►
Description
Keynote: What's Next? Getting Excited about Kubernetes in 2018 - Clayton Coleman, Architect, Kubernetes and OpenShift, Red Hat
The Kubernetes ecosystem has grown tremendously over the last three years. Each release pushes the boundaries of what we can accomplish and brings new participants and new success stories. That success has a price: how do we do what's best for the community and for our users, and what's on deck for 2018?
About Clayton Coleman
Clayton is architect and engineer on cloud orchestration and
containers at Red Hat, in charge of both technical direction for
Kubernetes and OpenShift (Red Hat's platform as a service built on top of Kubernetes) as well as the broader container and container content efforts at Red Hat. Clayton is one of the top contributors to both Kubernetes and OpenShift and has been involved in many projects in the container, platform-as-a-service, and ci/cd ecosystem over the last four years. He enjoys sleeping, but rarely has time to do it anymore.
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
So
yesterday,
I
talked
a
little
bit
about
why
we
should
make
kubernetes,
boring
and
honestly
talking
about
boring
things
is
boring
so
today,
I
wanted
to
talk
a
little
bit
about
how
we
can
make
this
ecosystem
in
this
community
really
exciting,
and
it's
because
kubernetes
is
part
of
something
much
bigger,
it's
bigger
than
container
orchestration
and
it's
getting
bigger
all
the
time,
as
evidenced
by
everybody
here
this
week,
there's
a
few
things
that
I
would
like
to
highlight
that
are
coming
in
the
next
year.
You
may
have
heard
of
them.
A
You
may
not
have
heard
of
them
and
I
want
to
make
sure
that
people
are
aware.
You
know
this
is
kind
of
at
the
end
of
the
conference,
but
over
the
next
year,
if
you
can
find
and
identify
these
technologies
and
these
tools
and
these
projects,
you
can
orient
yourself
and
maybe
get
an
idea
of
what's
coming,
get
involved
or
participate.
If
you're
already
an
expert,
you
might
have
find
something
surprising
in
my
list.
A
If
you
contribute
to
or
use
one
of
the
projects
I
mentioned,
maybe
you'll
hear
about
others
that
could
benefit
from
the
work
you're
doing
if
you're
using
kubernetes.
Maybe
one
of
these
projects
will
help
solve
a
problem,
for
you
inspire
you
to
contribute,
and
if
one
of
the,
if
every
one
of
these
projects
is
known
to
you,
speak
up,
we
actually
need
you
to
help
so
I'm
gonna
start
with
a
call
to
action.
We
should
be
making
everything
better.
A
This
is
somewhat
broad,
so
we,
the
people
in
this
room
this
conference,
this
ecosystem,
I,
assume
a
fair
amount
of
folks
here
when
I
build
things.
What
are
we
actually
going
to
build?
Well,
everything
is
somewhat
broad,
but,
let's
be
honest,
software
is
pretty
broad
now
and
I.
Try
actually
tried
to
scope
this
down
and
I
couldn't
come
up
with
a
reasonable
definition
of
software.
That
didn't
include.
Most
of
the
things
that
we
were
talking
about,
everything
is
networked.
Everything
is
on
a
different
type
of
device.
A
A
We
already
are,
judging
by
the
4300
people
here
at
cloud
native
con
and
cube
con.
Judging
by
the
sessions
I.
Think
it's
pretty
clear
that
we're
are.
Everyone
is
already
passionately
engaged
in
this
in
this
effort.
So
there's
easy.
Everybody
give
yourself
a
pat
on
the
back
great
we
won,
but
if
you
haven't
been
engaged
or
if
you'd
like
to
get
more
engaged
or
you've
been
heads
down
in
one
particular
area,
I
think
this
is
the
right
talk
for
you,
so
I
don't
have
time
to
describe
everything.
Everything
is
huge.
A
A
We
have
a
cloud
native
landscape
chart
that
doesn't
fit
on
an
8k,
monitor
Dan
like
that,
but
out
of
these
2,200
projects,
some
of
which
are
on
github,
some
of
which
aren't
there's
some
highlights.
So
if
you're,
building
new
applications,
expanding
last
year's
applications
or
integrating
a
new
technology,
2018
is
gonna,
be
exciting.
There's
tools
that
help
you
build
and
solve
problems
faster
and
better.
That's
been
true
for
a
long
time.
No
surprise
there
does
everybody
remember
Ruby
on
Rails,
maybe
some
of
you
are
too
young
I,
don't
know,
and
I
actually
had
that
thought.
A
I
was
like
oh,
my
gosh,
I'm,
so
old
that
people
may
not
remember
Ruby
on
Rails.
That
was
terrifying,
so
for
those
of
you
who
are
too
young
or
for
those
of
you
who
may
not
know
what
Ruby
on
Rails
is
Ruby
on
Rails
was
a
framework,
a
bunch
of
people
who
were
very
passionately
involved,
a
very
particular
aspect
of
technology,
building,
highly
dynamic
websites
that
empowered
their
clients
and
their
users
to
very
quickly
spin
up
ideas.
Rubies
on
Rails
was
let's
take
that
and
open
source
it.
