►
From YouTube: CNCF SIG Network 2021-01-07
Description
CNCF SIG Network 2021-01-07
A
A
C
C
Maybe
actually
in
a
lot
of
respects,
not
the
meetings
that
the
open
source
meetings
that
we
hold
in
the
layer
five
community
are
attended
by
more
companies
than
usually
show
here,
but
yeah.
It's
it's
good.
It's
a
healthy
collection.
It's
a
great
venue
to
engage
with
others
from
different
companies
for
sure,
because.
D
D
C
F
C
C
C
All
right,
fair
enough,
very
good,
so
we're
about
six
minutes
past.
Let's
get
going.
This
is
the
cncf
sig
network
meeting,
well
the
first
one
of
2021.,
so
very
nice
to
have
mert
thanks
for
coming
rodolfo.
Thank
you
for
coming.
We've
got
a
topic
very,
very
relevant
to
the
two
of
you.
We
may
have
some
others
that
join
along
the
way.
C
The
given
that
this
is
the
first
time
for,
I
think
the
two
of
you
to
be
on
the
call,
it's
probably
good,
to
give
a
brief
lay
of
the
land
so
to
speak.
Actually,
one
point
of
clarification
that
ralpha
was
just
highlighting
is:
is
that
it's
commonly
confused
that
this
meeting
is
the
same,
that
there's
a
there's
one
sig
network
in
the
world,
and
and
and
that's
that
actually
turns
out
not
to
be
the
case.
C
C
We're
not
always
getting
into
the
weeds
here,
but
but,
but
I
think
today
we're
going
to
be
able
to
a
little
bit
so
ada,
so
so
there's
within
the
cncf
sig
network.
There
is
a
working
group
called
the
service
mesh
working
group.
The
service
mesh
working
group
has
a
few
work
streams,
and
so
you
can
see
the
the
charter
and
some
some
slides.
We
tend
to
give
an
update
at
each
of
the
last
cube
cons
on
things
that
we're
doing.
This
is
the
charter
of
the
group.
C
This
there
are
two
sort
of
two
service
mesh
working
groups
or
two
service
mesh
groups
within
the
cncf.
Currently,
so
one
is
this:
one
which
tends
to
be
tends
to
be
a
little
bit
more
of
vendors
represented
or
projects,
and
their
maintainers
tend
to
represent
here
and
come
here
and
try
to
help
advance
the
state
of
tech.
C
The
other
group
is
falls
within
the
end
users
area
of
the
cncf
and
so
that
that
service
mesh
that
tends
to
be
the
service
mesh
user
group,
which
tends
to
advance
the
state
of
the
art
in
terms
of
how
you
run
these
things
and
run
them.
Well,
so
there
are
a
number
of
initiatives
within
the
service
mesh
working
group.
We
want
where
some
of
these
some
of
you
might
be
familiar
with
already.
We've
we've
talked
about
about
a
lot
of
these.
C
Having
passed
that
private
due
diligence,
it
now
opens
up
to
public
requests
for
comment
so
whether
you're,
a
cncf
member
or
not,
if
you've
got
an
opinion
or
a
public
comment,
you
can
review
the
docs
and
chime
in
right
now.
There's
a
discussion
going
on
with
respect
to
the
name
of
the
project
ambassador
and
it
it's
it's
to
be
renamed
and
right
now.
The
proposal
for
that
name
is
ingress
controller
for
envoy
proxy,
or,
I
guess
more
sort
of
specifically
that
it
would
be
an
acronym
ic,
4e.
C
C
C
It
is,
I
think,
you
know,
probably
a
tentative
name
get
nighthawk.
I
think
think
that
makes
sense
for
all
those
that
are
involved.
No
one
has
voted
otherwise
or
come
up
with
anything
smarter,
so
the
project
itself
is
well.
It
focuses
on
nighthawk
and
making
nighthawk
accessible
to
the
rest
of
us.
What
is
nighthawk,
you
say:
well,
it's
a
load
generator
written
in
c
plus
plus
it's
a
sub
project.
If
you
will
of
envoy.
C
C
I
think,
to
a
number
of
parties,
including
some
folks
that
including
engineers
at
various
tech
companies
that
we're
all
familiar
with
also
potentially
including
the
the
istio
project
that
is
both,
I
think
think
sort
of
is
dipping
a
toe
into
you,
the
use
of
nighthawk.
So
so
this
is
a
project
that's
doing
well.
