►
From YouTube: CNCF Serverless Working Group 2019-12-02
Description
Join us for Kubernetes Forums 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
CNCF Serverless Working Group 2019-12-02
A
B
A
We
got
lucky
both
my
parents
and
my
wife's
parents
both
seem
to
take
a
similar
approach
where
they'll
pick
like
a
different
holiday.
Like
my
mom
used
to
always
pick
Thanksgiving
is
you
know
her
holiday
where
she
wants
everybody
to
show
up,
but
they
both
seem
to
agree
that
Christmas
was
you
know,
that's
where
you
guys
spent
time
at
home
and
just
relax
and
stuff
and
I'm
so
glad
they
both
agree
with
that,
because
it
means
it's
so
much
easier,
much
much
more.
Relaxing
that
way.
C
A
A
A
A
A
A
E
E
A
A
You
all
right
17,
all
right,
let's
go
ahead
and
get
started.
Maybe
we
can
end
this
thing
early,
nothing
new
and
the
a
is
I
really
need
to
actually
find
out
from
Austin.
If
he
wants
to
keep
this
one
going,
they
looked
into
the
other
one
we
did.
We
did
actually
talk
to
the
CNC
FTO,
see
about
whether
we
want
to
start
up
a
service
sig
versus
get
incorporated
into
one
of
the
other
SIG's
that's
getting
created
or
create
a
brand
new
one.
A
A
That's
why
we
kind
of
been
letting
it
take
a
backseat
to
our
discussions,
but
I
just
want
to
let
you
guys
know
that
it
is
sort
of
still
lingering
out
there
and
we
will
try
to
figure
out
what
the
Hat,
what
we're
gonna
do
going
forward.
What
the
TOC
wants
us
to
do
going
forward,
but
as
of
right
now,
nothing's
really
changed
and
it
doesn't
in
fact
our
day-to-day
activities
anyway.
So
it's
not
a
big
deal
all
right.
A
Clearly
time
is
there
anything
from
the
community
that
people
like
to
bring
up
as
a
topic
for
discussion?
That's
not
on
the
agenda.
Typically,
this
is
for
newcomers
all
right,
not
hearing
any
I
just
want
to
mention.
Obviously
everybody
knows
about
ku
Connie
you
coming
up.
The
call
for
proposal
is
behind
us.
However.
I
did
notice
just
today
that
the
service
practitioners
summits
announcement
page
or
what
you
want
to
call
it
was
actually
went
online,
and
so
they
are
planning
and
having
one
on
March
30th,
which
I
think
is
a
another
day.
A
Zero
type
of
event,
kind
of
thing
for
ku
gone,
I,
don't
believe
the
CFP
for
that
is
open,
but
I
did
want
to
mention
this,
so
you
guys
can
start
thinking
about
ideas
if
you
do
want
to
submit
a
proposal
for
the
service
practitioners
from
it.
So
just
keep
an
eye
out
for
that
as
I
notice,
the
CFP
comes
online
I'll.
Definitely
let
you
guys
know,
but
obviously
coupon
itself
is
more
sturdy
at
the
April
28th
of
April.
Second,
if
you
want
to
start
making
plans
so
make
sure
everything
is
aware
of
that.
A
All
right
just
reminder.
We
did
agree,
no
calls
next
week
the
week
after
that,
or
do
we
get
to
that.
So
this
is
the
last
call
for
this
week
and
then
we'll
meet
again
what
January
9th
through
the
next
one.
So
just
a
quick
reminder
right,
SDK
call,
I
I
think
we
actually
did
have
a
call
last
week,
but
I
don't
remember
anything
too
exciting
happening
there.
If
somebody
was
on
that
call
who
wants
to
speak
up
feel
free
to
now
otherwise,
I'm
gonna
keep
going
I,
don't
think,
there's
anything
too
exciting
happened.
A
Last
week's
call:
can
anybody
think
of
anything
all
right
moving
forward,
I,
don't
see
Cathy
on
so
there's
no
I,
don't
see
anybody
from
the
work
flow
group
on
the
call
I,
don't
think,
let's
check
so
I'm,
not
sure,
there's
anything
to
mention
their
relative
to
going
on
anything
going
on
with
the
work
flow
stuff.
