiar
with
attribute
based
access
control
are
back
a
back.
There's
a
proposal
out
there
where
there's
something
collaboration
being
done
with
NIST
for
energy
in
G
ACK
next
generation
accidents
control,
all
those
when
the
videos
are
opposed
to
be
that's
something
to
look
into
your
stable.
It's
kind
of
interesting,
some
other
proposals
and
thoughts
around
a
presentation
to
Mary
Peterson
of
the
ona.
B
With
respect
to
really
this
concept
that,
like
merging
service
provider,
great
network
use
cases
layers,
one
through
four
focused
things
with
much
of
where
I
considered
the
rest
of
us
or
those
that
are
well
represented
on.
This
call
can't
focus
on
you
know
latest
fire
on
fire
at
five
to
seven,
so
that
was
interesting
because
certainly
there's
kind
of
a
service
provider
grade
their
service
provider
world
as
be
enough
sand
use
different
virtualization
area.
So
there
was
a
presentation
by
the
network
service
smash
group,
which
is
a
project.
B
That's
been
proposed
for
review
as
an
incubation
project
yeah
having
spoken
with
Fredrik
they're,
one
of
the
founders
of
that
project
I
finally
got
an
answer
is
the
light
of
day
why
they
named
it
network
service
mesh
so
generically,
and
that
was
in
part
something
of
a
devious
move
to
to
have
it
be
the
canonical
reference
for
a
network
service.
Mention
they've
even
got
a
chuckle
out
of
that,
but
I
get
a
drone
on
about
the
I
ended
up
resenting
fishery,
which
is
an
open
source
project
that
I
mentioned
during
our
last
community
called
yeah.
B
E
The
Li
Li,
do
you
mind
answering
some
questions
about
the
event
sure
in
your
since,
since
you
seem
to
be
a
pretty
accurate
estimator
of
numbers,
what
percentage
of
the
people
do
you
think?
Were
there
either
behind
an
open-source
project
or
a
commercial
project
that
helps
to
implement
service,
miss
at
service
mesh
and
what
percentage
were
either
using
or
actively
considering
using
service
mesh
for
their?
You
know.
B
A
E
E
B
B
You
know
you
know
meet,
meaning
that
it
meaning
mean.
How
do
you
want
more
things
on
the
mesh
that
are
out
on
the
edge
there's?
A
lot
of
you
know:
hey
there's
a
cell
tower
here
we
can
come,
allocate
some
compute
next
to
it.
Do
some
intelligent
things
there.
It
would
be
even
better
if
that
was
on
a
match
and
that
wasn't
necessary
in
context
of
grantees
and
then
network
service.
H
E
Yeah
I
think
it's
one
of
the
interesting
areas
that
that
we
will
see
as
as
enterprises,
look
at
adopting
this
I
think
we're
gonna,
see
two
strategies
and
I
and
I
know
of
enterprises
right
now,
employing
both
some
people
who
are
using
this
as
a
reason
to
move
the
containers
and
and
then
thinking
about
service
matching
things,
but
yet
others
who
are
saying?
No.
No,
no.
No.
We
can't
move
everything
right
now,
but
we
want
serve
as
much
right
now.
Yeah.
H
H
G
B
Yeah
yeah
I
couldn't
I,
didn't
hear
one
bubble
out
more
than
the
next
thing,
maybe
because
again
I
was
out
on
the
hall,
but
for
sure
of
the
conversations
I've
had
over
the
last
18
months,
it's
been
taken.
You
know
that
the
ability
to
take
the
blinders
off
and
have
observer
billion
see
what's
going
on
and
it
was
the
observability
that
was
the
most
significant
attractant
I.
J
B
And
that
may
have
been
another
from
my
part
as
we've
looked
to
engage
with
petty
Court,
along
having
a
build
up
and
actor
for
mystery,
that
open
source
project
they've
been
off
mostly
off
and
then
recently
very
on
in
terms
of
wanting
to
contribute
there
and
my
sense
of
them
being
off
was
to
your
point
about
that.
They
weren't
hearing
that
a
lot
their
customers
weren't
banging
on
the
door.
Yeah.
K
D
Have
a
large
customer
in
Canada
that
is
waiting
for
one
that
one,
because
they
have
a
pile
of
fishes.
