►
From YouTube: CNCF Serverless Working Group 9/21/17
Description
Join us for KubeCon + CloudNativeCon in Barcelona May 20 - 23, Shanghai June 24 - 26, and San Diego November 18 - 21! Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.
A
B
B
C
B
C
Would
be
that
would
be
cool
yeah
I
just
got
to
sign
off
from
your
boss,
IBM.
Yes,
that's.
B
Quite
a
few,
if
you
look
at
the
agenda
doc,
you
can
see
all
the
meetings
there.
It's
I
don't
know,
I
think
maybe
twenty
or
so
I
think
we've
probably
three
or
four
demos
left
yeah.
B
E
B
B
B
D
We
realized
that
we're
thinking
about
very
similar
things,
and
so
we
wanted
to
contribute
what
we've
learned
and
not
duplicate
some
of
the
work
that
the
community
is
doing
to
standardize.
You
know
name
functions,
have
had
us,
get
them
up
and
running
and
find
them,
and
so
that's
kind
of
how
we
found
Doug
and
I
sent
a
message,
and
so
we
wanted
to
kind
of
show
you
what
clays
about
and
then
discuss
a
little
bit
of
the
work
that
we've
been
doing
around
the
standardization.
D
And
then
we
can
tell
you
a
little
bit
about
some
of
the
next
steps
that
we
think
we
can
help
with
the
working
group.
So
with
that,
let
me
just
kind
of
give
you
a
quick
overview
of
what
clay
is
so
clays
a
way
to
put
up
code
really
quickly,
as
I
mentioned
and
I'll
show
you
two
things
so.
First
just
start
off
by
showing
you
how
we
create
a
function
and
then
I
will
show
you
some
of
the
ways
in
which
we
think
people
will
use
it.
D
So
when
you
log
into
clay,
you
get
a
list
of
all
the
services
that
you've
created
and
then
your
can
also
see
categories
of
functions
that
people
have
put
into
various
groupings.
And
here
we
have
the
groupings
around
different
companies.
So
all
the
functions
around
getting
started,
working
with
slab
or
clear,
MIT
or
cloud
meri
and
I'll
just
show
you
now
how
to
create
your
function
really
quickly.
So,
in
order
to
create
a
function,
all
you
need
to
do
is
give
it
a
name.
D
And
you
don't
have
to
specify
anything
extra
for
a
function
in
place,
so
you
don't
have
to
you
know,
tell
it
which
provider
you're
sending
it
to
or
what
the
size
you
know
what
memory
requires
and
you
don't
have
to
connect
it
with
an
API
gateway.
It's
basically
we've
done.
You
know
all
of
those
choices
for
you,
so
you
get
up
and
running
as
quickly
as
possible
and
then
the
things
that
we've
added
around,
that
is
that
we
have
some
metadata
around
what
parameters
every
function
expect
and
we
do
some.
D
We
help
developers
get
started
really
quickly.
Setting
up
what
their
you
know.
Function
definition
looks
like
so
that
they
can
start
coding
around
that.
So
now
that
we've
created
this
function,
you
can
see
it
exists
at
this
URL
exact
that
clay
that
run
with
my
username
and
then
the
name
of
the
function,
and
so
you
can
really
quickly.
You
know,
install
our
clay
client
in.
D
Be
able
to
call
that
function
by
just
saying
clay
that
run
green
clay,
WG
function
or
you
can
kernel
it
and
turn
it
off,
so
you
can
edit.
You
know
we
have
a
set
of
parameters
that
we've
set
up,
that
you
can
add.
So
you
can
add
a
text
box,
a
boolean
box
and
address.
We
also
have
the
ability
to
upload
a
file
which
makes
it
much
easier
to
handle.
You
know
writing
code
around
file
uploads.
D
This
is
how
you
know
getting
into
the
meat
of
it
when
you,
when
you
setup
a
function
play
we
create
like
a
little
editor
in
the
cloud
where
you
can
edit
the
code
directly.
So
this
is
just
a
little
bit
of
help
around
how
we
set
it
up,
but
really
all
you
need
to
notice.
Is
you
get
an
event
with
some
variables
that
are
passed
to
it,
whether
you've
hit
it
as
a
post
or
again
you
get
them
in
the
same
way,
and
then
you
have
two
functions.
D
D
Send
them
to
right
now
we
send
them
to
AWS
lambda
and
Google
cloud
functions,
but
we're
kind
of
set
up
to
abstract
away
from
any
of
the
cloud
providers.