C
It
is
not
easy
to
get
to,
though,
necessarily
like
there's
a
there's,
a
build
of
nighthawk
in
order
to
build
nighthawk
you've
got
to
have
oh
you've
got
to
have
a
copy
of
envoy
and
then
use
a
lot
of
envoys
tool
chain
use
envoy.
In
your
tool
chain
so
to
speak,
to
compile
and
build
and
make
a
release
of
a
nighthawk,
I
think
the
the
there's
a
single
release
of
nighthawk.
Today
it
comes
in
a
form
of
a
container.
C
C
G
Support
you
know,
arm
64
is
getting
some
popularity
lately,
so
maybe
at
some
point
that
would
be
kind
of
interesting.
I
know
that
employees
building
also
against
various
power,
pc,
etc,
etc.
So
I
mean
like
the
beauty
infrastructure,
at
least
for
envoy,
is
there
I
mean,
maybe
that's
something
that
can
be
of
interest
yeah.
H
C
G
I'm
not
sure
if
they
have
mips
or
whatever
I
mean
they
already
have
arm
64
power
pc,
and
if
this
project
is
close
to
the
beauty
infrastructure
of
envoy,
I
guess
that
something
can
be
reused
from
their
built
infrastructure
to
also
bring
it
to
this
other.
I
don't
know
how
how
interested
interesting
it
would
be
for
someone
to
run
it
on
other
architectures,
but
at
least
arm
64
probably
would
be
of
interest
to
someone
at
some
point.
G
C
C
C
Rather,
it's
been
really
challenging.
It's
taken
it's
taken
a
few
months
to
get
there
so
yeah,
so
some
of
the
goals
of
the
project
are
yep.
It's
really,
you
know
about.
It's
focused
initially
on
get
nighthawk
proper
sort
of
on
distribution
of
the
project,
making
it
easier
to
get
to
what
we
were
just
talking
about,
but
part
of
the
justification
for
why
to
spend
the
time
to
do.
C
This
specification
is
trying
to
help
provide
a
common
format
for
doing
so
and
so
helping
bring
nighthawk
and
that
closer
together
is
is
sort
of
part
of
the
thinking
here.
Nighthawk
itself
is
set
to
be
support,
horizontal
scaling
or
to
say
it
differently,
most
the
load
generators.
Today,
if
you
wanted
to
run
multiple
instances
of
the
load
generators,
you
would.
C
Nighthawk
is
trying
to
up
level
a
bit
of
its
intelligence
to
be
cognizant
of
different
instances
of
itself,
so
that
that
makes
it
more
interesting.
It
also
has
a
little
bit
of
a
plug-in
model
around
the
use
or
the
around.
This
thing
that's
referred
to
as
the
adaptive
load
controller,
which
is
to
say
that
nighthawk
has
the
ability
to
run
run
a
load
run
a
test.
C
Do
some
statistical
analysis
of
the
per
you
know
the
characteristics
of
that
test
that
the
performance
under
that
load
and
then
through
this
adaptive
load
controller,
which
you
could
plug
in
different
algorithms
or
different
decision
points
that
then
nighthawk,
would
automatically
kick
off.
Another
test
and
sort
of
and
then
interpret
you
know
generate
some
results,
interpret
the
results
and
then
kick
off
another
test
and
sort
of
in
an
adaptive
way,
control
the
load
so
that
you
could
maybe
run
some
optimization
algorithms
or
try
to
answer
questions
about
like
hey.
C
What's
your
what's
the
optimal
load
for
this
environment
for
this
app,
so
these
two
things
are
forthcoming
in
nighthawk
and
make
it
even
more
interesting
and
part
of
the
justification
for
wrapping
some
support
around
it,
so
yeah
so
anyway,
these
are
high
level
goals.
Some
other
reasons
are
that
hey,
there's
a
service
mesh
patterns
book
that
we've
just
talked
about
a
little
bit
before,
with
about
30
patterns
going
in
there
about
service
meshes
and
their
use.
C
C
Envoy
itself
is
available
in
certain
formats,
but
there
was
an
a
similar
initiative
created
to
help
build
a
cli
get
envoy
to
and
different
packages
to
help
get
onboard
into
people's
hands,
and
so
I
mention
it
just
as
a
sort
of
a
core
or
a
parallel
analogy
to
get
nighthawk
and
the
fact
that
the
the
group
isn't
looking
to
create
a
get
nighthawk
cli
rather
to
capitalize
on
the
management
that's
already
provided
by
mastery
ctl
of
this
cli.
So
so
we
would
be
enhancing
this
cli.
C
There
are
and
then
there's
a
collection
of
folks
who
are
coming
to
bear
so
on
both
the
continuous
integration
work.
