►
From YouTube: CNCF Service Mesh Interface Project 2021-04-28
Description
CNCF Service Mesh Interface Project 2021-04-28
B
Awesome
teamwork.
We
got
there
eventually
so
welcome
to
the
smi
community
meeting
april
28th,
and
this
is
the
the
usual
meeting
where
we
kind
of
discuss
more.
So
the
specification,
not
the
the
bi-weekly
meeting,
where
we
are
discussing
things
like
service
mesh
federation.
B
So
we've
got
only
a
couple
of
discussion
items:
an
smi
controller
update
from
myself
a
request
for
a
bit
of
bike
shedding
on
a
pr
from
from
bridget
chat
about
some
kubecon
eu,
which
is
is
coming
up
next
week,
of
course,
which
smi
being
a
project.
Is
this
playing
a
big
part
in
and
also
lee
chatting
about,
aligning
smi
smp
and
meshary?
So
pretty
good
agenda
there?
Do
we
do
we
have
any
new
folks
on
the
call
who
would
like
to
introduce
themselves?
B
So
the
approach
that
we're
going
to
take
in
the
the
short
term
is
that
the
the
existing
go
sdk,
which
is
used
by
pretty
much
all
of
the
existing
adapters,
is
going
to
be
maintained
alongside
the
the
new
controller
sdk
and
we're
going
to
figure
out
an
elegant
way
that
those
objects
can
can
eventually
share
the
same
space
without
any
any
duplication.
But
we
might
need
a
little
bit
of
you
know
just
some
some
clever
github
actions
or
something
in
the
in
the
short
term.
B
But
but
the
intention
is
that
both
the
new
controller
sdk
and
the
existing
go
sdk.
The
objects
which
are
defined
in
there
will
will
co
will
coexist
and
will
be
interchangeable,
which
will
make
the
upgrade
process
smoother.
B
It's
pretty
much
michelle
and
I
have
been
plugging
away
through
that
and
next
up
we
have
a
request
from
bridget
and
I'm
just
opening
opening
this
up
so
bridget
has,
I
don't
even
try
doing
the
screen
sharing.
So
I'm
just
going
to
copy
and
paste
the
the
pr.
B
Into
the
chat
there,
but
bridget
has
raised
a
sort
of
a
request
for
for
comments,
which
is
clarification
of
scope
for
service
mesh
interface
and
that
clarification
of
scope
is
coming
from
the
multi-cluster
working
group,
and
it's
intended
that
we
start
think
about
smi
as
a
as
a
specification
which
encompasses
other
workloads
than
just
kubernetes,
specifically
virtual
machines,
bare
metal
and-
and
things
like
that.
B
Now
that
that's
not
necessarily
going
to
to
change
the
the
direction
but
but
it
would
allow
for
elements
of
the
specification
which
are
specific
to
to
those
platforms.
I
think
this
is
a
really
good
thing
personally,
because
it
it.
It
enables
the
work
that
we're
trying
to
kind
of
get
through
around
multi-cluster
and
federation,
and
it
also
brings
the
the
capability
that
smi
is
wider
spread
than
than
kubernetes.
So
we
can
support
all
of
those
other
workloads
that
folks
are
using.
So
please
take
a
look
at
that.
B
A
Can
we
maybe
time
box
that
a
little
bit
because
the
current
point
in
time
it
sounds
like
you
know,
as
we
say
in
the
fullness
of
time,
which
is
kind
of
like
whatever
can
we
maybe
say
you
know
some
expectations.
B
I'm
bridgette
doesn't
mention
what
the
expectation
is
on
there
now
generally,
the
way
that
we
we
we
kind
of
run
run.
Those
things
is
that
if
it
gets
a
a
sort
of
a
quorum
of
votes
from
core
maintainers,
then
then
generally
those
things
sort
of
go
in.
In
my
personal
opinion,
I
think
this
is
a
fairly
sort
of
inoffensive
suggestion
and
something
that's
probably
going
to
happen.
But
let's
say
by
the
next
the
next
meeting
that
we
can.
We
can
merge
that
in
in
seven
days
time.
C
If
that
sounds
sensible,
yeah,
which
would
be
before
well
so
michael
we're
not
having
that
multi-cluster
conversation
next
week,.