So
you
know,
even
if
it's
a
kubernetes
stack
in
the
background,
it
might
be
that
in
the
end,
will
allow
people
to
run
it
in
their
own
infrastructure.
But
for
now
we
kind
of
managed
where
it
goes.
F
D
So
I'll
just
go
to
this,
explore
page
we're
building
these
categories
of
different
functions
and
different
categories,
around
specific
SAS
providers
that
exist,
and
so,
if
you
wanted
to
get
started,
writing
code
with
github
API
really
quickly.
Instead
of
going
and
setting
up
your
local
environment
and
then
figuring
out
how
to
push
that
up
into
product,
you
would
just
find
something
that
you
know
it's
approximately
what
you
want.
D
D
So
you
can
test
out
what
this
function
does
you
can
see
that
it
returns
the
data.
You
know
this
particular
data
and
if
you
think
this
is
a
good
starting
point
for
yourself,
you
can
click
for
it
and
make
a
copy
of
this
function
instantly
to
your
account.
And
then
you
can
start
editing
that
code
right
there
and
publish
it.
So,
basically,
the
way
that
we
think
people
will
use
this
is
one
to
interact
to
build
on
top
of
existing
API
is
really
quickly
and
then
share
these
reusable
building
blocks
with
the
world.
D
So
we
have
like
a
new
API
that
gets
medium
user
posts
which
doesn't
exist
today,
but
it's
something
that
others
would
find
useful
and
then
to
we're.
Also
thinking
of
it
as
a
way
takes
so
that's
kind
of
the
extending
existing
SAS
products
with
API
is
on
top
of
there
yet
and
then
the
second
one
is
we're
using.
We
have
an
M
mint
feature
where
you
can
write
some
codes
and
then
you
can
embed
it
into
a
blog
post.
D
G
One
question:
I:
have
it's
your
own?
You
mention
about
being
able
to
work
with
different
cloud
providers,
so
Google
and
Amazon,
maybe
a
bit
similar,
but
with
Asia,
for
example.
Events
is
something
else,
and
it
sort
of
more
structured
and
derived
from
a
variety
of
sources
and
all
that.
So
how
do
you
plan
on
abstracting
those
differences,
yeah.
A
D
And
I
think
you
know
just
to
add
to
that.
You
know
the
way
we're
thinking
of
it
is
that
each
you
know
like
we
have
our
own
current
model
today,
where
you
just
you
get
the
event
which
is
parsed
from
you
know.
The
expectation
is
that
you
hit
this
URL
as
eh
as
an
HTTP
POST,
and
then
we
pass
those
variables
into
our
event
variable.
So
we're
thinking
of
these
circular
functions
as
HTTP
endpoints
exclusively
at
development,
and
so
all
of
our
events
are
just
JSON
objects.
So
there's
no
we're
not
really
thinking
of
them.
D
G
D
Yeah,
of
course,
I
mean
it's
a
we're
kind
of
limiting
the
use
case
on
purpose,
just
a
starting
point-
and
you
know,
like
the
positioning
that
we
have
this
like
this-
is
a
really
fast
way
to
put
code
up
in
the
cloud
without
thinking
and
doing
it
around
like
a
very
specific
set
of
use,
cases
initially
and
then
starting
to
expand
potentially
with
you
know
some
something
that
would
hold
events
and
then
call
functions
as
needed.
So
we
want
to
just
get
a
little
bit
of
usage.
C
Do
like
CY
I
think
this
looks
really
clean
and
especially
the
idea
of
having
the
catalog
I
think
that's
something
that
a
lot
of
different
projects
are
looking
at
I
guess
with
abstracting
everything
away,
there
is
still
a
cost
to
pay
and
whether
the
choices
that
you've
made
are
actually
going
to
suit
everybody's
needs
is
a
is
another
question.
I
guess,
so
that's
something
that
at
some
point
in
your
story,
you
might
need
to
think
about
yeah.
D
One
thing
that
I
should
show
you
as
well,
as
that
you
know
we
found
it
really
frustrating
to
work
with
lambdas
logs
currently,
and
so
we
started
to
do
a
few
things
around
that
that
might
be
interesting
and
and
it's
it's
a
little
bit
because
we
can
control
the
UI
and
we're
controlling
kind
of
the
experience
here.
So
you
know,
for
example,
if
I,
if
I
just
type,
you
know
console.log
bars
here,
I,
don't.
D
D
So
you
know
we
can.
We
can
highlight
the
error
in
the
code.
We
can
tell
you
that
this
was.