That
is
supposed
to
be
fixing
one
that
one
but
I
just
had
a
cold
and
when
they
said
like,
we
just
put
a
little
bit
in
the
back,
because
they
couldn't
go
in
production
with
the
previous.
Because
there
was
a
lot
of
issues
they
had
at.
E
Scale,
certainly,
that's
not
uncommon.
I've
seen
some
I
don't
know
if
everyone's
seen
the
the
new
performance
page,
but
the
the
performance
page
was
updated
with
with
new
numbers
and
there's
it's
much
improved,
so
I'm
interested
to
see.
If
that's,
where
we're
working
closely
with
a
couple
of
couple
of
companies
who
were
implementing
pure
open
source
and
I'm
working,
and
we
can
talk
about
them
publicly
and
not
too
long
because
they're,
it's
it's
doing
much
better
at
a
large
scale,
where
a
large
scale
is
thousands
of
services,
some
of
which
have
thousands
of
pods.
J
J
G
J
G
J
E
I
I
have
a
customer,
actually
they've
publicly
talked
about
their
their
usage,
the
they
hadn't
give
me
permission
each
other
their
numbers,
and
they
saw
vastly
vastly
almost
always
that
I
much
vastly
vastly
reduced
CPU.
That
was
their
big
concern,
was
CPU
and
they
said
actually
I
do
see
things
like
that.
G
E
So
if
you
are
running
on
that
one-
and
this
is
any
sort
of
concern
at
all-
is
a
highly
highly
recommended
upgrade
people
who
are
doing
it
are
seeing
great
great
improvement
and
what
was
one
of
our
I
think
our
biggest
areas
of
concern
that
we
gave
people
the
increased
namespace
isolation
has
some
good
side
benefits
to.
Not
only
is
it
better
for
performance,
but
it's
also
better
for
isolating
your
teams
from
each
other
Minh
pattern
that
we're
seeing
in
implementations
is
allocating
namespaces
protein
and
the
namespace
isolation.
G
C
F
E
G
Well,
we're
not
we're
not
running
it
as
a
requirement
at
the
moment,
I'm,
not
surprised
that
it's
coming
up
and
I
mean,
especially
now
with
coop
114,
we're
hearing
more
and
more
requests
for
Windows
support,
given
that
it
is
now
official,
so
everyone's
going,
ok,
when's
everybody
gonna
support
Windows.
Now,
ultimately,
this
is
gonna.
This
is
something
we
have
to
take
through
the
TOC
and
and
work
through
the
requirements.
It
will
be
a
multi
argh
build,
so
our
entire
build
process
will
have
to
change.
G
The
testing
process
will
have
to
change
any
components
that
run
on
those
nodes
will
have
to
support
Windows,
which
is
predominantly
Envoy,
which
shouldn't
be
a
problem
right.
Hopefully
we
should
run
on
Windows
just
behind,
but
that's
that's
something
that
we
would
have
to
investigate
and
look
at
and
take
through
the
TOC.
C
E
C
G
F
C
M
C
G
And
that's
the
pet
that
I
would
expect
that
we
would
take
as
we
if
we
would
incrementally
roll
this
out,
but
the
control
plane.
We
would
not
try
it
because
that's
a
much
larger
chunk
that
we
would
have
to
buy
it
off
and
support
it's
easier,
not
easy,
but
easier
for
us
to
support
the
data
playing
yeah.
A
K
Yeah
it's
so
it's
kind
of
like
a
shameless
plug,
but
I
originally
wrote
it
in
response
to
universe
previous
question
about
case
in
it
so
I've
been
working
on
a
brand
new
terraform
plugin
for
kubernetes,
which
is
aside
from
the
official
one.
My
version
in
particular
deals
with
supporting
all
the
kubernetes
resources,
as
opposed
to
the
very
limited
set.
The
official
one
supports
and
the
most
challenging
part
is
the
sources
based
on
co,
DS
and
so
SEO
has
lots
of
co,
DS
and
I
it.
K
It's
spent
the
last
couple
months
using
SEO
as
sort
of
my
like
my
proving
proving
ground
and
I
finally
got
it
working
last
week,
so
I
just
thought
that
did
I,
let
everyone
know,
and
so
you
using
using
this
work,
I'm
able
to
first
of
all
take
the
is,
do
demo
llamo
and
which
is
like
a
plain
llamo
generated
from
ahem
shot
and
I'm
able
to
automatically
convert
converted
to
tell
form
format
and
then
deploy
it
like
a
regular
telephone
module
with
pretty
easy
right.
