►
From YouTube: Serverless WG BoF - Doug Davis, IBM (Any Skill Level)
Description
Want to view more sessions and keep the conversations going? Join us for KubeCon + CloudNativeCon North America in Seattle, December 11 - 13, 2018 (http://bit.ly/KCCNCNA18) or in Shanghai, November 14-15 (http://bit.ly/kccncchina18).
Serverless WG BoF - Doug Davis, IBM (Any Skill Level)
About Doug
Doug works in IBM's Digial Business Group. He's been working on Cloud related technologies for many years and has worked on many of the most popular OSS projects, including OpenStack, CloudFoundry, Docker and Kubernetes. He's currently co-leading the CNCF's Serverless WG, the Cloud Events specification work and the Open Service Broker API WG.
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.
Join us for KubeCon + CloudNativeCon in San Diego November 18 - 21. Learn more at https://bit.ly/2XTN3ho. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.
A
All
right
welcome
everybody
to
the
CNC,
f
service,
working
group,
BOF
or
birds
of
a
feather.
Basically,
we
don't
have
much
of
an
agenda.
It's
really
a
free-form
discussion.
If
you
guys
ask
us
questions
about
what
we're
doing
and
the
serverless
working
group
and
stuff
like
that,
what
I'd
like
to
do
as
a
sort
of
a
starting
point
is:
have
everybody
who's?
A
A
member
of
the
working
group,
basically
introduce
themselves
and
give
their
opinion
in
terms
of
why
we're
even
here
and
what
their
purpose
is
in
the
working
group,
what
they
want
to
get
out
of
the
stuff
like
that?
If
it's
okay
I'd
like
to
start
off,
though,
with
a
sort
of
a
brief
introduction
in
terms
of
what
I
think
we're
doing
here
and
and
how
we
got
to
where
we
are
about
a
year
ago,
the
CNC
F
technical
oversight
committee
decided
they
wanted
to
do
some
investigation
on
service.
A
It
was
still
a
relatively
new
technology
and
they
wanted
us
to
figure
out,
or
they
sort
wanted
a
working
group
to
figure
out.
You
know
what
it
is.
You
know
what
is
service?
How
does
it
relate
to
other
technologies?
Like
function
of
service
or
all
the
other
asses
out
there,
basically
what's
going
on
in
the
community,
and
so
they
started
up
a
working
group
to
investigate
that
and
as
a
result,
our
working
group
produced
a
white
paper
and
a
landscape
document.
The
white
paper
basically,
does
everything
I
said
it
lists
out.
A
You
know
what
is
service?
How
does
compared
to
fast
as
I
as
what
is
the
common
service
framework?
What
are
the
use
cases
for
service
all
the
general
stuff,
as
well
as
some
recommendations
in
terms
of
what
potentially,
the
cnc
has
to
do
going
forward?
And
that's
really
what
we're
talking
we're
gonna
talk
about
next,
but
the
other
thing
we
produced
was
a
landscape
document
which
is
pretty
much
just
a
spreadsheet
of
all
the
various
technologies
out
there
in
the
service
space
they're
doing
something
with
service.
A
So
we
have
open
source
service
projects,
proprietary
service
projects,
toolkits
services
that
interact
or
we
relates
service
functions
all
the
types
of
stuff
just
dump
out
there,
so
people
can
see
what's
out
there,
not
trying
to
say
which
one's
good
or
bad
just
what's
out
there
for
people
that
see
it
and
they
can
start
playing
with
them
themselves.
So
I
guess
I
should
say
my
name
is
Doug
Davis
I
work
for
IBM
and
our
purpose
in
the
working
group
was
really
to
promote
the
idea
of
interoperability
and
portability
right.
A
We
don't
want
people
to
get
locked
into
one
particular
vendor
for
service,
and
so
we're
our
objective
in
the
working
group
is
to
look
for
opportunities
to
create
community
standards
around
these
things,
so
that
people
can
move
their
code
applications
from
one
platform
to
another.
So
if
you
have
well-defined
or
proprietary
event
formats,
proprietary
function
signatures,
that's
going
to
lock
you
into
one
particular
platform,
and
we
want
to
try
to
fix
that.
So
if
you
will
meaning
our
customers
have
a
choice
as
they
move
around.
A
B
Everyone,
my
name,
is
Austin
Collins
in
2015,
I
created
a
popular
open-source
project
called
the
serverless
framework
and
then
went
on
to
found
a
company
called
serverless
inc.