You
know
that
it's
undefined,
so
we
can
help
people
start
to
write
code
in
the
serverless
way
really
quickly,
because
we
control
the
whole
experience
and
then
and
another
thing-
that's
interesting
as
part
of
just
making
the
coding
experience
really
simple.
You
know
we
do
things
were
a
node,
for
example,
you
don't
even
need
to
add
the
dependencies
if
you
type
underscore
it'll.
D
Add
that
immediately,
so
you
can't
make
mistakes
in
certain
ways
in
within
this
framework,
and
it's
all
about
you
know
the
way
we
we
think
about.
It
is
how
we
improve
the
developer
user
experience
and
how
they
use
these
advances
in
infrastructure
technology
like
serverless
to
improve
the
technics
user
experience
around.
You
know
limited
set
of
use
cases
for
now.
So
is
this
open
source?
D
And
so
this
is
not
open
source,
but
we
we
have
like
an
open
source
CLI
that
uses
it
and
we're
starting
to
think
about
which
parts
of
this
we
can
open
source.
You
know
like
which
part
of
this
is
our
IP
and
which
things
we
can
just
openly
share
with
the
community.
That's
part
of
why
we
wanted
to
discuss
this
with
you
and
see
like
maybe
we
can
participate
and
have
the
function
definition
open
source.
D
We
can
make
it
so
that
if
you
want
to
use
like
clay
and
run
it-
and
you
know,
run
a
version
of
this
at
yourself-
it
can
all
be
open-source,
and
if
you
just
want
to
use
our
Postal
Service
for
convenience-
and
that's
that's-
you
know
what
will
charge
for
in
the
future.
So
we're
still
in
the
early
stages
I'm
thinking
about
how
to
which
which
parts
can
be
made
available.
B
C
D
Part
of
one
thing
that
might
not
have
been
super
clear
in
my
demonstration
so
far
is
that
one
of
the
things
that
were
pretty
excited
about
this
that
a
lot
of
these
you
know
the
fact
that
you
can
have
these
little
functions
means
that
people
can
share
almost
like
these
templates,
even
if
they
require
private
parameters,
like
you
know,
private
keys
and
stuff
that
we
can
allow
you
to
get
started
with
those
functions
quickly.
So
one.
D
Quickly
here
is,
you
know
we
were
working
with
cloud
Neri
and
creating
functions.
That
would,
for
example,
take
except
an
image
and
then
using
AWS
s,
image
recognition
run
some
analysis
on
that
and
tag
it
and
then
upload
it.
So
that's
kind
of
a
useful
feature
for
their
developers
so
that
they
have
this
new,
almost
combines
API
that
that
they
can
make
available
and
have
us
build
a
community
around
their
api's.
A
D
Just
use
production
so
when
you
click
fork
to
use
it
identifies
all
the
API
keys
that
are
necessary
for
this
function
to
be
used,
and
if
you
have,
if
you're
able
to
enter
them
or
in
the
future,
hopefully
we'll
be
able
to
enter
them
for
you.
If
you're,
storing
them
with
us,
then
you
can
get
started.
Creating
things
really
quickly
and
not
have
to
you
know,
even
do
any
setup
or
worry
about
where
your
keys
are,
how
you're
gonna
set
them
up.
D
So
just
to
go
back
to
that
earlier
question
of
like
how
are
we
different,
then
you
know
server
lists
or
other
frameworks
that
exist,
we're
basically
a
way
where
we
were
we're
not
giving
we're
not
saying
here's
a
simpler
way
to
configure
the
system.
We're
saying
this
is
full
death
experience
around
how
to
create
these
functions,
and
all
you
have
to
do
is
write
the
code
yeah.
D
H
A
D
A
A
Like
but
think
so
we
could
see
being
helpful
to
have
is
like,
what's
like
a
good
way
to
define
what
what
the
inputs
to
a
function
are
like
right.
Now
we
have
this
kind
of
metadata
where,
where,
where
we're
saying
hey
here's
the
parameters
that
are
there
and
while
we
haven't
done
it
yet
that
like
provides
the
opportunity
in
the
future
to
do
really
nice
things
like
being
able
to
like
provide
validation
on
that,
for
example,
so
that,
like
the
developer,
just
taps
it,
you
know,
go
in
and
like
write
all
the
validation
code
themselves.
G
G
D
We
wanted
to
be
able
to
autocomplete
against
various
object
properties
that
you
should
expect
to
receive
and
to
be
able
to.
You
know
help
you
understand
what
kind
of
things
this
web
hook
will
expect
to
make
that
experience
better,
and
so
we
started
to
look
at
like
a
schema
around
like
how
should
vents
looked
like,
and
we
noticed.