A
Actually,
what
I
will
do,
as
any
item
I
saw,
was
I'll
try
to
ping
somebody
over
there
to
get
them
to
at
least
join
the
cause,
and
they
think
something
worthy
of
mentioning
happens
in
the
group
that
might
be
of
attention
to
people.
The
one
thing
that
I
could
think
of,
because
I
am
kind
of
watching
it
out
of
the
corner
of
my
eye.
Is
they
are
working
on
a
governance
stock
to
describe
how
you
can
get
you
know,
committer
rights
and
all
that
our
maintainer
rights
in
that
little
subgroup?
A
So,
if
that
is
of
interest
to
you,
feel
free
to
look
at
the
PR,
that's
opened
up
over
there.
If
you
want
to
do
that,
got
a
chance
to
take
a
look
at
it
myself,
but
I
suspect
it's
pretty
probably
pretty
similar
to
what
most
people
would
expect,
but
anyway,
take
a
look.
If
you
want
to
do
that.
Alright,
quick,
go
through
some
issues
here.
These
two
issues
were
there
from
last
week.
I
think
thank
you
Scott.
A
Even
though
I
don't
he's
on
a
call
for
responding
to
both
of
those
and
I'm
going
to
double
back
with
the
author's
to
see
if
we
can
close
those
two
issues,
although
the
first
one
may
lead
to
a
syntactical
wording,
change
nothing
normative
I,
don't
believe
in
the
spec,
but
I'll
work
with
the
authors
of
those
issues
to
see.
If
anything
needs
to
happen
there
and
just
to
keep
the
backlog
moving
along
Klaus
since
you're
on
the
call,
yes,
is
wondering
if
you've
had
it,
yeah
wrap
this
one
from.
D
A
A
Yeah
and
so
anybody
anybody
else
who
may
have
opened
up
an
issue
be
warned
at
some
point:
it
will
ping
you
there.
What
do
you
want
to
do
with
your
issues?
I,
get
that
backlog
down
all
right,
cool,
let's
say
I,
don't
see
mark
on
the
call,
so
I
can't
make
him
about
his.
So,
okay,
we'll
keep
moving
them.
We
have
no
open
PRS,
so
that's
good
only
discuss
there
so
all
right
on
to
the
meat
of
the
meeting
itself.
A
So
as
a
refresher
on
last
week's
call,
we
talked
about
all
the
various
options
in
terms
of
what
to
work
on
next
and
there
are
lots
of
obvious
that
we've
obviously
had
lots
of
different
topics
that
were
brought
up
and
we
kind
of
narrowed
it
down
to
two
that
people
thought
were
worthy
of
discussion
or
possible
work
items
or
workstreams
at
this
point
in
time.
That's
not
to
say
that
some
of
the
other
ones
that
people
thought
were
important
should
not
be
worked
on
at
all.
A
It's
just
sort
of
a
timing
perspective
because
we
seem
to
think
we
can
only
do
one
one
bigger
thing
at
a
time
and
in
fact,
I
actually
am
hoping
that
for
some
of
the
items
there
may
be
some
background
discussions
going
on
in
preparation
for
maybe
working
on
it
after
whatever
the
next
thing
is.
So
we
do
a
little
bit
of
parallel
work
in
terms
of
thinking,
but
in
terms
of
hard
core
work,
we're
probably
gonna
have
to
serialize.
A
You
know
what's
in
scope,
but
that
was
out
of
scope.
Unfortunately,
didn't
actually
give
me
a
concrete
answer
for
that,
but
I
what
he
did
say
without
that
was
interesting
was
he's
actually
in
favor
of
looking
at
subscription
and
discovery
first,
so
well,
I.
That
wasn't
quote
an
official
vote
for
him
towards
which
way
he
go
I'm
going
to
interpret
it
and
in
that
way,
for
the
purposes
of
us
deciding
so
I
just
got
to
throw
that
out
there
to
help
you
guys
as
you
go
through
your
thought
process.
A
F
Is
my
case,
I
talked
to
Jim.