A
Right
next
week
is
it's
kubecon
you
correct
correct,
so
yes,
so
that
would
actually
be
nice
if
we
could
add
cubecon
next
wednesday.
I
understand
that's
where
we
have
our
little
online
session,
maybe
already
report
on
that.
If
someone
asks
like,
is
that
korean
specific?
Well,
we
recently
decided
to
broaden
this
code
yeah.
B
Yeah,
just
because
we
defined
the
specification
in
yaml
doesn't
doesn't
limit
the
the
use
cases.
You
could
write
a
controller
for
anything
which,
which
I
think
is
really
really
really
cool.
To
be
to
be
honest,
and-
and
that's
all
from
from
these
two
points
lee
next
week
is-
is
kubecon
I'll
hand
it
over
to
you
yeah.
He.
C
A
Oh
yeah,
like
sorry
I
need
to
yes,
let
me
allow
multiple
participants.
Try
again.
C
There,
it
is
very
good
so
on
this
topic
next
week,
if
you,
if
you
haven't
seen
a
tweet
about
this
next
week's
kubecon
and
like
like
last
time,
I
think
it
was
only
last
time
we
ended
up.
We
have
a
smi,
it
has
project
office
hours
for
an
hour
actually
which,
on
the
surface
of
it,
feels
like
a
long
time.
C
C
C
Yes,
oh
very
good,
so
there's
a
there's
there's
a
session,
that's
reserved
for
that
hour
I'll.
If
someone
else
doesn't
I'll
drop.
The
link
into
that
session
here,
question
for
those,
so
so
I
anticipate
michael
will
is
gonna.
C
Do
his
best
to
be
there
so
he's
gonna,
be
you
know
on
the
list
so
so
next
next
week,
when
we,
when
we
host
these
office
hours,
we'll
end
up
engaging
with
people
talking
about
the
questions
they
have,
we'll
also
have
some
slides
in
the
background
to
use
as
fodder
for
a
conversation
and
to
michael's
point,
one
of
those
slides
would
ideally
and
hopefully
be
that
there's
been
the
slightest
verbiage
tweakage
in
the
charter
that
says
kubernetes
and
beyond,
and
so
what
I'm
looking
for
is
confirmation
for
those
that
think
that
they'll
be
there
to
you
know
to
sit
on
the
panel
right.
A
I
would
say
the
only
challenging
thing
or
if
it's
not
entirely
clear,
because
I
was
just
saying
you
know
central-
this
is
kubecon
eu
right,
so
this
would
be
7
a.m
for
u.s
west.
So
if
you're
on
u.s
west,
you
either
get
up
very
early
or
you
skip
that
east.
If
you're,
you
know
somewhere
in
u.s
east,
not
a
problem,
but
for
for
west,
it's
really
early.
D
Six
a.m:
okay,
good.
6
a.m.
As
far
as
the
slide
with
folks,
I
know
those
are
the
maintainers.
I
wonder
if
it
needs
updating
that
might
be
a
separate
conversation
yeah,
the
more
the
more.
A
The
merrier,
I
think
we
should
you
know,
but
that's
I
think,
that's
the
panel
right
yeah
we
were
earlier
talking
about
the
sort
of
the
thing
from
6am,
pacific
or
or
2
a.m
might
to
be
in
my
team.
This
is
the
office
hour.
We're
talking
about
right
and
what
you
see
here
is.
C
A
C
The
topic
of
that
is
another
topic
to
address,
though,
about
to
to
review
the
those
that
are
maintaining
and
potentially
have
that
list
updated
or
have
have
some
move
to
emeritus
or
invite
new
ones
or,
and
part
of
that
has
been
a
topic
of
conversation
in
the
past
about
what
are
those
criteria.
It
was
there's
different
ways
for
us
to
approach
it.
One
is
to
define
it
ourselves.
C
Two
is
to
lean
into
sid
contributor
strategy
and
part
of
the
guidance
that
they're
giving
for
like
what
would
constitute
engagement
and
maintainership,
and
what
is
maintainership
and
these
things
right
now.
That's
universally
left
to
the
and
will
continue
to
be
left
to
the
individual
projects,
but
they
might
provide
some
some
guidance
as
to
what
that
is.
We
have
an
open
issue
to
define
what
it
means
to
be
active.