We
owned
the
infamous
server
list
comm
domain,
so
some
of
you
may
know
that,
let's
see,
why
are
we
here?
The
working
group?
That's
a
great
question.
B
You
have
basically
massively
scalable
paper
execution,
compute
resources
available
now,
so
we
think
event-driven
design,
I
t's
been
around
forever,
but
even
more
so
in
the
future
is
going
to
be
incredibly
important
and
events
are
going
to
be
coming
from
all
types
of
sources
and
being
transported
across
environments,
so
events
everywhere
and
then
functions
everywhere.
For
a
long
time,
we
felt
like
everyone's
going
to
go
out
and
create
some
type
of
service
compute
option,
and
that's
what
we've
seen
public
cloud
providers
all
have
a
fast
offering.
B
B
Although
we're
not
supposed
to
say
the
word
standard
within
the
within
the
CN
CF,
so
harmonizing
these
serverless
concepts
so
that
what
what
developers
are
focused
on
is
simply
just
writing
functions
one
way
and
deploying
them
wherever
they
want
to
deploy
them,
whether
it's
in
a
container,
whether
it's
in
some
proprietary
fazz
offering
having
to
shift
their
focus
to
become
platform-specific,
not
really
having
to
learn
kind
of
platform,
quarks
and
all
that
stuff,
so
deploying
functions
everywhere
and
then
of
course,
events.
Wouldn't
it
be
great.
B
I
mean
everybody's
publishing
events
right
now,
but
they're
all
publishing
them
in
different
formats-
and
this
just
like
makes
the
developer
experience
really
frustrating.
You
have
to
go.
Relearn,
what
event
for
my
people
are
using
and
so
what,
if
event,
data
had
a
consistent
format
or
a
consistent
envelope,
so
you
could
always
see
you
always
know
as
a
developer,
where
to
check
for
the
event
type
the
event
version
you
can
start
putting
in
tracing
information,
so
you
know
where
the
event
has
been.
B
Maybe
you
could
put
in
security
rules,
so
you
can
know
where
the
event
should
not
go
all
that
stuff
needs
to
be
standardized
or
harmonized,
and
then
you
know,
maybe
we
should
look
into
plumbing.
It's
like
how
do
we
actually
start
rigging
up
entire
systems,
functions
and
event,
Riv
and
workflows
and
stuff,
and
how
do
we
do
that
in
a
standard
way?
So
all
that
stuff
I
think
there's
plenty
of
opportunity
for
us
to
harmonize
these
concepts
and
there's
no
better
place
to
do
that
than
the
CNC
F
right
now.
So
that's
why
I'm
here?
C
So
a
few
of
us
have
been
involved
in
the
service
working
group
because
we're
my
team
is
building
the
infrastructure.
The
with
that
creates
the
events
that
drive
Google
Cloud
functions,
and
what
we're
seeing
is
that
these
are
used
for
much
wider
breadth
of
use
cases
than
traditional
messaging
products
and
regular
old
application
developers
who
make
web
applications
are
routinely
running
into
the
complexity
of
a
distributed
system.
It's
not
okay
and
you
want
to
just
have
a
web
application
that
just
has
your
person's
data.
C
C
D
D
Our
use
of
our
we
reason
for
being
in
a
server
Louis
working
group
is
all
around
the
same
aspects
as
IBM.
We
want
portability
of
services
of
functions
across
services
we
won't
be
able
to
as
service'
be
able
to
have
our
customers
be
able
to
use
any
cloud
write.
The
write
the
function
once
run,
many
and
I'd
say
that
that's
one
of
the
areas
that
as
we're
seeing
containerization
and
functions
come
about,
is
where
customers
are
having
the
most
problems,
they're
trying
to
take
their
monoliths
applications
and
convert
them
into
containers
and
functions.
D
It's
problematic
as
they're
writing
these
new
functions
I
really
want
them
to
be
able
to
only
have
to
write
them
once
and
not
for
once
for
every
cloud,
and
so
that's
part
of
the
reason
that
we
want
to
look
at
this
harmonization
be
able
to
to
work
with
the
industry
with
everyone
in
this
room
to
understand
how
we
can
develop
a
good
set
of
specifications
to
harmonize
things.
So.
E
E
Let
me
tell
you
both
those
things,
so
my
name
is
Clemens
masters
I'm
here
with
my
two
colleagues
Justin
Conway
and
Baram
bani-sadr,
and
we
are
messengers,
an
azure
we
own
about
your
services
that
have
to
do
with
good
old
messaging
as
well
as
new
messaging
and
we've
been
we've
built
out
a
platform
over
the
last
two
years.
