►
Description
Join us for Kubernetes Forums Seoul, Sydney, 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
The Serverless Landscape and Event Driven Futures - Dee Kumar, CNCF
Serverless design patterns have grown in popularity amongst developers and enterprises alike and the ecosystem is exploding. Developers like moving faster by focusing on business logic without worrying about the underlying infrastructure. Today, there are umpteen solutions and OSS projects in the market and the space needs some organization to maximize effort. There is a lot of curiosity and confusion around serverless computing. What is it? Who is it for? Is it a replacement for IaaS, PaaS, and containers? Does that mean the days of servers are over? The CNCF created the Serverless Working Group to explore the intersection of cloud native and serverless technology. The first output of the group was creation of serverless landscape. The landscape lists some of the more common/popular Serverless projects, platforms, tooling, and services.
https://sched.co/MPeI
B
B
You
know,
let's
look
at
the
bot
top-left
first
of
all,
serverless
does
have
servers,
but
those
are
not
the
ones
that
you
manage
from
your
perspective,
you
build
your
application,
you
deploy
your
application
and
then
you
don't
know
know
who's
managing
those
servers,
so
you're
really
focusing
on
the
application.
That
is
your
core
competency.
That
is
what
you
excel
in
and
let
somebody
else
manage
your
service.
So
there
is
no
managing
of
servers.
There
is
no
provisioning
of
servers,
no
patching
of
servers.
B
None
of
that
has
to
happen
from
your
side
of
it
go
on
the
top
right.
You
are
given
a
knob,
a
dial
to
run
your
application
and
if
you
want
to
scale
your
application,
if
all
of
a
sudden,
your
application
becomes
popular
and
instead
of
like
10
requests
per
second
you're,
getting
thousand
requests
per
seconds,
your
app
automatically
scales-
you
don't
have
to
scale
it.
There
is
no
cluster
provisioning
that
needs
to
happen
behind
the
scene.
You
know
that
has
to
happen
automatically,
for
you
then
go
on
the
highly
available
side
of
it.
B
So
we
going
in
a
clockwise
direction.
The
app
is
highly
available.
I
mean
I
work
at
AWS
at
AWS.
We
have
this
concept
of
a
region
which
is
a
physical
location
around
the
world,
and
that
region
has
typically
multiple
availability
zones.
So
you
can
architect
your
application
to
be
highly
available,
but
then
you
need
to
make
those
calls.
B
Let's
talk
about
some
of
the
patterns
that
we
have
seen
across
our
customers
own.
What
does
it
really
mean
so
on
the
bottom
left?
What
you
see
is
basically,
where
your
typical
classical
monolithic
applications
are
setting
over
there.
So
we
look
at
you
know
how
customers
are
migrating.
What
they
want
to
do
is
they
want
to
get
to.
They
want
to
get
to
the
top
right
quadrant
here,
so
the
first
set
of
customers.
What
we
see
is
they
lift
and
shift
the
application
into
the
cloud.
B
And
then
there
are
customers
like
Fannie
Mae,
which
is
an
Amazon
customer
who
just
say
no,
what
we're
gonna
go
full
blow
into
server
less,
so
we're
gonna
start
away
right
away
start
using
lambda
functions,
deploy
everything
into
the
cloud
actually
build.
Our
application
said
that
it
is
micro,
services,
oriented
and
everything
is
even
driven,
so
you
store
something
in
a
queue
or
a
notification
or
API,
and
then
I'm
gonna
call
that
lambda
function.
So
that's
the
common
pattern
that
we
have
seen,
then.
B
The
second
pattern
is
where
we
have
seen
where
in
the
IT
business
typically
what
you
do
is
you're
doing
one-off
jobs,
and
you
know
that
job
runs
once
a
day
or
maybe
at
a
certain
point
of
time.
You
don't
want
to
manage
your
cluster
for
that.
So
in
case
you
don't
want
to
manage
your
cluster.
For
that,
that's
exactly
where
lambda
fits
in.
So
you
can
say
hey
by
the
way.
I
just
want
to
run
this
function
once
hour
once
every
six
hours.
B
So
why
pay
unnecessary
cost
for
that,
so
it
automatically
just
triggers
that
lambda
function
and
then
runs
it,
and
then
you
get
it
back
now,
as
you
are
moving
along
in
that
you
realize,
oh
by
the
way,
this
job
is
actually
getting
done
in
a
lot.