A
A
B
Right
well,
I'm
gonna
have
to
call
tile
on
this.
You
know
it's
very,
very
cool
and
I'd
I'd
love
to
see
more
and
I'll
be
playing
with
it.
Move
on
to
the
doc,
but
I
want
to
thank
you
guys
for
presenting
and
I
would
love
it
if
you
guys
could
stick
around
not
just
for
today's
call,
but
for
other
phone
calls.
If
you
have
any
input
on
the
on
the
way,
prove
that
were
working
on
we'd
love
to
get
your
input
on
that
as
well.
Of
course,
yeah.
D
B
B
For
your
time,
all
right
cool,
thank
you
guys
very
much
for
presenting
I
was
really
cool.
Alright.
So
let's
switch
back
to
the
white
paper
itself.
There
a
couple
things
that
I
want
to
discuss
on
today's
call.
First,
when
was
the
call
I
guess
it
was
Tuesday
the
TOC
call.
There
was
a
request.
It
may
have
been
from
Brian
grant
the
camera
for
sure,
but
he
somebody
was
asking
for
a
comparison
of
the
coupe
hosted
functions
of
service
offerings
that
run
a
Coronet
eight
against
like
serverless
vision
or
some
of
the
other
ones.
B
That
can
read
all
the
names
he's,
not
a
comparison
of
those.
Those
are
something
it's
not
too.
Controversial
might
be
a
good
thing
to
add
to
the
white
paper,
and
so
first
I
want
to
know
whether
people
thought
that
was
a
good
idea
and
if
so,
if
someone's
willing
to
volunteer
to
write
that
up
relatively
quickly,
I
would.
C
Be
slightly
wary
of
doing
that,
Doug
main.
The
main
reason
is
because
of
the
velocity.
These
projects
are
moving
a
lot
of
them,
don't
even
have
1.0
yet
so
while
they
may
be
stable
enough
to
run
with
a
production
workload,
they
may
not
be
stable
in
the
in
the
way
that
their
API
is
breaking
and
changing.
I
I
would
be
I.
C
Guess
worried
that
one
of
the
frameworks
pick
one
I,
don't
know
open,
whisk
didn't
work
for
Cuba
Nettie's
and
then
the
Red
Hat
work
came
through
and
now
it
does
and
people
are
looking
at
the
document
and
thinking
oh
well.
No,
this
is
the
one
source
of
truth,
and
then
we
have
to
keep
up
to
date.
I
think
what
was
good
was
where
we
have
that
summary
section,
and
we
say
we
said
what
our
frameworks
are
doing.
I
think
that's
something
that
that
probably
covers
some
of
it
already.
Okay,.
B
I
There
needs
to
be
some
sort
of
a
bar
for
which
ones
you
wanna
compare
against,
which
ones
you
would
in
and
it
will
be
kind
of
an
ever-growing
list
that
might
have
you
know
you
might
get
a
single
organization
that
spins
up
some
version
of
this
and
then
another
organization
spins
up
their
own
version
of
it
and
so
like.
How
do
you
decide
which
ones
of
those
are
credible
and
which
ones
aren't
right?.
F
And
I
think
this
goes
also
to
what
we
talked
about
last
week
or
the
week
before.
We've
got
that
CN
CF
cloud
native
landscape
I
think
there
was
also
an
idea
to
do
that.
The
storage
working
group
as
well
as
this,
if
anything,
maybe
it's
it
just
copies-
that
model
doesn't
necessarily
do
a
matrix
comparison,
but
maybe
it's
just
a
view
into
the
landscape.
What's
there
that's
regularly
updated,
okay.
G
F
B
G
Valuable
to
have
a
table
instead
of
each
one
of
the
project
listing
you
know
just
couple
of
bullets,
but
maybe
tables
of
which
you
know
which
languages
are
supported.
You
know
which
serve
and
sources
etc.
You
know
we
can
create
several
categories
and
then
is
project
list
just
like
in
in
a
Wikipedia.
Well,.
B
C
B
K
This
is
basically
kind
of
what
Brian
and
the
TOC
is
expecting
this
working
group
to
come
up
with
and
I
think
you
have
a
lot
of
that
information
already
there
and
the
white
paper.
It
just
needs
to
be
distilled
in
a
Google
sheet
and
I.
Think
once
you
have
it
there,
it's
a
little
bit
easy
to
constantly
update,
and
so
that's
that's
it
I.