Really
that
is
the
the
substrate
underneath
the
outer
platform
that
makes
the
Azure
platform
event-driven
and
out
of
that
effort.
So
now
the
storage
team
can
go
into
mid
events
about
things
that
are
happening
in
their
system.
E
A
Remember
you
during
the
intro
did
I
mention
cloud
events.
I
know,
I
talked
about
the
server's
workgroup,
it's
up,
I,
don't
think
I've
ever
been
to
cloud
events.
Did
I
so
I
did
okay
before
we
keep
going
because
people,
if
I,
keep
talking
about
inventing
some
of
you
guys
who
knew
the
group
may
not
understand
why
so
one
of
the
outputs
of
service
working
group
was
recommendations.
A
I
know,
I
said
that,
but
one
of
the
recommendations,
Euron's
next
I
think
was
to
focus
around
harmonization
around
the
events
about
transferred
between
the
various
components,
and
so
we
started
working
on
this
cloud
event
specification
which
we
just
released
version
1.0,
I'm
sorry
0.1
and
as
clement
alluded
to,
we
have
an
HTTP
and
adjacent
mapping
defined
for
those.
So
that
was
our
first
attempt
is
some
sort
of
harmonization
effort
still
very
basic
alpha
level,
but
please
take
a
look
at
it.
D
A
F
H
Hi,
my
name
is
kathy
drunk
I'm
leading
the
service
party
at
huawei
embassy
in
Silicon,
Valley
I.
Think
my
goal
of
this
is
to
work
in
this
workgroup
is
to
make
the
developers
life
easier,
I
think
when
aspect
of
that
is
to
if
we
can
have
a
consistent
and
generic
model
for
the
user
to
configure
its
server.
This
application
model
like
to
specify
you
know
the
event
and
function
and
the
relationship
between
event
and
function
so
I
think
that's
when
I
think
I'm
interested,
of
course.
Well,
that
is
the
event
cloud
events,
the
format.
Let's.
I
I
Guess,
as
we
talk
about
events,
one
of
the
things
that
it's
still
very
I'd
say
undefined
in
service
right
now
is
the
application
module,
because
now
you
have
like
all
these
functions
and
all
these
events,
but
you
still
don't
have
like
a
well-defined
established
way
to
package
one
application
that
is
composed
of
like
multiple
functions,
for
example.
That's
one
thing
that
I'm
very
interested
in.
Hopefully
we
get
to
that
after
we
nail
the
event
in
part.
J
Need
very
much
I'm,
also
interested
in
interoperability.
Of
course,
I
think
you
brought
up
an
interesting
point,
so
several
is
is
much
more
it's
evolving
rapidly,
so
I'm
also
in
the
group
to
see
you
where
else
it's
heading
and
I
mean
apart
from
functions
and
events,
and
so
how
this
whole
approach
and
architectural
style
may
evolve.
L
Hi
I'm
Chad,
Inari
Mira
from
Oracle,
formerly
with
iron
IO,
and
we're
doing
the
FN
project,
I
think
a
lot
of
what
I
agree
with
everything
that's
been
said
so
far.
One
thing
that
I'm
really
interested
in
is
seeing
that
the
that
server
list
doesn't
go
from
this
high
peak
cycle
down
to
the
trough
of
disillusionment
and
stay
there,
because
I
think
there's
a
challenge
with
server
list
that
it
hasn't
really
delivered
on.
I.
L
Think
all
the
promises
out
there
right
now,
which
is
a
big
part
of
the
hype,
and
it's
really
interesting,
to
see
a
group
of
diverse
companies
that
don't
always
get
in
the
same
room
to
talk
about
what
the
future
looks
like
and
we
how
we
can
try
and
avoid
that
together.
So
that's
a
big
part
of
the
reason
why
I'm
there
yeah.
L
A
A
M
A
M
Thank
me
for
that
that
we,
you
know
there
are
others
that
you,
mostly
others
that
are
represented,
have
some
kind
of
a
dog
in
the
fight,
so
to
speak,
or
have
a
bit
more
skin
in
the
game
is
maybe
a
better.
You
know
way
of
putting
that.
I
would
say
that
interoperability
is
certainly
core
to
the
value
that
SolarWinds
provides
or,
or
maybe
benefits
from
to
the
extent
that
it's
third-party
to
a
lot
of
things,
we
do
provide
some
amount
of
monitoring
for
functions
that
folks
run.