He
is
not
able
to
join
us,
but
yes,
he
is,
or
we
are
on
the
opinion
that
we
should
a
little
bit
postponed
with
the
security,
because
there
is
a
large
number
of
potentially
rather
complex
issues.
Yep
and
the
perhaps
engaging
some
people
who
are
really
focusing
on
on
depth
of
security
will
be
appropriate
at
some
point
later.
Okay,.
A
A
It's
actually
merit
phrase
that
differently
with.
Is
there
anybody
who
would
like
to
speak
now
in
favor
of
one
or
the
other
I?
Think?
Last
time,
last
week's
call
we
had
quite
a
few
people,
speaking
in
favor
of
description
and
discovery.
I'm,
not
sorry
a
subscription
discovery,
but
is
there
anybody
in
the
call
who
would
like
to
speak
to
doing
and
to
end
security
before
we
do
something
like
discovery
in
subscription?
A
D
C
Hello,
if
you
allow
I'm
not
going
to
speak
for
it,
but
I'm
speaking
for
staging
these
things,
because
I
think
we
need
to
go,
do
both
mm-hmm,
but
so
we
will
need
a
some
sort
of
of
registry
API
that
we
agree
on.
That
has
the
ability
to
store
you
know
a
catalogue
of
events,
and
that
has
the
ability
to
store
schemas
and
it
has
to
store
secrets
of
some
sort
and
I.
Think
of
this
of
the
secret.
The
secrets
are
the
less
important
one.
C
That's
why
I'm
favorite
discovery
subscription,
but
ideally
we'll
have
something
that
is
uniform
for
end-to-end
security.
It
will
be
important
to
have
a
model
where
we
can
store
our
secret
information
of
some
sort,
because
we're
doing
we
need
to
be
doing.
You
know,
broadcasts
of
effective
broadcasts
of
data
grabs,
which
means
you
can
especially
key.
C
We
can't
negotiate
session
keys
so
and
security
is
going
to
be
a
fairly
involved
thing
that
requires
distributing
story
and
distributing
key
information,
and
so
we'll
that's
something
we'll
have
to
tackle
I
think
we
have
to
go
in
and
deal
with.
You
know
what
is
the:
what
is
that?
Our
registry
model
look
like
what
is
our
registry
interface
look
like,
and
it
will
probably
also
have
to
go
and
think
about
this
for
multiple
protocols
and
unders,
HTTP
and
I.
C
Think
once
we
have
that
as
a
template
and
that
once
we've
made
that
robust
using
something
that's
a
little
simpler
like
the
the
schema
registry
and
or
the
event
registry,
then
we
can
go
in
and
and
move
ahead
and
start
thinking
about
how
a
cue
ball
will
look
like.
But
ideally
the
interface
is
not
very
different.
C
C
A
And
just
for
clarity's
sake,
we
used
when
you
talk
about
the
registry
stuff.
Are
you
talking
about
the
registry
from
our
that
that's
sitting
on
top
of
an
event
producer,
so
you
can
see
what
he's
producing
or
are
you
thinking
about
some
standalone
registry
someplace?
Or
do
you
actually
see
one
solution
be
able
to
couple
both
so.
C
I
think
I
believe
there
are.
There
are
three
things
that
fall
into
the
discovery
and
subscription
bucket,
and
that
is
and
discovery
discovery
is
I
can
walk
up
to
I,
can
walk
up
to
a
publisher
or
a
delegate
of
the
publisher
and
I
can
ask
which
events
shall
be
expected
from
a
particular
source
and
then
I
can
go
and
subscribe
to
them,
but
then,
once
I
go
to
walk
up
to
that
catalog.
That
catalog
should
also
give
me
a
information
about.
You
know
what
is
contained
in
them.
C
C
G
Be
like
just
curious
when
you
think
about
the
scheme
on
a
registry.
Are
you
thinking
about
an
API
proper
or
are
you
just
thinking
about?
The
actual
format
of
the
scheme
occurs
in
the
cloud
events
we
have
to
ski
my
URL,
it's
a
pointer
right
is
that
is
there?
Is
there
actually
a
proper
API
that
is
necessary,
or
can
we
just
utilize
a
URL
and.