J
G
B
I
was
definitely
hearing
some
concern
about
the
doc
or
putting
something
like
this
into
the
white
paper,
because
it's
gonna
be
stale
immediately
gonna
be
at
a
date,
could
be
mislead.
The
people
as
bad
things
are
rapidly
changing,
does
putting
this
into
some
sort
of
Google
Doc
spreadsheet
thing
where
it's
implied
it's
supposed
to
be
a
sort
of
living
doc.
Does
that
soften
anybody's
opinion
as
to
whether
we
should
do
something
similar
to
what
Chris
has
listed
here
for
the
storage
guys.
G
K
It's
it's
surprisingly
worked
well
from
our
different,
not
only
the
CNC,
F
landscape
diagram.
That's
been
a
bit
of
a
an
entertaining
thing
to
deal
with,
but
you'd
be
surprised
how
many
vendors
and
folks
want
to
make
sure
their
stuff
is
up
to
date
on
that
list,
and
then
same
goes
with
a
storage
thing.
It
kind
of
it
kind
of
acts
as
a
forcing
function,
because
it
is
somewhat
of
a
diagram.
People
go
to
to
kind
of
learn.
What's
on,
what's
going
on
so
I,
don't
think
we'll
have
trouble
actually
keeping
it
up
the
day.
F
K
F
K
K
B
B
L
F
L
B
B
Interesting
yeah,
alright,
so
moving
on
then
I
think
we've
got
that
one
done.
Thank
you
guys.
So
what
I
wanted
to
do
is
I
know,
there's
still
several
comments
in
the
doc,
but
there
was
one
section
in
particular
that
I
want
to
call
out
except
I,
don't
might
be
worth
having
a
kind
of
discussion
on
it
and
from
better
or
worse
Chrisman's.
I
picked
on
your
section.
The
fast
vs.
service
discussion,
the
doc,
because
that
seemed
to
be
going
back
and
forth
a
bit.
I
I'll
leave
it
up
to
the
group,
I
mean
I,
think
you
know
going
back
a
little
I
volunteered.
That
I
would
write
out
what
I
could
based
on
kind
of
you
know
the
listview
point
that
I
have
based
on
my
employer
and
trying
to
make
it.
You
know
a
vendor
neutral
thing.
It
does
create
a
box.
It
does
mean
that
some
things
are
in
the
box.
Some
things
are
out
of
the
box.
I
I
think
it's
good
to
have
bounds
to
a
definition,
so
I
mean
I'm
open
to
discussion
with
folks,
obviously
open
to
hear
what
what
you
know.
People
want
to
suggest
instead
of
it
but
I
I
do
think
it's
good
for
us
to
have
some
firm
guidelines
as
to
again
what
fits
in
the
box
and
what
doesn't
fit
in
that
box.
Okay,.
G
I
I
think
that
it
depends
on
what
the
customer
wants.
You
know,
even
in
Amazon,
you
could
go
and
get
a
spot
instance
for
a
VM
or
you
could
go
and
provision
stuff
up
front
for
some
applications
which
are
latency
sensitive.
They
don't
wanna,
wait
200
over
to
a
second
milliseconds.
The
second,
the
a
customer
may
just
say:
yeah
I
want
this
function,
I'm
willing
to
pay
a
bit
more
and
I
want
this
function
to
be
warm
and
ready.
G
B
H
I
think
it's
just
an
aspect
of
what
you're
getting
like.
There
is
always
a
cost,
whether
the
customer
or
whether
the
users,
the
one
that's
sort
of
paying
for
it
sort
of
upfront
I
think
is
maybe
the
question
you
know,
there's
a
you
know
the
causes.
You
know
your
trade-off
between
sort
of
high
responsive.
You
want
the
function
to
be
whether
you're
paying
to
sort
of
keep
sort
of
containers
or
whatever
running
in
the
background,
I
think
it's
it's
a
it's
not
a
clear
binary
choice.
There
I.
I
Would
wholeheartedly
disagree
and
I
mean
it's
it's
when
we
look
at
kind
of
the
space
of
what
people
have
considered
service
here,
industry-wide,
the
the
know
paying
for
Idol
is
a
key
aspect
of
what
draws
people
to
these
technologies
and
I.
Think
that
comments
about
well,
there's
always
someone
paying
for
something
was
truer,
but
again
from
the
perspective
of
the
consumer.
The
technology
is
from
an
organizational
perspective,
going
from
hey,
I,
run,
servers
or
I
run
instances
or
I
run
containers.
Oh
I
run
nothing
but
I
have
an
application
that
works.
I
Of
aspects
like
performance
because
again
from
the
various
cloud
providers,
the
various
service
platform
providers,
I,
should
say
out
there.