Lesser
cost
lot
more
efficient,
highly
available
I'm
only
paying
for
my
execution
time.
That's
where
it
starts
becoming
strategic,
your
CT
or
your
CI.
You
kind
of
starts
looking
at
it,
oh
how
about,
instead
of
just
doing
in
a
job
in
an
IT.
B
Let's
take
a
look
at
the
entire
department
or
an
entire
business
application
and
see
we
can
start
converting
this
into
sort
of
the
serverless
landscape,
part
of
it
and
that's
where
you
start
getting
looking
at
agility
and
how
rapidly
I
can
deploy
and
deliver
value
to
my
customers
and
the
third
pattern
that
we
have
seen
is
called
as
a
strangler
pattern.
Now
strangler
wines,
you
know
this
is
a
pattern
that
Martin
Fowler
has
written
about.
B
Strangler
wines
are
the
wines
that
he
found
at
the
coast
of
Australia
with
his
during
his
one
of
his
vacation
with
his
wife.
Strangler
wines
really
go
on
the
tree,
they
start
from
the
top
and
then
they
go
down
to
the
root
and
then
they
start
growing
on
the
tree
itself,
where
they
become
the
host
and
the
tree
disappears.
That
is
a
pattern
that
we
have
commonly
seen
in
our
customers,
where
they
start
with
the
monolithic
application,
the
chip,
a
part
of
the
monolithic
application.
B
They
create
an
API
to
it
and
make
a
micro
service
out
of
it
and
then
the
chip,
another
part
of
it
and
another
part
of
it
and
so
on
and
so
forth.
Until
the
monolith
itself
is
kind
of
decomposed
into
multiple
applications
and
potentially
each
of
those
micro
services
are
being
run
as
a
server,
less
function
or
a
container
or
whatever
way.
You
want
to
call
it.
But
it's
a
true
micro
services
oriented
pattern
that
we
have
seen
in
our
customers
being
used
all
over
the
place.
So.
A
Last
year
new
stack,
they
did
a
survey
and
five
hundred
people
responded
to
the
survey.
So
this
essentially
is
the
response
that
they
receive
and
basically
what
the
shows
is
fast
is
pretty
much
for
developers.
It
changes
the
way
developers
are
creating
and
building
applications,
and
based
on
that,
this
essentially
shows
the
you
fast.
As
you
can
see,
here
are
the
top
10
different
vendors
and
again
fast,
is
just
one
part
of
the
service
stack
and
it's
not
just
not
the
whole
server
story
in
itself
and
AWS,
lambda
I
think
from
a
usage
perspective.
A
B
I
mean,
as
you
rightly
said,
you
know
fast.
You
know,
people
are
people.
Often
confuse
it
is
fast.
Civilized
fast
is
just
a
compute
part
of
server
less,
and
if
you
realize
the
title
of
the
talk
is
event-driven
future's,
so
fast
is
just
a
compute
part
of
it
where
you're
taking
a
function,
and
somebody
else
is
managing
that
function
for
you
in
terms
of
the
deployment.
But
if
you
look
at
oh
go
back
slightly,
so
if
you
look
at
AWS
lambda
AWS
lambda
is
a
foundational
service
for
Amazon.
B
The
service
was
launched
in
2014,
roughly
90
to
95%
of
our
roadmap
is
driven
by
customers
and
then
five
to
ten
percent.
You
know
where
we
think
that
oh
this
is
what
customers
might
need.
So
AWS
lambda
is
a
classical
example
on
how
we
thought
the
customers
might
need
it
and
seems
like
they
loved
it.
So
it's
a
foundational
service.
It's
run.
That
means
it's
run
on
all
AWS
region,
anytime.
We
launch
a
new
region.
The
service
starts
over
there,
a
classical
use
case.
A
customer
use
case
I
like
to
quote
home.
B
There
is
FINRA
FINRA.
Is
financial
regulation,
a
nonprofit
that
basically
99%
of
equity
trade
in
u.s.
goes
through
FINRA
and
65%
of
option?