C
B
B
Why
don't
we
table
a
conversation
for
what
would
be
well
14
days
time
from
today,
where
we
actually
have
that
discussion
as
an
agenda
item
about
the
core
maintainers,
because
I
think
when
we
originally
set
the
guidelines
out,
there
was
some
very
loose
stuff
that
we
put
in,
which
was
around
if
you're
not
active
within
three
months,
but
but
I
think
it
also
gives
us
the
opportunity
to
bring
new
folks
who've
been
contributing
to
to
s
mine
participating
in
the
community
who
are
who
are
maybe
not
in
in
the
sort
of
core
maintainers
group
and
and
also
the
opportunity
to
welcome
them
into
to
the
group
in
a
more
formal
aspect
as
well.
B
So
why
don't
we?
Why
don't
we
add
that
to
the
agenda
for
our
first
meeting
back
after
coop
corner,
because
I
think
that's
a
great
a
great
thing
for
us
to
to
talk
about
and
formalize
and
and
if
you
can
share
some
experience
on
that
as
well.
If
you
wouldn't
mind
can
I
can
I
ask
you
to
to
lead
that
agenda
point,
because
I
literally
know
nothing
about
the
the
way
of
the
other
sig
working
groups.
C
So
the
other
thing
that
the
project
has
is
a
booth,
a
virtual
booth
in
the
project
pavilion
and
so
for
any
of
you
looking
to
coordinate
on
potential
virtual
handouts.
C
Let
me
know
we're
over
the
deadline,
but
my
understanding
is
that
that
may
not
be
so
long
as
we're
willing
to
do
it
ourselves.
There's
still,
I
think
the
avenue
to
get
those
in
so
so.
For
the
rest,
I
guess
I
guess
the
ask
here
is
to
move
on
is
for
those
maintainers
who
might
be
available
for
that
office
hour.
Please,
let
me
know
so
we
can
represent
you
appropriately
in
the
deck
for
those
maintainers
that
are
going
to
be
there
as
well.
C
If
there's
additions
or
changes
to
the
deck,
please
engage
jump
into
the
deck.
I
just
20
minutes
ago
quickly
copied
some
slides
from
last
year
last
time
to
this
time.
So
so
there's.
E
Is
this
house,
is
there
somewhere
specific
that
folks
can
find
this
document.
F
Can
you
I'll
drop
a
link
to
that
on
chat
or
something
yeah?
You
bet
it's
also
it's
in
the
last
couple
of
meeting
minutes.
I
think
okay
cool,
not
that
that's
a
great
question.
B
Yeah,
so
back
back
over
to
to
you
lee
so
now
you
I
believe
you
kenny
we're
going
to
talk
about
service,
mesh
interface,
service,
mesh
performance
and
measuring
and
how
those
three
three
items
align.
C
Yeah,
so
the
genesis,
so
we're
only
going
to
make
it
so
far
into
this
particular
discussion,
but
when
I
I've
got
the
wrong
deck
ups.
Otherwise
I
I'm
sure
that
I
cannot
look
for
it
here,
but
so
so
some
of
you
reviewed
a
blog
post
that
went
out
yesterday
about
meshary
being
the
conformance
tool
for
smi
and
that's
been
way
too
long
in
the
works.
C
There's
more
engagement
with
each
of
the
service
mesh
projects,
each
of
the
service
mesh
teams
about
how
they're,
empowered
to
self
you
know,
there's
more
about
that
tool
and
to
engage
with
each
of
the
teams,
the
mesheri
as
a
tool
and
service
mesh
performance
as
a
specification
centered
on
performance.
C
Those
have
both
been
proposed
a
couple
of
months
ago
for
adoption
into
the
cncf.
They
were
up
for
review
a
month
ago.
The
cnc,
the
toc
didn't
make
it
down
that
far
on
the
list
to
evaluate
them.
They
made
it
down
that
far
yesterday
and
and
they
have
an
open,
ques
and
well.
They
have
an
open
question
as
they'd
like
to
understand
how
the
projects
align
and
if
they
should
come
together
or
if
they
should
be
apart
and
how
do
they
align
and
and.
C
And
so
we
won't
really
make
it
very
far
in
this
discussion
today,
but
I
want
to
bring
that
up
as
consideration
fodder
for
consideration
for
all
of
you.