Any
one
of
us
could
be
working
on
features
that
make
cold
starts
a
non-issue
and
so
I,
don't
think.
I
would
write
a
definition
that
is
respective
of
a
current
implementation.
H
That
would
suggest
that
we
came
to
sort
of
you
know
that
would
kind
of
sort
of
basically
rule
out
anybody
sort
of
hosting
their
own
platforms
to
provide
sort
of
like
service
capabilities
across
like
so,
if
you're
a
large
corporation
and
you
maybe
sort
of
have
some
capacity
dedicated
to
providing
this.
The
consumers
within
your
organization,
maybe
aren't
saying
a
direct
cause,
but
the
organization
is
its
yep.
It's
I.
I
G
G
I
Yeah
you're
you're,
taking
you
you're,
taking
your
personal
experience
with
a
given
product
in
the
funny
across
the
white
paper
right.
What
I'm
saying
here
is
that
this
definition
is
meant
to
be
broad
and
widely
scoped
so
that
it
fits
many
different
platforms.
Many
different
tools
that
meet
these
rounds
don't
take.
G
Your
limit
increase
I'm,
not
limiting
I,
think
your
definition
is
limiting
in
wherever
I
wrote.
You
know
anything.
I
wrote
in
the
dark.
I
took
a
lot
of
examples
from
Amazon
from
major
from
most
of
the
providers
and
I
try
to
generalize
what
you're
doing
is
not
generalizing
is
limiting
and
I'm
trying
to
generalize.
So,
if
you're
saying
I'm
limiting
you
serve
the
other
way
around
right.
Well,.
I
G
I
G
Going
after
limited
use
cases-
and
you
know
I'm,
better
I'm,
going
to
put
personal
back
with
you
that
once
you
start
seeing
those
use
cases
Google
trying
to
push
a
you
know,
lambda
into
higher
performance
aspects
and
real-time
aspects,
you'll
see
that
immediately
you're
gonna
jump
in
and
allow
this
level
of
SLA
and
any
of
those
providers.
I'm
good
and
I'm
willing
to
put
a
personal
bet
on
that.
So.
B
Let
me
question:
Chris
cuz,
I
I
feel
like
I'm,
one
of
the
more
newbie
guys
here
in
this
space
and
I'm.
Sorry,
another
question.
I
know
there
are
some
serverless
frameworks
out
there
that
don't
go
all
the
way
down
to
zero
instances.
They
go
down
to
one,
and
so
you
I
think
you
would
end
up
paying
something,
because
your
application
is
just
sitting
there
waiting
for
like
an
HTTP
request
to
come
in
right.
It
may
be
maybe
minimal,
but
it
is
still
something
there.
B
G
People,
you
know
what
we
we've
done
in
our
solution.
It's
a
customer
can
choose
if
you
want
zero
or
if
he
wants
a
higher
number.
Maybe
even
he
wants
five,
not
even
one
just
because
he
knows
that
personally
becoming
and
wants
to
be
preferred
and
provide
better
silly,
so
you're
saying
decision
that
anyone
that
will
make
a
choice
I
was
serving
the
customer
needs
for
some
new
types
of
application
is
not
serving
us
just
because
of
business.
B
So
trying
to
figure
out
a
way
to
forward
here,
because
obviously,
there's
very
I
feel
he's
a
large
gap
between
these
two,
because
what
and
I'm
trying
to
figure
out
if
there's
some
wording,
he
can
add
to
this.
That
I,
don't
want
to
say,
softens
it,
but
allows
for
perhaps
an
explanation
of
these
different
points
of
views.
B
Or
is
this
impossible
because
because
Chrissy
you're
taking
a
very
firm
position,
especially
with
the
book,
absolutely,
is
there
any
wiggle
room
here
to
add
clarifying
language
that
doesn't
necessarily
invalidate
some
of
the
innovations
that
are
out
there
today
that
the
our
service
and
people
do
claim
or
service?
Well.
I
Again,
I
think
we
run
into,
and
this
is
something
we've
seen
the
industry
for
the
last
couple
years
with
terms
like
cloud
and
DevOps
and
Big
Data
and
stuff
like
that,
is
that
people
are
going
to
apply
a
term
to
a
technology.
Despite
maybe
what
the
rest,
the
industry
sees
it
as,
and
so
I
mean
this.
This
document
can
go
a
certain
direction
that
it
wants
to.
It
may
not
align
with
what
the
major
cloud
providers
or
the
major
you
know,
Gartner's
and
4-5-1
researchers
of
the
worlds.