C
So
so,
yes,
so
right
now,
we're
happy
to
have
just
the
URL,
but
I
think
you
should
be
able
to
so
subscriptions
is
effectively
the
ability
to
go
and
and
ask
for
a
for
a
new
subscription
be
created
right.
So
that's
a
really
neat
yeah
and
if
there's
middle
work,
so
if
you're
a
publisher
but
you're
using
some
middleware,
you
probably
want
to
go
in
and
define
a
you
want.
You
want
to
publish
into
that
middleware
what
shape
your
events
are,
so
which
events
you
offer
and
what
shape
your
events
are.
So,
yes,
I,
believe.
G
C
G
C
There's
a
I
think
there
is
a
lightweight
and
and
I
don't
think
it's
it's
it's
a
spectacular.
You
know
new
invention,
it's
effective,
crud
API,
which
is
going
to
be
for
HTTP,
very
simple
rest
service,
but
the
very
simple
rest
service.
We
have
to
go
and
agree
on
the
shape
of
that
very
simple
restrooms
and
the
simpler
we
can
make
it
the
better.
It
is
yeah.
G
I
understood
I
was
just
trying
to
the
the
only
tricky
bit
then
now
we
have
to
deal
with
the
the
earths
and
everything
else.
So
I
just
want
to
understand
if
it's
strictly
necessary
or
if
we
might
be
able
to
go
ahead
and
and
focus
and
reap
first
but
anyways.
Now
we
had
I'll
take
the
conversation
elsewhere,
I
just
Stan
I
think
the
reeds
are
also.
A
Okay,
so
I
feel
like
there
is
something
I
have
to
say
here,
just
as
sort
of
like
a
moderator,
the
call
well
that's
saying
whether
I
agree
or
disagree
with
anything
Clemens
said
here,
even
though
it
would
be
a
foolish
to
ever
disagree
with
them.
I
think
it's
important
to
say
that
everything
that
Clemens
just
said
is
him
speaking
for
himself
and
or
Microsoft.
Yes,.
A
So
if,
for
example,
we
choose
to
work
on
discovery
and
subscription
discovery
on
subscription,
the
exact
scope
of
that,
the
exact
thing
that
what's
in
or
out
is
TVD
right,
we
will
decide
what
that
is.
The
agreement
here
is
not
to
say:
yes,
everything
come
in
just
said,
we're
going
to
work
on.
Rather
it's
organic
we're
agreeing
on
that
general
direction
and
then,
as
the
process
goes
along,
we'll
figure
out
exactly
what's
in
tirana,
scope
and
whether
aligned
to
look
Clemens
just
said
or
not,
is
yet
to
be
determined.
A
Anybody
else
want
to
speak
up
then
in
favor
of
or
talk
about
and
end
security
at
all,
because
the
reason
I'm
asking
it
this
way.
I
know
it's
a
little
bit
of
a
leading
question
is
because
it
seemed
like
most
people
in
last
week's
call
did
want
to
say
yes
discover
inscription
was
the
choice,
but
I
don't
want
to
downplay
and
then
security.
If
somebody
really
wants
to
promote
that
one
as
the
one
is
the
way
to
go.
A
A
All
right,
so
that
is
approved,
work
on
that
okay.
So
then,
the
next
steps
so
a
while
ago,
I
put
together
this
Doc
simply
as
a
thought
exercise
for
myself
just
to
see
what
I
thought
might
be
included
as
a
starting
point
for
mainly
around
the
subscription
API
I
didn't
I,
don't
think
I
really
much
include
things
around
discovery
and
I
definitely
didn't
include
schema
register,
hey
they're,
the
stuff.
A
How
do
people
want
to
take
the
next
steps
here
generally
going
forward?
I
suspect
people,
though,
probably
not
a
whole
lot
of
time
before
they
beginning
a
year,
so
obviously
you
can
think
about
it
over.
You
know
holidays
coming
up,
but
do
we
want
to
start
with
a
clean
sheet
of
paper?
Do
people
think
that
this
is
enough
of
a
starting
point?
A
We
could
start
with
this
and
then
people
just
start
editing
it
like
crazy
through
Google
Doc
and
then
once
it
settles
down,
we
can
move
it
over
to
mark
down
and
do
PRS
is
their
preference
people
have
in
terms
of
how
we
go
forward
here.