Trades
goes
through
FINRA
and
the
response
time
for
FINRA
is
super
critical
because,
based
upon
the
rules
that
they
run
on
each
trade,
they
need
to
quickly
say
that
if
the
market
is
good
to
go
or
the
market
needs
to
shut
down,
it
can
actually
impact
the
US
stock
market
before
migrating
to
AWS
lambda.
B
They
were
taking
about
three
minutes,
average
response
time
and
seven
minute
peak
response
time
when
they
migrated
to
lambda.
They
were
consistently
getting
a
minute
of
risk
time
and
the
scale
we
are
talking
about
for
AWS
lambda
is
they're
running
about
half
a
trillion
that
is
500
billion
validations
per
day,
so
I
would
encourage.
You
know
if
you're
looking
at
that
scale.
Take
a
look
at
that
case
study
how
it
talks
about
why
AWS
lambda
is
so
popular
and
how
customers
are
using.
It
now
talk
about
the
developer
experience.
B
It's
super
important
that
the
languages
that
matter
to
you
are
supported
in
lambda.
So
in
lambda,
for
example,
you
know
a
bunch
of
languages
are
supported
out
of
the
box,
so
there
is
no,
there
is
Java,
there
is
Ruby,
it
is
go.
There
is
PowerShell,
you
pick
your
language
favorite
language
and
is
supported
over
there
last
year
at
AWS
reinvent.
We
also
introduced
something
collars,
bring
your
own
language
and
basically
what
we
introduces
a
new
lambda
or
runtime
api
and
using
that
lambda
runtime
api.
We
open
sourced,
c++
and
rust
runtime.
B
So
now
developers
have
written
C++
functions
that
can
be
managed
as
a
lambda
function
or
rust,
and
we
have
lots
of
examples
by
partners.
So
COBOL
is
back,
PHP
is
back,
Fortran
is
back
if
you
are
any
of
the
fanboys
for
any
of
those
languages
for
Tron
used
to
be
my
college
language,
so
those
custom
runtimes
can
actually
be
now
managed
in
the
cloud
for
you.
A
So,
let's
talk
about
service
within
the
cloud
native
computing
foundation,
so
just
to
get
the
community
involved,
the
server
list,
there's
a
working
group
that
was
created
within
services
and
then
the
goal
for
this
working
group
was
to
figure
out
the
intersection
between
cloud
native
and
servers,
and
so
the
outcome
of
the
other
work
that
came
out
of
the
service
working
group
was
essentially
a
white
paper.
So
CN
CF
sponsored
the
white
paper
and
was
authored
by
a
few
specialists
from
new
stack
and
this
white
paper.
A
If
you
guys
haven't
seen
it
I
would
highly
encourage
that
you
read
the
white
paper.
It
really
is
a
has
a
lot
of
recreate
content
just
in
terms
of
breaking
down
the
whole
stack
of
servers,
and
it's
not
just
limited
to
fast,
because
that
essentially
is
somewhat
the
majority
of
the
understand
today
and
the
other
artifact
that
came
out
of
the
work
from
the
surveillance
working
group
is
the
service
landscape.
A
The
most
of
you
might
have
known
about
the
CNC
F
landscape,
so
the
CNC
F
landscape
is
essentially
to
represent
all
the
cloud
native
technologies
that
exists
and
today
the
landscape.
There's
a
joke
that
goes
around:
it's
called
hellscape
because
we
represent
close
to
600
products
or
projects
within
the
landscape,
and
then
the
civiles
landscape
is
a
subset
of
it.
It's
essentially
to
create
more
awareness
for
folks
were
not
involved
with
serverless
or
just
new
to
server
less
and
just
to
understand
all
the
different
companies
and
products
and
projects
that
belong
to
the
serverless
landscape.
A
I
also
want
to
give
out
a
shout
out
to
Andre
I'm,
not
sure
if
you're
here,
hey.
Thank
you
so
he's
the
rock
star
who
actually
builds
the
service,
landscape
and
I
think
what
is
magical
about
the
landscape
is
I
would
say
three
years
ago.
The
landscape
was
very
static
in
nature.
So
if
there
was
someone
a
company
wanted
to
represent
or
put
a
logo
and
represent
all
the
details
into
the
landscape,
it
was
very
manual
and
it
kind
of
limited
participation
from
the
community,
and
so
now
the
landscape
has.