Rudolph
rudolfo,
that's
sort
of
eyeballing
you!
If
you
can't,
if
you
don't
feel
that
and
there's
some
other
contributors,
who've
raised
their
hand
to
create
a
small
website
static
website
to
help
promote
the
project
a
bit
more
and
today.
If
you
want
to
learn
about
envoy
you
cruise
on
out
to
envoy
proxy
slash,
you
know
nighthawk
and
it's
unfortunate.
C
C
Go
into
the
docs
there.
If
you
do
a
search
for
nighthawk,
I
think
you'll,
I
think
you'll
find
one
one
mention.
I
believe
what
a
best
part,
and
so
my
point
is
outside
of
just
trying
to
build
the
bits
and
get
the
bits
into
people's
hands.
It'll
also
really
help
facilitate
documentation
around
the
capabilities
of
the
thing
and
how
it
can
be
used.
C
This
is
probably
the
first
time
you're
really
getting
introduced
to
the
project
itself
and
to
kind
of
what
the
purpose
and
why
clarifying
questions.
I
know
there's
a
lot
of
a
lot
of
things
you
get
yet
to
be
able
to
think
about,
but
just
does
the
the
logic
behind
the
justification
kind
of
the
logic
behind.
Why?
Why
we're
spending
time
here,
some
of
the
things
that
we're
looking
to
do
in
scope
to
those
make
sense.
C
I
don't
know
if
this
is
is
as
interesting,
but
so
nighthawk
like
envoy,
written
in
c
plus
well,
the
lingua
franca
of
the
rest
of
the
cloud
native
world
could
be
argued
to
be
golang,
and
so
there's
a
little
bit
of
a
lack
of
like
part
of
what
we're
trying
to
achieve
here
is
maybe
bridging
between
the
two
that's
actually
described
in
one
of
these
line
items
yeah
here
and
so
it's
a
bit
tentative
like.
Maybe
there
needs
to
be
a
different
repo.
C
C
D
C
It's
it's
a
it's
a
smaller
project
in
nature,
like
hey,
no
con
no
contributor
is
ever
turned
away.
Yeah
it
like
in
terms
of
the
scope
of
what
the
project
is
trying
to
achieve.
There's
kind
of
there's
some
big
big
blocks.
If
you
will
there's
or
the
bigger
blocks,
are
a
project
site
which
you
could
depending
upon
how
pretty
you
want
to
make
it
look
it
it's.
C
C
There
is
some
court,
you
know
coordination
on
like
a
bunch
of
things
to
kind
of
define
like
what
does
it
mean
if
you
have
nightly
builds,
how
are
we
doing
facilitating
sort
of
integra?
How
are
we
uplifting
that
project
and
doing
maybe
some
integration
tests
with
it
between
meshary
and
nighthawk,
because
the
potentially
there
could
be
breakage
between
those
two
components.
C
C
He
pratia
banerjee.
He
is
an
open
source
contributor
and
he's
hot
hot
to
trot
on
the
project
site.
So
he's
he's
kind
of
been
putting
effort
into
helping
figure
out.
What
to
do
here
he's
an
open
source
contributor.
That's
helped.
He
helped
create
the
s.
P
spec
site
he's
a
maintainer
of
that
site,
so
he's
coming
to
bear
here
as
well.
D
C
Yeah
yep,
we
need
to
I
like
this,
for
this
particular
forum
isn't
the
best-
and
this
has
been
my
struggle,
nikolai.
I
don't
know
if
you've
sort
of
watched
me
struggle
for
the
last
few
months
about
like
how
deep
to
get
into
some
of
the
service
mesh
working
group
projects
here,
because
I
feel
like
we're
commandeering
sig
network's
time
with
it,
but
each
time
that
I've
asked
I've
been
told,
like
hey,
just
continue
to
use
this
time
up
until
there's
some
other
topics
and
so
okay
cool
but
yeah.
C
But
so
we
need
to
do.
You
know
much
more
like
in
the
weeds
things
about
some
of
what
nikolai
was
highlighting.
Actually,
like
hey,
you
know,
let's
get
more
familiar
with
the
build
tool
chain
of
envoy
itself
on
the
nighthawk
as
a
project,
it
doesn't
have
regular
meetings
to
to
sort
of
lean
into
and
attend,
and
so
there
are
a
couple
of
other
projects,
a
few
other
people
that
I
thought
would
be
on.
Today's
call
that
are
related
to
this.
C
C
Yeah,
so
so
mert
I
don't
have.