I
G
G
Is
about
the
zero
if
I'm
forced
to
put
and
not
charge
anything
because
I
want
to
provide
the
service
a
link
to
CERN
application
and
I
think
you
know
I've
been
sitting
with
for
a
lot
of
committees
and
a
lot
of
standardization
bodies
and
and
if
there
is
service
disagreement,
you
try
to
generalize
you're,
not
trying
to
specify
based
on
one's
opinions
in
the
community.
So.
H
It
sounds
like
it's
describing
a
business
model
rather
than
a
technology.
At
this
point
you
know,
I,
don't
think
anybody
thinks
that
there
is
absolutely
zero
cost
sort
of
somewhere
along
the
chain.
It
may
be
the
case
that
you
know
sort
of
Amazon
or
Google
sort
of
absorb
the
fact
that
they're,
paying
maybe
a
micro
amount
for
storing
your
tiny
little
piece
of
code,
but
it
has
cost
at
some
some
place,
unlike
whether
you
absorb
it,
is
more
about
a
business
model
in
the
technology.
H
I
They
to
intertwine
right
we're
talking
about
technology
that
specifically
aligns
to
the
benefits
to
its
core
consumers,
and
so
I
think
to
say
that,
oh
because
a
series
of
vendors
across
the
industry
have
a
capability
and
do
something
behind
the
scenes
that
that
doesn't
impact
the
end
customers
drastically.
That
doesn't
make
any
sense
than
to
talk
about
this
term.
Is
anything
different
from
what's
existed
in
cloud
before
it?
So
would.
B
It
be
useful
to
maybe
split
this
out
and
say
from
an
abstract
definitional
point
of
view.
Serverless
means
no
cost
one
Idol,
however,
in
practice,
because
certain
offerings
or
products
or
you
want
to
call
them,
may
need
to
offer
a
certain
level
of
siela
SLA
or
whatever
want
to
call
it
in
practice.
There
actually
may
not
ever
get
you
down
to
a
complete
idle
state.
Therefore,
your
cost
may
actually
not
be
zero,
but
that
doesn't
necessarily
change
sort
of
the
abstract
state
of
what
service
means
from
a
definitional
point
of
view.
B
I
The
question
is
to
me
to
Christmas.
My
answer
is:
I
still
think
that
it
is
extending
out
beyond
what
one
of
the
main
reasons
are
that
people
look
towards
service
technologies,
and
you
know
the
question
around
things
like
SLA
is
that's
actually
more
of
a
back-end
implementation
type
of
a
thing
versus
it
is
a
customer
decision,
and
so
I
I
disagree
that
that
should
be
a
factor,
there's
things
that
could
happen
behind
the
scenes.
That
would
make
any
SLA
possible
hypothetically.
That
becomes
a
business
decision
right.
B
But
what
I
think
that
does,
though,
is
allows
people
then
say:
okay,
I,
understand
service
should
be
zero
cost
an
idle.
However,
when
I
go
choose
my
particular
provider,
I
need
to
pay
attention
because
not
everybody
is
going
to
be
complete,
zero
cost
for
whatever
reason,
and
so
that
allows
them
to
take
that
into
account
when
they
go
to
evaluate
their
particular
hosting
environment.
B
I
If
you
get
rid
of
that
rule
and
you
have
the
other
rules,
how
do
the
other
rules
in
a
sense
impact?
What
becomes
server
list
so
you
take
away
idle
cost
and
you're
left
with
a
zero
management
platform
for
the
end
consumer
and
then
you're
left
with
the
instant
scalability.
Are
we
really
saying
that
those
things
still
apply
the
same
way
when
you
are
paying
for
resources
that
you're
not
using
well.
B
To
be
clear,
I'm
not
saying
remove
the
no
cost.
What
I'm
saying
is
split
that
number
two
item
into
sort
of
dream:
dream,
state
of
service
and
a
perfect
world
should
be
zero
cost,
but
in
reality
not
everybody
can
or
will
offer
that,
and
so
you
need
to
take
that
into
account.
When
you
make
your
hosting
choice,
I,
don't
even
that
becomes.
B
A
service,
so
let
me
ask
you
a
different
question,
then,
because
I
know
that
a
lot
of
people
who
use
lambda,
for
example,
I,
think
you're
on
mentioned,
this
will
occasionally
ping
their
stuff,
so
they
never
actually
get
down
to
zero
right.
Does
that
mean
that
they're
not
using
your
service
technology
in
a
service
way?
No.
G
It
means
that
Amazon
absorbs
the
cost.