A
M
A
A
If,
from
the
outset,
are
there
any
questions
you
guys
have
about
the
working
group
in
general
or
about
the
cloud
events
or
what
we're
doing
otherwise,
I
may
have
to
start
randomly
picking
ideas
out
of
my
head
and
that's
always
a
scary
thing,
any
questions
so
far
based
upon
what
you've
heard
you
guys
are
gonna
kill
me.
Okay,
in
that
case,
let
me
let
me
do
oh
go
ahead,
sir.
K
K
K
J
N
So
that's
a
really
interesting
one,
because
I
think
when
I
looked
at
the
service
white
group,
when
I
first
joint
way
back
last
year,
mark
invited
me
I
saw
this
chat,
ops
there
and
I've
not
really
thought
about
it,
and
then
it
connected
with
an
idea
that
I
had
on
open
fires.
We
had
all
of
these
contributors
coming
in
on
github
and
I,
wanted
to
give
them
fine-grained
access
permissions
so
that
they
could
comment
on
on
things
and
then
a
bot
could
do
stuff,
instead
of
giving
them
full
write
access.
N
So
you
guys
should
fix
your
permissions,
but
I've
created
this
bot.
That
means
I
can
take
anyone
like
panels
who
sat
here.
That's
contributed
to
open
fires,
giving
permission
to
open
closed
issues
to
merge,
pull
requests
tree
artery
of
labels,
and
that's
all
done
with
an
open,
fast
functioning
go
so
it
just
gets
JSON
from
github
verify
the
H
mark,
and
it's
a
really
nice
integration
with
github
amps.
N
Taking
that
a
step
further
and
started
a
little
project
called
open,
fast
cloud.
The
idea
is
that
you
can
deploy
a
bunch
of
functions
anywhere
on
top
of
open
files
and
then,
when
github
gets
a
push
notification,
it
will
tell
us-
and
you
end
up
with
a
Heroku
style,
HTTP
endpoint
with
let's
encrypt
so
I-
think
it's
a
super.
It's
a
super
cool
use
case
and
there's
so
much.
We
could
do
for
automation,
so
you
should
check
it
out.
O
Hello,
I'm
Panos,
I
work
for
SUSE,
Linux
I'm,
a
QA
engineer
and
I
was
recently
looking
open.
Thousands
serverless
and
basically
one
of
the
use
cases
is
that
since
we
have
many
departments,
many
people
working
in
their
own
different
clouds,
many
people
have
their
own
scripts.
Doing
the
same
thing
using
serverless,
basically
is
a
way
to
unify
so
I
can
access
somebody
else,
functionality
without
really
having
to
know
how
it
works
in
the
in
again.
They
just
have
a
question
and
I
have
the
answer.
That's
it,
and
another
second
use
case
is
a.
O
We
are
looking
into
how
we
can
use
it
for
having
really
fast
testing
in
that
case,
for
the
products.
Just
to
give
you
a
very
quick
example:
I
assume
you
everybody's
familiar
with
this
webpage
like
is
it
offline
or
it
just
me?
So
imagine
something
like
since
we
see
a
distribution
that
case
we
would
like
to
have
the
knowledge
if
our
packages
are
installable
in
that
case.
P
Hi
yeah
first
regarding
the
server
last
application
sharing
AWS
did
sir
I'm,
not
with
AWS
I.
Do
consulting
AWS
said
the
server
list
repository
application
thing,
which
is
a
great
step.
It's
kind
of
like
the
same
thing
as
cube.
Apps
I
think
that's
something
that's
worth
looking
into
now
regarding
workflows,
I
do
consulting,
and
the
two
biggest
issue
I
see
are
a
lot
of
companies
use
server,
less
land
I,
especially
to
automate
infrastructure.
P
They
do
smaller,
one-time
event
related
stuff
or
creating
users
and
stuff
like
that,
and
they
have
a
lot
of
smaller
projects
like
this
and,
more
often
than
not
I
get
into
a
huge
mess
of
functions
and
no
idea
how
how
to
do
and
how
to
manage
a
large
number
of
smaller
functions
that
happen.
We've
seen
again
no
precedent
and
this
and
no
way
to
organize
it
and
there's
still
a
lot
of
issues
on
that.
The
second
use
case
I've
had
clients
who
are
have
applications.
P
That
would
be
great
as
event-driven,
but
still
since
there
is
no
obvious
or
direct
way
to
do
to
build
the
full
application.
That's
server
less.
There
are
some
framework
like
chalice,
which
is
in
Python
and
so
forth.