This
is
like
still
it's
kind
of
a
struggle
for
me
because
we
like
I've
been
I've,
helped
create
and
nikolai
might
have
been
there
as
well.
But
like
the
cloud
events
like
as
a
project
more
or
less
born
of
the
cncf,
I
mean
it
was,
and
it
was
kind
of
built
within
as
a
spec
sort
of
built
within
the
server
this
working
group.
C
So
we
had
precedent
for
creating
things
you
know
sort
of
directly
within
the
cncf
rather
than
externally
and
bringing
them
to,
but
I
think
I
I
think
we
would
do
well
to
find
probably
a
separate
meeting
from
this
to
go
through
all
the
particulars
of
to
to
advance
smp
to
advanced
nighthawk
part
of
like
advancing
on
get
nighthawk.
It
will
help
advance
the
service
mesh
performance
spec.
The
service
performance
spec
is
intended
to
be
up
for
review
under
sandbox
and
so
that'll
be
another
meeting
to
help
advance
this
in.
D
Sorry
I
asked
that
question
the
workshop
or
hackathon,
because
c
plus
plus
is
building
itself
daily
and
so
like
11
and
17
the
standards.
So
I
don't
know
which
standards
you
follow.
I
didn't
read
the
thought
code
source
code
and
I
guess
it
will
be
fine
to
secure
to
fix
the
bugs
within
the
source
code
with
new
c
plus
plus
standards
and
the
own
projects
code
standards
for.
C
That's
a
great
question:
if
otto
or
jacob
were
here
the
two
you
know
a
couple
of
the
nighthawk
maintainers,
they
could
speak
to
that
more
directly.
Also,
if
matt
klein
were
here
the
one
of
the
maintainers
on
envoy,
he
could
speak
to
that.
I
I
don't
know
much
of
what
we're
trying
to
accomplish
in
get
nighthawk
is
more
packaging,
docks
and
distribution,
much
more
of
that
than
it
is
slinging
a
bunch
of
c
plus
plus.
C
G
I
was
thinking
I
thought
that
this
is
going
slightly
off
of
the
boundaries
that
you
you
kind
of
draw
here
around.
All
the
interesting
gets
get
tight
hulk
here,
but
how?
How
would
this
tool
be
useful?
In
I
don't
know
staging
can
even
production
environments,
they
is
there,
like,
I
don't
know,
set
of
use
cases,
or
I
don't
know,
example
configuration
so
I
I
don't
know
this
could
be,
could
be
something
interesting
to
this
project,
but
I
mean
because
okay
testing
in
all
up
is
great
good.
G
C
I
think
I
think
it's
a
great
question.
Let
me
let
me
see
if
I
can
answer
it
with
something
of
an
example.
I
think
very
much
so
very
much
so
like
integration
into
ci
pipelines
is
something
that
it's
in
part.
Why
we're?
Not?
That
project
is
not
trying
to
write
a
new
get
nighthawk
cli,
because
because
it's
already
a
significant
investment
in
those
use
cases
for
measuring
ctl
to
do
so.
The
cli
here
measuring
ctl
perf
to
run
a
performance
test.
You
could
then
pass
in
your
passing
you
here.
C
Because
to
your
point
like
hey
the
yay
or
nay,
you
don't
necessarily
need
a
human
to
look
at
a
colored
chart
to
determine
what
to
do.
You
know
you
might
just
want
to
know
hey.
You
know
the
baseline
for
our
project
either.
Maybe
our
the
workload
that
we're
the
application
we're
creating
or
for
the
mesh
that
we're
creating.
C
It's
been
looking
like
this.
It
looks
like
this.
We
can
describe
that
in
a
standard
way
using
service
mesh
performance.
We
can
have
some
repeatable
tooling
that
we
can
just
build
into
the
ci
system
to
run
that
and
then,
in
accordance
with
the
baseline,
that
we
have
now.
You
know
different
projects
like
kuma
can
figure
out
if
they're
having
regression
issues
or
if
they're
advancing
in
performance.
C
C
C
So
with
that,
and
actually
nikolai
just,
I
think
quickly
to
that
point-
you
know
how
we
we've
discussed
the
smi
conformance
project
a
bit
to
to
run
some
some
conformance
tests.
The
the
same
thing
is
true
there
as
well
that
you
would
be
able
to
do
the
same
program
at
run,
measuring
ctl,
smi,
conformance
tests
using
mastery,
ctl
or
using
mescheri's
rest
api,
either
way
in
in
the
ci
process
that
you
don't
have
to
go
run
mesher
a
user
doesn't
have
to
go
run
master.