A
This,
essentially,
is
how
the
landscape
has
grown,
where
I
talked
about
the
five
or
six
hundred
different
companies
that
exists,
and
today,
due
to
the
work
of
Andre
I.
Think
is
a
lot
of
enabled.
What
I
would
say
collaborative
editing,
it's
very
dynamic
in
nature.
We
do
get
a
lot
of
requests
in
terms
of
how
do
I
represent
or
how
do
I
add
my
product
to
the
landscape.
It
is
super,
easy,
I
know,
all
of
you
are
developers
and
github
is
something
you
must
be
using
all
the
time.
So
this
should
be
extremely
easy.
A
There
is
a
llamo
file
where
all
you
need
is
to
follow
some
really
strict
specifications
for
lines
is
all
we
ask.
As
you
enter
your
company
name
URL,
we
do
ask
for
a
CrunchBase
URL,
because
that's
where
we
get
all
the
financial
stat
stats
and
then
we
also
ask
that
the
logo
should
be
in
SVG
format,
and
the
for
that
is.
This
logo
is
a
help
scale
across
different
modern
browsers.
A
So
you
don't
have
to
worry
about
pixelation
or
anything
of
that
sort,
and
the
other
thing
you
can
do
is
you
can
fork
your
own
branch
from
the
master
branch.
You
can
try
it
out.
You
could
also
look
at
it
from
staging
server.
We
use
neckla
5,
which
spins
up
a
staging
server,
where
you
can
actually
see
how
your
product
or
project
is
represented
in
the
server
list
landscape
and
then
the
last
step
is,
if
everything
looks
good
to
you.
You
write
a
comment
and
then
you
push
that
request
and
then
someone
from
C&C
F.
A
We
just
look
at
it
to
make
sure
that
it's
represented
correctly
and
then
the
product
actually
shows
up
in
the
landscape.
And
if
there
is
time
today,
I
think
it's
really
simple,
but
I
came
happy
to
share
a
demo
in
terms
of
how
you
can
actually
represent
your
company
your
product
into
a
very
specific
category
and
how
you
could
just
do
that
on
the
fly
and
I
can
happily
take
you
through
that,
and
so
with
that,
so
I
will.
Can
you
talk
about
exactly
what
the
different
stacks
are
in
server
less
and
then
how?
B
Sure
so,
if
you
look
at
it,
you
know
look
at
different
cloud
providers,
you
know
they
have
their
own
storage,
they
have
their
own,
you
know
API
gateway,
they
have
their
own
functions.
So
if
I
look
at
say
AWS,
for
example,
there
is
s
3,
there
is
API
gateway.
That
is
lambda.
Now
the
idea
is,
you
know
we
talk
about
the
event-driven
future,
so
the
whole
aspect
is
where
the
functions
are
really
the
gluing
between
all
of
these
different
services.
So
something
happens
in
the
storage.
B
You
call
a
function,
something
happens
in
the
database,
you
call
a
function,
something
happens
in
the
API
management.
Somebody
is
calling
you
a
REST
API,
you
call
a
function
and
then
you
return
the
response
back.
So
that's
sort
of
the
whole
idea
between
different
but
pick
a
cloud
that
you
want
to
run
it
now.
Let's
say
you
want
to
have
this
up
and
running
across
multiple
clouds
and
what
up?
B
But
by
that,
what
I
mean
is,
let's
say
somebody
is
storing
an
image
in
a
Google,
Cloud
storage
and
they
want
to
run
a
roast
lamb
function.
You
know
in
AWS
cloud
or
somebody
is
storing
an
image
in
s3,
for
example,
and
I'm
going
to
call
as
your
function.
How
does
that
work?
Basically,
so
that's
that
start
of
the
missing
piece
right
now,
because
within
the
cloud
of
course,
it
works
very
well
and
that's
sort
of
the
whole
premise
of
cloud
events
and.
A
So
now
that
we
have
the
service
working
group,
the
TOC,
which
is
the
technical
oversight
committee
of
who
are
a
part
of
C
and
C
F
or
represent
CN
CF.
Essentially
what
they
have
done
is
now
they've
created
a
C
bank's
type
of
working
group
to
further
along
what
our
own
was
saying
is
to
enable
events,
a
platform
or
I
would
say
more
like
a
specification,
because
we
all
live
in
a
multi
cloud,
multi
service,
type
of
environment.