You
say
so.
It's
a
business
model
decision
of
Amazon,
absorbing
the
cost
of
someone's
serve
abusing
their
service,
and
you
know-
and
someone
else
may
not
want
to
have
someone
abuses.
You
know
business
model,
it's
a
business
thing,
it's
not
a
technology
thing,
because
you
do
keep
the
resources
up
in
those
cases,
but
as.
I
I
A
bit
exaggerated
and
a
bit
dramatic
and
I
can
say,
give
you
points
that
I
have
of
the
entirety
of
land
of
running
globally,
that
a
very
small
percentage
of
customers
ever
do
this.
So
again,
let's
not
go
back
to
what
lambda
has
to
do.
Let's
look
across
the
industry
and
all
of
the
major
service
platforms
that
exist
out
there
and
you
see
them
all
talking
about
the
need
for
you
not
to
pay
for
idle
resources.
I
B
So
I
think
what
I'm
going
to
do
here
is
I,
don't
think
we're
gonna
come
to
a
decision
on
this
call
right
now,
but
I
do
think.
This
is
something
we
probably
need
to
hash
through
before
the
white
papers
finished,
so
I'm
gonna
try
to
start
either
an
email
thread
or
something
within
the
doc.
Here
through
comments
and
see,
we
can
come
to
some
sort
of
resolution
here,
but.
B
I
B
Up
the
entire
call
with
this,
because
I
do
wanna
at
least
get
to
the
conclusion
section,
and
we
only
have
ten
minutes
left.
So
we
will
continue
this
discussion
through
some
of
the
means,
if
nothing
else,
next
week's
call,
but
it
but
I,
think
it
is
something
we
have
to
resolve.
So
what
I'd
like
to
do
in
the
last
ten
minutes
is
talk
about
the
conclusion
section
and
then
I
copy
the
link
and
paste
it
into
the
chat.
B
So
there's
the
conclusion
section.
What
I
wanted
to
do
was,
as
over
the
over
last
day
or
so
I
put
together
a
rough
draft
of
what
a
conclusion
section
may
look
like
I
tried
to
distill
out
what
I've
heard
from
other
people
when
people
had
written
in
the
section
before
in
terms
of
ideas
and
I
wanted
to
put
it
out
there
for
people
to
look
at
review
and
throw
stones
at,
but
I
wanted
to
draw
people's
attention
to
it,
because
I
do
think.
This
is
obviously
the
one
of
the
more
important
sections
of
the
doc.
B
L
B
Obviously,
I
wasn't
gonna
try
to
ram
it
down
anybody's
throat
right
now,
I
try
to
see
if
there's
anything
that
jumped
out
is
just
Doug
you're.
So
far
off.
Don't
even
bother
writing
another
word,
a
text
in
here
again
kind
of
thing.
So,
okay,
so
tell
you
what
why
don't
we
take
time
to
review
it
like
I
said
that
my
main
purpose
here
is
just
to
bring
it
up
to
make
sure
people
are
aware
that
something
is
now
in
there
that
I
want
reviews
on,
and
that's
actually
it
for
the
agenda
today.
B
L
A
question
regarding
that
this,
a
concluding
section,
so
it
encouraged
more
service
technology
to
join
the
CSF
right
for
parties
to
see
in
the
open
source
service
section.
So
if
there
are
new
open
short
minutes,
it
should
be
added
to
that
section
right
that.
B
Well,
whenever
we,
whenever
we
claim,
is
done
hopefully
within
the
next
couple
weeks,
but
one
of
the
things
we've
talked
about
is
potentially
taking
this
documents
and
putting
it
in
in
the
github
repo
for
CN
CF
has
marked
down
documents
that
way
it
can
be
more
of
a
living
document.
So,
as
new
open
source
projects
are
created,
people
can
issue
a
pull
request
to
update
that
particular
section
and,
in
particular,
split
this
document
out
into
separate
documents
inside
the
github
repo
that
way
it's
more
easily
consumed.
L
B
Definitely
and
again,
I'll
say
this
again.
I
know
I'm
becoming
a
broken
record,
but
please,
if
you'd
like
to
add
something
the
doc
or
you
all
make
textual
changes,
do
not
make
a
comment.
Add
the
actual
text
itself
to
the
document
as
a
suggestion
that
way,
people
can
know
exactly
what
you
want
to
see
added
and
we
don't
have
to
go
back
and
forth
about
the
exact
wording
that
you're
looking
for
just
add
it
straight
in
there,
and
then
people
can
accept
it
or
reject
it.