Potentially
next
time
we
meet
like,
or
maybe
in
advance
of,
that
is
to
send
out
some
thoughts
on
what
that
might
look
like
one
way
or
the
next.
But
to
help
each
of
you
independently,
contemplate
that
to
lead
the
question
a
little
bit
or
I'll
say
that
that
I
think
all
of
the
above
is
so.
They
want
the
projects
in
the
cncf.
C
And
then
the
question
is
just
you
know:
how
do
these
line
marry
up
and
and
how
complementary
and
or
overlapping
are
they
in
and
from
my
perspective,
they
are
well
completely
complimentary.
That
performance
is
quite
long
and
deep
and
wide,
and
if
some
intel
folks
were
here
on
the
phone
or
some
red
hat
folks,
that
the
maintainers
of
s
p
were
on
the
phone
which,
by
the
way
they
need
to
be
consulted
and
the
same
for
measuring
and
the.
C
But
the
talk
with
the
toc
was
sort
of
giving
this
guidance
and
the
toc
members
were
doing
quite
a
lovely
job
of
trying
to
quickly
understand
a
bunch
of
projects,
don't
focus
on
the
space
and
are
looking
to
sig
network
for
some
of
this
guidance,
and
so
so
I
want
to
bring
it
up
today
as
a
food
as
fodder
for
consideration.
C
It's,
I
think
the
myopic
perspective
or
the
very
quick
understanding
of
what
measuring
is
to
the
the
toc
members
that
we're
considering
this
is
that
mescheri
does
two
things.
It
is
the
canonical
implementation
of
s
p
and
that
it
is
the
conformance
tool
for
smi,
so
it's
sort
of
a
spec
tool
and
not
realizing
that
it
is
a
lot
bigger
than
that
it.
It
also
implements
ohm.
It
also
manages
web
assembly
filters.
It
also
there's
there's
a
list
that
doesn't
mean
that
it
is
inappropriate
necessarily
to
bring
them
together.
C
Maybe
that's
a
part
of
their
think.
There's
some
brainstorming
and
sort
of
thinking
a
lot
I
mean
immediately.
It
strikes
me
as
not
not
what
the
projects
were
designed
to
be
and
do.
But
the
last
thought
that
I'll
part
with
here
is
that.
B
B
The
performance
of
a
system
is
absolutely
essential
and
then
I
think,
with
with
service
mesh
complexity
like
there's,
there's
a
heck,
a
lot
that
can
be
done
by
by
having
a
specification
which,
which
gives
you
a
common
view
on
that,
and
there
is
a
definite
sort
of
analogy
to
the
service
match
matrix
capability
which
exists
in
that's
mine
and
yeah
and,
of
course,
mastery
great
great
tool
and
a
good
sort
of
good
reference.
Implementation
of
smi,
as
well
as
a
a
consumer
of
the
specification.
So.
C
Yeah
the
oh,
I
remember
what
I
was
going
to
say.
Part
of
the
the
thinking
allowed
by
a
couple
of
few
of
the
toc
members
was
well,
and
this
was
really.
I
think
I
think
the
va
like
two-thirds
of
the
functionality
of
mastery
are
unbeknownst
to
them,
even
though
you
know
just
be
just
because
they're
short
on
time.
C
But
it
was
the
notion
that
well
hey
if
we're
the
cncf
and
we're
trying
to
help
you
inform
the
world
about
what
a
service
mesh
is
and
what
it
does
like.
If
these
projects
amalgamated.
C
Would
this
be
the
sanctification
by
which
it's
articulate,
you
know
articulated
that
here's
all
of
the
meshes
here's
what
it
takes
to
be
a
mesh?
Here's
all
their
capabilities
and
whether
or
not
they're
in
compliance
with
these
specifications
and
and
it
was
their
holes
and
their
like
discrepancies
or
overlaps
and
under
lapse
in
that
type
of
a
thought,
but
but
that,
but
I'm
just
trying
to
convey
to
the
best
of
my
ability
sort
of
the
sentiment.
C
C
It's
about
twice
as
big
as
sms
or
smi
smp
another
one
and
another
one.