We
talk.
A
We
have
to
make
sure
that
there
is
a
common
standardized
way
or
specification
of
events
that
B
that
is
being
able
to
talk
to
different
cloud,
different
types
of
clouds
or
platforms
or
systems.
So,
and
one
of
the
things
that
so
essentially
the
specification
is
a
metadata
that
can
be
used
regardless
of
the
cloud
or
the
platform
of
the
system
that
you're
using
and
then
one
of
the
things
that
what
also
want
to
mention
is
the
cloud
events.
A
It
was
came
to
the
CN
CF
as
a
sandbox
project,
and
now
that
the
specification
has
furthered
along
there
is
a
proposal
in
flay
place
for
it
to
move
from
the
sandbox
stage
to
the
incubating
stage,
as
we
have
three
different
stages
within
C
and
C
F,
just
depending
on
the
adoption,
as
well
as
the
community
health
and
growth.
We
move
these
projects
along
through
different
stages
and
the
last
stage
is
graduation
and
in
just
in
terms
of
who's
adopting
the
cloud
event
spec.
A
As
you
can
see,
it's
Microsoft
s,
ap
we're
getting
a
lot
of
support
from
other
vendors
as
well.
Arun
is
going
to
touch
on.
How
can
a
dev
is
actually
using
cloud
events
and
how
they
are
complementary
to
each
other
and
I.
Think
I
touched
on
the
incubator
status
and
now
with
all
of
this,
are
we
can
you
talk
about
containers,
civilus
infrastructure
and
how
does
that
all
fit
in
yeah.
B
So
I
mean
this
is
cube
con
right,
so
we
could
talk
about
not
just
several
eyes.
How
is
the
serverless
world
relevant
in
terms
of
kubernetes
world?
So
let's
talk
about
that
now.
One
of
the
things
that
our
customers
came
to
us
is
that
it's
cool
that
you
are
deploying
lambda
functions,
you're,
giving
us
lambda
functions.
We
love
it,
but
we
have
already
before
lambda
function
came
in
or
around
lambda
function
came
in,
we've
already
invested
quite
heavily
into
the
containers
land
of
it.
B
So
then
we
looked
at
something
called
as
AWS
for
gate
where
we
wanna,
we
don't
want
the
customers
to
do
undifferentiated
heavy
lifting
for
them
for
their
perspective.
They
just
want
to
get
it
up
and
running
on
how
easy
it
is
to
fire
up
a
container,
and
you
manage
it
for
me.
So
just
very
much
like
lambda.
We
have
those
different
for
gate
configurations,
so
you
know
you
pick
a
V
CPU.
You
pick
like
the
memory
that
you
need
and
you
get
that
one
dial,
so
there
are
about
40
different
configurations
that
are
available.
B
That
can
get
you
up
and
running
with
four
gate.
So
you
say
here
is
my
container
here
is
my
V
CPU.
Here
is
my
memory,
and
then
we
manage
it.
It's
just
like
lambda.
It
scales,
you
know
to
whatever
extent
you
want
to
scale
it
to,
and
then
it
works
like
a
charm.
Now
what
about
kubernetes
in
that
case?
So
that's
where
the
project
aqui
native.
This
is
an
open
source
project.
B
Not
a
CNC
have
sponsored
project
at
this
point
of
time,
but
as
this
is
how
you
do
several
s
with
kubernetes
essentially
so
this
is
a
plat,
a
kubernetes
plaid
built
on
top
of
kubernetes
kind
of
aligning
with
the
theme
that
you
saw
in
the
keynote
this
morning.
Kubernetes
is
a
platform
for
platforms,
so
that's
where
sort
of
canary
was
sitting
on
top
of
it,
where
you
can
build,
deploy
and
manage
your
server
list
workloads.
So
all
you
do
is
you
know
here?
Is
my
parts
pack
go
run?
B
It
go
manage
it
for
me,
works
very
much
like
a
charm,
and
another
part
that
it
focuses
on
is
how
it
promotes
the
idiomatic
developer
experience
so
you're
not
kidding
about
the
language,
typically
you're,
carrying
about
I'm
building,
an
application
using
Ruby
and
rails
or
Django
or
spring
boot.