A
Some
basic
ideas,
are
it's
so
easy
to
add
one
more
thing
where
that
one
more
thing
is
actually
valuable
to
your
users
that
I
think
more
than
any
other
project.
In
recent
years,
Ruby
on
Rails
set
a
tone
for
the
open
source
ecosystem
because
it
wasn't
just
about
Ruby
on
Rails.
It
was
about
all
of
the
people
on
top
of
Ruby
on
Rails,
who
built
gems
and
extensions
and
plugins
and
engines
and
templating
in
a
sense,
Ruby
really
saw
the
birth
of
the
modern
web
app.
Well.
A
A
It
can
connect
services
across
clusters,
operating
systems
and
clouds,
and
that
key
idea
is
rather
than
building
all
of
these
capabilities
into
each
application,
you're
on
a
proxy
next
to
each
one
and
dynamically
program
that
proxy,
based
on
what
the
application
author
or
the
infrastructure
knows
about.
What's
going
on,
this
makes
integration
much
easier.
You
don't
have
to
write
one
client
library
for
a
framework
or
for
an
idea
across
every
programming
language
that
people
use
basically
JavaScript
in
Java,
but
taking
that
out
means
I,
don't
have
to
worry
about
that
problem
and
just
like
rails.
A
Kubernetes
is
adding
things
like
GPU
support,
as
just
one
more
thing
that
belongs
to
the
cluster,
which
means
consuming
and
using
GPUs
on
cloud
or
on
bare
metal
should
go
from
possible,
but
a
little
fiddly
sometimes
to
be
diddly.
If
you
do
something
really
weird,
which
is
it
actually
a
huge
win
for
most
people?
Well,
we
probably
aren't
all
going
to
be
machine
learning
experts
at
the
next
cube
con
I
do
hope
that
we
continue
to
focus
on
taking
people
not
just
into
technologies,
but
through
technologies.
A
A
A
A
I
think
there's
a
huge
opportunity
for
the
technologies
in
this
space
to
kind
of
blur
the
line
between
function
and
container
you
know
functions
sit
down
at
one
extreme
and
if
you
think
about
what
a
container
is,
it's
really
just
a
function
that
grew
up
and
got
a
whole
bunch
more
functionality,
because
it
was
just
easier
and
so
that
just
easier
is
something
that
should
be
true
at
either
end
of
that
spectrum.
Many
of
the
common
patterns
and
serverless
fast
start
up
simple
billing
event-driven
behavior
idling
integration
are
also
features
that
normal
containers
could
use.
A
If
you
have
a
function,
it
needs
to
talk
to
my
services,
regardless
of
where
they
are,
and
if
I'm
in
one
cloud
provider,
that's
really
easy
today,
but
I'm
still
willing
to
bet
that
for
a
good
portion
of
the
people
here,
one
cloud
provider
isn't
enough
on
the
application
side.
As
Brendan
noted
there's
still
a
lot
of
work
to
go.
Do
applications
are
becoming
complex,
I'm
even
more
complex.
That
also
has
been
true
for
the
last
5
10
50
100
years.
A
A
This
is
an
effort
to
kind
of
help,
bring
together
tools
and
ideas
that
make
everybody
else's
application
definition
easier,
maybe
not
to
solve
the
problem,
because
if
this
was
a
solvable
problem,
we
would
have
solved
it
already,
but
it
tries
to
at
least
move
the
whole
ecosystem
forward
to
take
the
best
ideas
from
people
who
are
actually
doing
this
in
the
real
world
and
to
find
the
tools
that
work
for
all
of
us.
Projects
like
helm
are
going
through
a
big
effort
for
a
new
300
roadmap
for
helm.
A
I
think
there'll
be
a
lot
of
exciting
things
out
there
case
on.
It
recently
announced
some
some
big
improvements
and
how
they
focus
on
people.
I
think
this
space
is
going
to
grow,
not
shrink
and
I
think
we're
all
going
to
get
better
at
reusing
common
tools.
Make
this
easier,
so
I
was
voted
most
likely
to
make
wildly
optimistic
addictions
at
cube.
Con
2017,
my
classmates,
really
really
did
know
me
I,
don't
either
that
or
they
should
have
played
the
stock
market
instead.
A
The
second
part
of
this
is,
if
we're
going
to
build
more
complex
applications,
understanding
and
managing
them
becomes
even
more
important
identity.
If
you
may
have
seen
there
was
a
few
sessions
at
the
conference
this
week
around.
How
do
we
give
containers
identity?
How
do
we
give
services
identity
is
do
in
a
sense
is
out
working
on
some
of
these
ideas.
A
There's
a
lot
of
existing
tools
in
this
space
like
Kerberos
and
Active.
Directory
are
a
really
good
example
that
have
existed
for
a
very
long
time
and
there's
people
who
have
massive
deployments
today,
where
they're
running
things
that
are
almost
like
containers
at
scale,
and
we
want
to
take
those
lessons
and
bring
them
forward
cuz
if
we
just
reinvent
everything,
I'm,
not
really
learning
from
history
in
the
policy
space.
This.