I'm.
To
be
honest,
even
though
I
wrote
this
I'm
perfectly
okay,
with
starting
with
this
fresh
piece
of
paper,
if
somebody
wants
to
do
that,
I,
don't
really
care
if
I
know
how
you
guys
feel
anybody
want
to
volunteer
an
opinion
us
Scott,
your
hands
up
and.
E
C
So
why
our
protocol
is
something
that
I
care
about
first
and
then
an
implementation
on
top
of
the
on
top
of
whatever
the
web
protocol
is
so
I
would
say.
Declarative
API
sounds
like
you're
writing
code,
which
is
great,
but
we
also
need
to
need
to
worry
about
how
the
pieces
talk
to
each
other
across
the
network.
C
F
A
Scott,
let
me
ask
you
this
I
I
do
agree
with
you
that
most
of
what
I
put
down
here
kind
of
had
an
RPC
mechanism
in
mind.
I
do
agree
with
that,
and
I
probably
should
not
have
done
it
that
way.
However,
there
are
I.
It
would
not
be
very
difficult
for
me
to
twiddle
this,
so
that,
instead
of
saying
okay
here
are
the
inputs-
and
here
are
the
outputs.
E
A
Kind
of
where
I
was
gonna
go
with
this.
Yes,
yes,
and
then
at
some
point,
though
I
do
think
we're
probably
going
to
have
to
have
a
discussion
so
which
one
do
we
promote,
so
that
we
have
interoperability.
That's
obviously
a
key
point
in
all
this
right,
because
if
half
the
world
uses
RPC
and
half
the
world
uses
declarative,
we
may
not
have
helped
the
community
right.
It.
A
But
my
point,
my
bigger
point
is
I'm,
not
sure
we
have
to
decide
that
today,
that's
right,
so
the
sky,
if
we
decide
to
head
in
the
direction
of
using
this
talk
as
a
starting
point,
I
will
take
the
excellent
action
I.
Don't
you
try
to
twiddle
this
around
to
not
assume
one
particular
wire
protocol
that
fair
yep?
Thank
you,
okay.
A
C
So
so
I
will
probably
bring
a
I'll,
probably
bring
a
protocol,
a
protocol
level
suggestion,
because
the
the
you
know
how
that
looks.
How
that
looks
up
at
the
programming
model
layer
doesn't
worry
me
as
much
as
it
does
cross
system
interoperability,
because
this
this
should
work
with
kubernetes.
But
that's
not
that's.
Not
everything
I
exist
in
the
world
that
neither
is
the
hole,
no
go.
Cn
CF
infrastructure
I
want
this.
This
should
be
as
masters
called
Vince
as
as
fairly
large
reach.
H
I
Traditionally,
if
you
look
at
this,
I
would
have
assumed
to
be
three
separate
guys.
Like
one
physician
our
discovery,
one
for
I
would
I
guess
my
gut
would
be.
We
should
spend
more
time
defining
the
discovery
piece
first
and
then,
once
we
go
from
there,
we
can
actually
then
define
the
model.
I.
Think
I
read
the
doc.
Actually
really
is
awesome,
I
didn't
notice.
It
was
a
doc
and
very
deep
into
this
subscription
to
hash
out
what
discovery
means.
I
A
G
Delay
I
think
we
should
I
mean
I,
know
I
I
think
that's
the
same.
Talk
that
I
have
already
made
comments
and
looks
like
other
people
have
as
well
I
preferred
that
unless
there's
a
very
strong
reason
to
start
from
blank
sheet,
we
use
that
as
a
starting
point
and
I
prefer
doing
it
using
Google
Docs
over
Gober
markup,
okay,.
A
Okay,
so
let
me
let
me
make
a
little
formal.
Is
there
any
objection,
then
to
starting
with
that
doc,
which
is
under
that
rough
idea
link?
That's
in
the
meeting
notes
and
we'll
obviously
keep
it
as
a
Google
Doc
for
down,
because
that's
easier
to
do
quick,
edits
going
back
and
forth
and
then
once
we
feel
like
the
doc
is
starting
to
settle
down,
then
we
could
switch
over
to
something
like
markdown
in
the
github
repo
and
do
PRS
and
issues
like
we
did
for
cloud
events
that
sounds
fair.