Whatever
comes
to
your
mind,
so
it
provides
that
developer
experience
and
then
it
also
adds
your
building
kit,
ops.
You
are
building
manual,
offs,
you're,
building,
docker
ops.
B
What
kind
of
ops
methodology
are
you
using
and
it
easily
incorporates
that,
as
part
of
you
know
the
entire
workflow?
So
essentially,
there
are
three
components
to
there's
a
building
part
of
it.
Where
is
you
know
how
you
build
the
container
part
of
it?
You
know
how
do
you
serve
the
container
and
then
the
eventing
part
of
it
and
in
the
inventing
is
where
the
cloud
events
become
relevant
soaked.
B
A
native
of
course,
you
know,
gives
you
that
ability
to
manage
your
service
workloads,
but
again,
just
a
random
container
running
somewhere
as
a
function
is
not
very
helpful.
So
you
want
to
make
sure
that
there
are
events
that
are
triggering,
that
function
and
being
in
a
cross
cloud
environment
depending
upon
how
your
kubernetes
cluster
is
configured.
That's
exactly
where
cloud
events
kind
of
kick
in
now.
B
Is
that
everywhere,
is
that
enough
in
terms
of
innovation?
Is
that
enough
that
we
are
doing
in
the
serverless
world
to
help
customers
be
successful?
What
about
the
infrastructure?
We
talked
about
sort
of
at
the
function
layer.
We
talked
about
at
the
container
layer,
but
then
you
know
last
year
at
reinvent
Amazon
introduced
firecracker.
Firecracker
is
a
brand
new
virtual
machine
manager.
Essentially-
and
this
is
a
project
that
we
looked
at
as
we
were
running
lambda
workloads
as
we
were
running
Fargate
workloads.
But
how
do
we
achieve
more
efficiency
in
terms
of
running
these
workloads?
B
I?
Don't
want
to
have
one
container
per
ec2
instance.
That's
not
very
efficient
for
me.
So
that's
exactly
where
we
introduced
this
micro,
this
open
source
vmm.
You
can
go
to
firecracker
micro,
BM,
dot
IO
to
get
all
the
details,
but
essentially
what
it
gives
you
is
is
gives
you
the
security
and
isolation
of
traditional
virtual
machines,
with
a
speed
and
density
of
containers
and
I'll.
Give
you
a
little
bit
more
background
on
that.
B
So
if
we
go
to
the
next
slide
here,
so
what
it
gives
you
is
a
it
gives
you
security
from
the
ground
up.
It
really
has
that
minimal
device
model,
like
you
know
a
word
I
Oh,
a
word
block
ones,
one
key
one
keyboard
one
block
keyboard
so
really
literally
three
or
four
devices
that
are
available
to
you.
So
because
the
exposed
area
is
not
that
big,
it
really
gives
you
a
lot
more
security.
B
It
gives
you
speed
by
design,
because
now
what
I
can
do
is
I
can
launch
a
micro
VM
in
less
than
hundred
and
two
five
milliseconds.
So
essentially,
what
I
can
do
is
instead
of
running
the
ec2
instance,
sure
it
runs
in
nested
virtualization,
so
job
ADA
actually
talked
about
firecracker
and
one
of
the
TGI
Kate's
series
running
firecracker
on
Google
Cloud.
But
the
real
advantage
of
this
is
when
you're
running
on
a
bare-metal
machine
in
your
on-prem
data
center.
B
So
what
you
would
do
is
essentially
you
will
fire
up
a
bare-metal
machine
and
then
you
can
launch
as
many
micro
biomes
as
you
want
over
there
and
that's
what
gives
you
the
density?
That's
what
gives
you
the
ability
to
fire
up
the
micro
VM
quickly
and
it's
very
well
sized
for
your
container
or
for
your
lambda
function
and
last,
but
not
the
least.
It
gives
you
that
scale
and
efficiency,
because
the
footprint
of
each
micro
BM
is
less
than
five
megabytes.
B
B
Fargate
is
heading
towards
that
direction
as
well.
But
then
what
we
were
excited
is
that
last
cube
con.
We
had
really
good
meetings
with
the
Carter
containers.
Folks,
now
Carter
containers
uses
Q
mu
as
a
default
vmm,
but
then
one
of
their
longest
customer
asked
has
been
that
Q
mu
mu
is
a
little
bit
heavyweight
vmm.