K
K
Yeah
I
mean
I,
mean
I
sort
of
most
of
my
communications
via
code
on
on
get
up,
so
any
anything
feel
free
to
log
an
issue
or
hit
me
up
directly.
If
you
want
to
have
a
one-on-one
demo
on
how
it
works-
and
this
is
a
comment
on
one
one
one.
It
works
a
lot
better
than
when
I
tried
it
previously
and
I'm,
not
a
heavy
user,
but
but
yeah
I'm
looking
forward
to
ultimately
integrating
its
deal
with
K
native.
A
Awesome
I
see
if
there's
some
links
in
the
dark
for
Canadian
events
that
are
coming
up,
which
is
awesome
and
then
it
looks
like
Lin
is
doing
a
deep
dive
webinar.
So
there's
a
link
there
for
folks
to
register
for
that
and
then
there's
a
link
to
the
new
sto
performance
page
as
well.
So
lots
of
good
things
to
check
out
in
the
dock
Julian
had
a
question
about
load,
shedding
functionality
to
mixer
workloads.
A
E
I
E
I
So,
at
a
very
high
level,
which
is
the
only
level
I'm
capable
of
operating
at
the
loadshedding
firm
mixer
allows
us
to
watch
the
latency
of
incoming
mixer
reports
as
well
as
policy
requests
and
if
they
get
over
an
established
threshold
which,
if
I
remember
by
default,
is
100,
milliseconds
mixer
will
simply
drop
those
requests.
That
does
mean
that
for
mixer
reports,
telemetry
trace
data
etc
will
not
be
collected.
I
It
also
means
that
for
policy
you
will
either
fail
open
or
fail
close
depending
on
how
you've
configured
your
policy
engine,
but
it
has
the
advantage
that
you
don't
have
a
whole
bunch
of
envoys
that
are
waiting
for
an
excessive
amount
of
time.
We
saw
certain
scenarios
where
mixer
gets
overwhelmed,
where
envoy
would
be
waiting
on
an
individual
request
for
more
than
five
seconds,
which
has
the
ability
to
impact
your
production
traffic.
So
it's
a
way
of
isolating
your
production
traffic
from
being
affected
by
a
mixer
failure.
I
It
is
on
by
default
in
1:1
and
I,
like
I,
said,
I
believe
the
threshold
is
100
milliseconds.
You
can
change
that
within
the
mixer
command
line
to
being
off
by
default
or
by
to
only
list
in
the
log
when
it
would
have
load
shed.
If
you'd
like
to
turn
it
on
provisionally
kind
of
to
test
out
what
it
would
look
like,
without
actually
affecting
your
workloads.
A
I
Okay
and
so
I
do
have
a
provisional
answer
for
load
shedding
on
telemetry
versus
policy.
We
do
run
two
different
instances
of
mixer
one
for
telemetry
and
one
for
policy.
You
can
change
the
load
shedding
mode
on
one
and
not
the
other
off
the
top
of
my
head.
I,
don't
know
if
we
have
like
helm
support
for
that
if
you
could
deploy
it
by
default,
just
using
a
helm
value
or
if
you
mean
to
manually
edit
the
deployments
to
achieve.
E
A
E
Will
put
in
one
more
plug
for
the
for
the
updated
docs
site,
I'll
think
right
rigs
for
a
lot
of
work
on
the
dock
site.
There
is
a
I
think
it's
in
the
about
action
in
the
community.
There's
a
a
page
for
people
who
are
using
a
steel
and
pr's
are
welcome
there.
So,
if
you're
using
this
geo
art,
if
you
have
customers
who
are
using
this,
do
encourage
people
there
welcome
to
throw
their
logo
up
there.
F
D
A
H
A
B
A
G
A
I,
don't
know
the
service
message
a
this
year
like
there
was
last
year,
but
of
course,
as
we
all
know,
with
conference,
those
things
come
together
at
the
last
minute,
so
that
could
always
change.
But
if
nothing
else
I
know
a
lot
of
us
will
be
there.
So
that's
another
good
opportunity
to
kind
of
meet
up
and
chat.
We
certainly
had
fun
last
year,
so
hopefully
we
can
all
get
together
again
this
year.