A
Is
there
anything
else
worth
discussing
then
on
this
call,
because
that's
technically
into
the
agenda
are
there,
do
you
guys
want
to
start
having
more
discussions
around
the
document
itself?
Would
you
rather
wait
and
start
making
comments
into
the
doc
over
the
next
couple
weeks?
It's
up
to
you
guys.
Maybe
we
can
end
the
call
early.
We
can
have
discussion
now
about
from
a
tech
perspective.
It's
clearly
up
to
you
guys.
We
have
half
an
hour
to
do
stuff.
If
you
really
want
to
I.
C
Am
a
little
worried
about
the
or
I'm
I'm
worried
about
making
sure
that
the
the
you
know
there
are
the
API
definitions,
whatever
they
are
map
into
wire
interactions
that
can
be
implemented
by
the
publishers
or
by
generic
middleware
in
a
way
that
you
can
go
and
realize,
though.
So,
if
you
have
a
declarative
API,
you
have
a
configuration
declaration
that
says:
I
need
to
go
and
subscribe
to
these
three
things.
C
Then
there's
will
presumably
be
an
engine
that
does
that
realizes
those,
but
the
engine
needs
to
go
and
talk
to
external
parties
and
so
that
API
that
that
engine
uses
to
talk
to
external
parties,
I
think
that's
the
more
difficult
piece
then
the
or
or
the
the
piece
that
really
needs
to
be
harmonized
rather
than
the
piece
that
really
says.
You
know:
here's
how
to
tie
up
a
subscription
into
a
particular
infrastructure,
because
that's
a
matter
of
taste
and
and
whatever
your
infrastructure
is.
A
Okay,
yeah
I
think
I
I
kind
of
focus
on
very
beginning
of
what
you
said
there,
which
was
you
need
time
to
go.
Look
at
that
dock
and
I
heard
everything
else.
You
said
there
too,
but
I
think
just
my
gut
reaction
based
upon
how
I've
seen
this
group
work
in
the
past
I
think
most
people
have
probably
not
really
looked
at
the
dock
very
much,
even
though
it's
been
out
there
for
I
think
at
least
a
week
or
so
so.
A
I
would
like
to
suggest
that
we
don't
go
into
a
deep
dive
discussion
here
about
anything,
but
rather
give
people
the
opportunity
to
look
at
it
over
the
next
couple
weeks
and
then,
when
we
come
back
when
in
the
January
9th
then
start
having
a
discussion
based
upon
the
comments
people
left
on
the
dock.
Does
that
sound
fair
to
people
yeah
I'm
coming
ahead
her
to
come?
Is
anybody
else,
want
a
voice,
an
opinion
yes
or
no
sounds
great.
Ok,
that
sounds
good.
A
A
That
way,
people
have
very
concrete
of
text
to
review
of
what
your
idea
is,
and
it's
not
just
some
abstract
thing
in
a
comment:
I
find
it
much
easier
for
people
to
say:
aha
I
see
what
you
mean,
I
like
it
or
oh
I.
Don't
like
that.
Now
that
I've
actually
seen
what
you
what
that
means
realized
and
then
they
could
say
no
I
think
it's
a
lot
easier.
A
If
you
just
put
your
actual
text
in
there
and
now,
we've
who
can
compare
and
contrast
the
various
ideas
in
a
more
concrete
form,
rather
than
just
an
abstract
idea
in
a
comment
so
don't
feel
like
you
have
to
build
nestle
upon
what's
already
in
there
and
keep
that
that
basic
idea,
you
even
have
a
completely
radically
new
idea,
just
put
it
after
the
previous
one.
So
people
can
compare
and
contrast
just
put
like
a
line.
A
A
Okay,
is
it
because
I
to
assume
that
anything
in
there
has
to
stay
if
the
entire
thing
could
be
tossed
out
over
time?
I
have
no
problem
with
that
whatsoever.
This
was
just
a
brainstorming
exercise
for
myself
to
get
a
little
more
concrete,
miss
in
my
mind
in
terms
of
what
I
thought
people
were
thinking.