It's
a
long
roadmap,
we're
being
asked
to
engage
with
the
chaos
engineering
team
and
write
a
white
paper
on
chaos,
engineering
in
part,
because
that's
part
of
mastery's
roadmap
as
well
as
cast
engineering
meshery's,
while
it
implements
smi.
It's
also
had
a
hard
time.
C
Well,
I
don't
know
how
to
like
it
like
osm,
maybe
as
an
example,
desires
to
imp
to
to
bring
forth
some
capabilities
that
aren't
defined
by
smi,
and
so
nick
has
an
example,
and
I
have-
and
michelle
have
talked
about
how
to
help
speed
along.
Maybe
certain
existing
specifications
or
potentially
you
know,
propose
new
ones
to
expand
its
scope,
such
that
you
know
such
that
these
things
stay
in
cohesion.
You
know
or
like
with
with
respect
to
measure
anyway
and
smi.
B
To
do
sorry,
I'm
just
just
interrupting
elite
just
to
do
a
quick
time
check
there.
Now
it
is
6
30,
but
technical
issues
in
the
meeting.
So
if,
if
anybody
wants
to
to
continue
the
conversation,
I'm
I'm
happy
to
to
run
along
for
another
another
10
minutes
or
so,
but
but
if
folks
do
have
to
to
drop,
then
of
course
you
know,
we
fully
fully
understand
about
those
folks
want
to
get
their
teas.
C
So
I
know
that
I
know
that
some
folks
might
need
to
go.
Actually,
since
sunku
came
on
it,
it
bears
repeating
for
his
part,
but
but
anyone
else
have
comments
or.
E
Before
I,
I
think
my
initial
take
is
that
it
feels
like
it
would
be
somewhat
useful
to
look
at
smp
independently
of
some
of
the
mystery
stuff,
just
because
it
potentially
is
like
a
much
broader
scope,
and
it
may
be
more
real,
like
I
think
there
may
be
some
stakeholders
that
could
have
an
interest
in
the
s
p
stuff,
whether
that
be
someone
like
the
open,
telemetry
groups
or
other
stuff
like
that,
that
it
may
be
focused
on
just
that
specific
part
without
trying
to
get
up
to
speed
on
the
entirety
of
like
what
the
mystery
project
is
doing.
E
B
And
I
think
is
originally
the
the
both
the
products
were
submitted
to
the
cncf
as
individual
items.
That's
correctly,
isn't
it
so
mesherie
was,
was
an
application
submitted
to
the
cncf
for
inclusion
and
smp
as
a
specification
in
the
same
way
as
smi.
G
Yeah,
I
agree.
Sorry,
it
got
delayed
if
it's
another
office
call
but
yeah.
I
agree
with
that
analysis.
I
think
yeah
I
mean
there's
should
be
a
good
correlation
with
work
going
on
in
smi
and
smp
and
in
a
measuring
by
itself
is
a
separate
entity.
Definitely
there
is
a
relationship
with
respect
to
what
we
could
do
together,
but
and
the
scope
itself
can
be
separated.
G
Out
so
maybe
I'm
missing
the
context.
So
are
there
any
next
steps
being
discussed
here
or.
B
I
guess
that's
that's
the
next
question
so
should
we,
I
think
maybe
the
best
thing
to
do
lee
is,
if
again
like.
I
think
you
did
this
to
me
last
time,
so
I
can
kind
of
get
my
own
back
in
some
ways.
But
like
would
you
be
able
to
to
give
very
quick
tldr
to
the
group
next
time
about
s,
p
and
measuring
and
then
maybe
what
we
do
is
we
go
away
for
a
session?
B
We
can
do
some
thinking
and
then
we
can
have
a
discussion
about
the
call
the
action
items
that
we
should
take.
Does
that
does
that
may
sound
like
a
sensible
course.
C
It
does
to
me
it
also
gives
an
opportunity
to
for
the
other
two
projects
implicated
and
I'm
glad
that
cinco
is
on
because
he
because
he
represents
one
of
them
very
directly
and
for
them
to
kind
of
go
through
the
same
sort
of
procedure
of
presenting
smi
and
what
it's
you
know
so
yeah
thanks
for
that.
Nick
that
helps.
A
E
D
So
I
I
should
tell
you,
I'm
looking
forward
to
updating
you
on
the
multi-cluster
progress
next
time.
We
we
are
making
progress
so
yeah,
I'm.