B
As
a
security
boundary,
you
can
truly
run
a
no
multi-tenant,
your
data
plane,
so
this
is
one
or
vogel's.
You
know
this
is
a
snapshot
from
him
from
reinvent
a
couple
of
years
ago.
So,
and
this
really
summarizes
for
me
that
serverless
really
means
just
write
your
application.
That's
all!
That's
all
you
should
care
about
you
know,
then
you
compose
your
application
on
what
events
it
would
trigger
and
that's
all
it
should
be
about
and
I'll.
Let
somebody
else
manage
the
servers
will.
Yes,
there
are
servers
in
server
less.
B
There
are
absolutely
servers,
but
somebody
else
is
managing
them
for
you.
But
from
your
perspective,
all
you
care
about
is
write
your
application.
Let
somebody
else
take
care
of
high
availability.
Let
somebody
else
take
care
of
scaling.
Let
somebody
else
take
care
of
storing.
You
are
only
running
your
application
and
only
paying
for
that
application.
A
So,
lastly,
in
terms
of
how
do
all
of
us
get
involved,
I
think
we
highly
encourage
you
start
building
our
civilus
apps
and
functions.
It's
always
easy
to
get
started.
The
second
thing
like
I
said
before
is
it's
easy
to
update
the
server
this
landscape?
It's
a
very
simple
llaman
file,
where
you
would
put
three
or
four
lines
of
tagging
that
needs
to
happen
and
it's
dynamic
and
it's
very
easy
to
do
so.
It
can
update
at
any
time
all
the
time
please
do
join
the
serval
is
a
working
group
like
I
said.
A
The
server
list
working
group
now
is
very
much
focused
on
the
see
events
spec
and
then
there
is
also
a
another
initiative
that
is
underway
where
they
talk
about
the
event
workflow
and
what
that
exactly
means.
So
that's
another
subset
project
which
is
a
part
of
the
overall
service
working
group.
That
is
where
they
primarily
look
at
see
events,
and
then
there
is
the
events
workflow
and
then
there
are
all
kinds
of
channels
that
are
made
available.
A
So
there
is
a
service
email
channel
that
has
access
to
so
cab,
access
to
the
slides
and
click
on
the
links
here,
you'll
be
able
to
join
these
different
channel.
There
is
a
weekly
call
service
that
happens
on
Thursdays.
There
is
a
separate
get
repo
that's
available
for
cloud
events.
Please
do
join
the
mailing
list
for
cloud
events,
it's
separate
cloud
events
as
well
as
Cerberus.
A
The
other
thing
that
I
was
told
the
co-chair,
our
dog
Davis
is
very
well
known
and
I
would
say,
he's
an
expert
in
service
as
well
as
mark
P
and
when
I
talked
to
them,
I
think
one
of
the
feedback
that
Cape
gave
me.
He
apparently
went
a
few
other
conferences
and
many
asks
people
in
terms
of
how
many
of
them
are
using
service
and
production.
I
think
there
were
very
few
hands
that
went
up.
So
one
of
the
things
that
we
would
really
appreciate
is,
if
you
could.
A
Please
share
some
of
the
key
use
cases,
pain,
points
or
challenges
that
customers
are
actually
finding
in
terms
of
moving
or
losing
services
in
production
and
the
way
you
could
help
help
help
us
in
terms
of
sharing
that
feedback
as
CN
CF.
We
also
have
our
end
user
community,
so
you
could
look
up
the
end
user
community,
Cheryl
Hong,
who
runs
the
end
user
community
she's,
always
looking
for
such
requirements
and
feedback.
A
So
that
would
be
one
place
for
you
all
to
share
any
specific
use
cases
or
customer
requirements
that
you're,
aware
of
and
some
of
the
challenges
that
they're
finding
in
terms
of
using
server
list,
because
we
want
to
make
that
journey
for
our
customers
really
easy
and
then
having
worked
with
AWS
and
other
vendors,
we
are
happy
to
kind
of
work
with
them
and
figure
out.
What
exactly
are
their
pain
points?
So
it's
really
important
for
us
to
understand
the
different
use
cases
of
pain
points
from
folks
were
actually
in
this
journey
of
serverless.