Zappa
if
I
remember
correctly,
there's
a
lot
of
them,
but
there
are
still
not
comfortable
doing
a
full
application
in
server
less
until
there
are
some
best
practices
or
more
documentation
on
this.
So
definitely
more
examples
and
workflows,
especially
around
organizing
functions
and
projects.
Okay,.
A
So
the
massive
question
is,
as
I
said,
one
of
the
recommendations
was
to
produce
a
specification
around
the
metadata
for
events,
so
you
get
some
commonality
there.
The
other
thing
we
talked
about
in
the
working
group
was
looking
at
the
function
signatures
and
trying
to
get
sitting
there
out
billy
around
that
I'm
curious
to
know
from
people
out
there
using
this
stuff,
who
actually
maybe
tried
different
platforms,
are
those
the
two
high
hitters
for
you.
A
Are
there
other
areas
of
harmonization,
you'd
like
to
see
us
tackle
or
look
at
or
where
your
pain
points
when
you
have
to
try
to
move
your
code
from
one
platform
to
another,
or
has
it
not
been
an
issue
for
you
at
all
and
we're
just
making
stuff
up?
Anybody
want
to
comment
on
trying
to
port
things
around.
Q
I'm
Marco
from
adidas,
so
basically
we
built
our
data
platform
on
AWS
using
just
lambdas.
So
basically,
everything
is
built
by
events
and
data
scientists
built
basically
they
built
their
code
they
commit,
and
after
that
is
trigger
of
the
process,
to
create
the
EMR
cluster
whatever
right.
So
the
problem
is
that
now
we
are
looking
in
another
second
provider
yeah,
but
the
problem
is
that
a
lambda
is
written.
All
the
brand
lambdas
are
written
in
Python
right.
So
this
is
standard.
Then
Python
is
nothing.
Q
So
we
are
looking
into
Google,
for
example,
or
the
providers,
but
basically
you
miss
all
the
underlined
infrastructure.
So
what
I
was
expecting
that
we
have
like
kind
of
API
calls
that
if
I
need
just
a
storage
or
and
I
do
framework
I
just
call
an
API
from
a
Python.
So
I
have
the
full
CI
CD
to
chain
prepared
directly
from
the
data
scientist
and
wherever
is
the
the
the
other
framework.
They
can
just
call
it
yeah.
So
this
is
really
an
event-driven
platform.
So.
A
N
So
that
sounds
really
interesting.
The
problem
that
you
have
about
Python
and
the
signature
being
the
same
everywhere,
I
think
some
of
the
open
source
projects
that
are
represented
here
could
solve
that
very
easily
as
long
as
they
covered
the
other
things
that
you
need.
There's,
obviously
just
one
format
you
deploy
cublas
nuclear,
open
fires,
wherever
you
want
it's
the
same
signature
right
and
an
iron
as
well,
so
I
think
that
could
solve
that
problem.
But
there
are
probably
things:
you'll
miss
from
lambda,
but
one
thing,
I,
don't
think.
N
You'd
miss
from
lambda
is
probably
like
this
60
Meg
binary
limit.
If
you're
doing
data
science,
how
are
you?
How
are
you
even
doing
that
with
a
60
mega
man,
because
as
soon
as
I
had
people
in
universities
or
in
actual
companies
do
they
need
gigs
right,
550
max
not
enough?
So
what
are
you
doing
about
that?
Wait
given.
Q
Actually,
we
are
trying
to
move
the
data
scientists
to
use
like
standard
CIC
to
to
Train
right
because
actually
the
data
scientists,
they
are
using
just
simple
our
code
yeah,
so
we
are
moving
them
to
Python.
This
is
another
step
that
we
are
doing
and,
of
course,
like
this
limiter
made
on
lambdas
is
something
that
will
come
so
actually
we
build
all
the
platform
that
is
driven
by
lambdas
and
yeah.
Q
F
F
E
What
one
quick
comment,
our
code
us
since
that
she
that
just
came
up
with
that
code
limit
for
me
services
is
some
people
making
decisions
for
you
around.
You
know,
operating
the
infrastructure,
and
also
doing
so
such
is
such
in
the
way
that
it
works
not
for
only
for
you,
but
it
works
for
very
many
people
and
then
also
doing
so
in
a
way
that
it's
profitable
and
so
we're
finding
that.
Sometimes
we
impose
quotas
that
seem
weird
for
certain
customers.
E
You
know
if
you
have
special
workloads
where
you
need
to
have
binaries
that
are
several
gigabytes
large,
then
you're,
probably
no
longer
a
customer
for
the
server
this
workload
that
can
be
run
that
can
be
run
in
a
general-purpose
public
cloud
with
shared
infrastructure,
then
you're,
probably
more
in
a
place
where
you
can
run
something
that
is
dedicated
for
you,
where
you
can
load
that
up
and
then
you
can
also
run
it
in
the
phone
in
land
stay
away.
So
there's
options.
If
you
go
different
platforms.
What
what's?
E
What
I
think
was
important
is
that
the
interfaces
to
those
infrastructures
can
be
the
same
so
that
you
can
go
and
put
little
functions
and
no
matter
you
don't
care
whether
the
functions
run
or
you
have
a
scenario
where
you
actually
control
the
play
to
deployment,
but
you
can
go
and
still
use
the
same.
Api's
and
harmonization
will
help
with
that,
but
the
quotas
quotas
are
for
someone
who
runs
public
cloud
services
at
you
know
trillions
of
her
of
events
per
per
day.
It's
important
to
have
quotas
in
place.
I
P
I
I
A
A
H
Has
anyway
used
AWS
the
step
function,
okay,
I
think,
okay,
I,
think
III
know
he
raised
a
good
question.
I
think
you
know
we.
We
need
a
way
to
organize
the
functions.
A
typical
application
is
you're
in
not
just
a
simple
function
or
a
simple
event:
Cheeto
one
function,
it
might
be.
You
know
a
combination
of
events
triggers
combinational
functions,
which
sequence
or
in
parallel,
or
maybe
you
know,
switching.
You
know
it
depends
on
the
function,
execution,
result
and
branch
to
different
next
function.
H
A
B
A
A
B
Off
I
forgot
to
introduce
my
colleague,
matcha
he's
our
lead
architect
on
our
event:
gateway
product,
it's
a
pretty
awesome
event:
router,
just
routes,
event,
data
faz
on
any
cloud
and
on
any
service
platform,
but
I
want
to
say
great
job
everyone.
At
the
beginning
of
this
year
we
started
our
the
first
effort
to
harmonize
one
of
these
concepts.
That
was
cloud
events.
We
just
got
the
first
version
out.
We
really
said
like
two
weeks
ago:
I
think
something
like
that.
B
Yeah
I,
honestly
think
of
what
we're
doing
just
standardizing
that
event
format
is
bigger,
could
be
bigger
than
serverless
itself.
I
think
it's
gonna
have
a
pretty
big,
interesting
impact
on
IT
in
general,
and,
if
you
don't
believe
me,
you
should
watch
Kelsey
Hightower's
talk
in
about
an
hour
here
and
if
anyone
who's
looking
for
validation
of
like
the
work
that
we've
been
doing
over
the
past
few
months,
check
out
that
talk,
it's
gonna
be
pretty
awesome
right.
C
So
firebase
has
a
persistent
connection
between
the
client
and
server
and
I'll
be
going
over
some
of
those
use
cases
in
my
talk
as
well
as
service
compute
ones,
and
what
I
think
the
majority
of
the
audience
thinks
of
is
server
list,
but
I.
Consider
these
things
really
the
same
thing
and,
as
you
think,
about
eventing.
This
kind
of
you
know
under
the
hood-
maybe
web
sockets,
but
how
you
transmit
events
is
sort
of
independent
of
the
events
themselves.
A
With
that
I
think
we're
technically
out
of
time
101.
The
last
thing
I
want
to
say
is
that
what
is
it
8:20
this
evening
we
actually
have
a
working
group
meeting
here
on
site.
If
you
look
for
it,
it's
called
CN,
CF
cloud
events
working
group
face
to
face,
so
you
can
actually
see
us
not
just
sing
Kumbaya
like
we're
doing
now,
but
actually
argue
a
little
at
8:20
this
evening,
I'm
sure
of
the
room,
but
you
can
find
it
on
the
schedule
mark.
Since
you
got
the
magazine,
something
you
wanted
to
say.
Yeah.
D
I
was
just
gonna
fall
on
about
persistent
connections,
which
is
that's
an
opportunity
to
use
some
of
the
open-source
faz
as
being
able
to
deliver
different
functionality.
You
know
not
having
to
worry
about
limits,
be
able
to
have
persistent
connections
wherever
you
need
them
and
not
have
your
functions
killed
after
five
minutes
or
you
know,
however
long
so
you
know
it
is:
there
need
for
being
able
to
run
open-source
projects
such
as
open
files,
riff,
etc.