A
This
is
an
intentionally
sparse
slide
because
we're
just
now
starting
to
add
the
extension
points
and
kubernetes
that
make
it
possible
for
people
to
plug
in.
In
an
open
way,
I
would
actually
hope
that
throughout
2018,
everyone
who
has
talked
about
adding
policy
to
Cooper
Nettie's
or
who
has
forked
kubernetes
and
added
policy
is
able
to
start
building
these
in
ways
that
that
inner
operate
or
make
it
easy
to
plug
them
in,
because
we
all
need
a
lot
more
knowledge
of
what's
going
on
or
control.
A
A
They
have
up
power,
a
simple
and
powerful
language
for
doing
policy,
rule
definition
and
they're,
using
they're
planning
to
use
the
hooks
that
are
being
exposed
in
kubernetes
to
be
able
to
intercept
when
someone
goes
and
creates
the
deployment
to
do
things
like
ok,
well,
I'm,
going
to
define
a
policy
that
says
you're
not
allowed
to
update
deployments
outside
of
weekday
work
hours
and
OPA
would,
in
a
fairly
generic
way,
say
nope.
You
can't
do
that
rejected
and
if
we
do
our
jobs
right
across
the
ecosystem,
these
tools
will
be
increasingly
easy
to
plug
together.
A
Ldap
is
another
one.
A
lot
of
people
use
LDAP
as
a
central
store
for
their
policy
or
their
organizational
identity,
and
I
would
really
might
very
much
like
to
see
better
integration
between
the
LDAP
side
of
the
fence
and
a
kubernetes
cluster,
so
that
we
can
do
a
little
bit
less
a
little
bit
more
automation,
a
little
bit
more
manual
work.
A
The
container
runtime
interface
in
kubernetes
was
a
very
deliberate
effort
started
over
a
year
ago
to
extract
out
the
connections
between
kubernetes
and
the
container
runtime,
both
for
practical
reasons
to
create
a
nice
boundary
to
make
kubernetes
easier
to
develop,
but
also
to
open
up
the
space
for
better
extensibility
and
for
other
people
to
innovate,
and
that
has
happened.
There's
a
ton
of
new
container
runtimes
out
there.
A
The
most
recent
was
Microsoft
announced
a
virtual
cubelet
that
could
do
much
of
the
same
things
of
calling
out
to
the
underlying
services,
but
even
under
that
khadak
containers.
I
have
them
up
here
as
hyper-v
was
an
effort
announced
to
take
Linux,
clear
containers
and
combine
them
with
the
existing
hyper-v
run
V
approach
of
running
a
pod
inside
a
VM
and
to
move
that
forward.
A
So
that
if
people
do
want
more
secure,
VMs
over
time,
they'll
be
able
to
use
those
in
a
kubernetes
cluster,
the
cryo
project
is
really
about
a
simple
and
fast
container
runtime
that
exists
only
to
serve
kubernetes
needs
and
to
open
up
the
doors
in
the
future
in
this
in
this
area,
for
even
more
exploration
and
optimization,
if
we
want
to
do
serverless,
we're
gonna
need
something
that
can
start
containers
pretty
darn,
fast
and
I.
Think
there's
a
lot
of
fruit
left
on
the
particular
optimization
tree.
A
debug
containers
is
a
long-running
thing.
A
I
had
to
throw
it
up
here,
because
I
feel
so
terrible
about
this,
because
I've
been
a
neglectful
reviewer
of
this
proposal,
which
is
really
the
idea
that,
if
you're
running
in
a
production
environment,
you
want
to
jump
into
a
into
a
pod
and
bring
your
debugging
tools
with
you
and
we've
gone
through,
like
15
iterations
of
this
lis.
If
you're
out,
there
I'm
really
sorry
I
apologize,
but
we
I'd
like
to
see
this
move
to
completion,
because
there
are
a
lot
of
interesting
things
that
we
can
do
if
we
can
get
visibility
into.
A
What's
actually
running
in
production
and
we
don't
want
to
take
tools
away,
but
we
want
to
be
very
deliberate
about
how
we
add
those
tools
and,
as
I'm
I've
gone
through
a
whole
bunch
of
technologies.
These
four
stood
out
to
me
because
they
really
are
awesome
examples
of
taking
the
ideas
and
Cooper
Nettie's
and
going
way
past
anything
that
we
would
have
thought
about.
A
You
know
the
possibilities
were
there,
but
each
of
these
projects
in
their
own
way
took
some
unique
aspect
of
the
ecosystem,
made
it
simpler,
easier
in
ways
that
we
couldn't
predict
when
we
started,
and
so
the
cube
D
project
by
apps
D
is
a
cluster
enabler
for
cluster
ops
on
kubernetes.
It
has
a
lot
of
great
ideas
around
the
dynamic
nature
of
kubernetes.
A
A
It
makes
it
really
easy
to
take
a
kubernetes,
namespace
or
set
up
kubernetes
namespaces
or
an
entire
cluster,
and
because
of
the
effort
that
we
put
in
to
have
a
simple
and
declarative
model
for
a
api's
in
kubernetes
arc
was
able
to
just
go
and
grab
everything
by
discovering
all
the
api's.
The
kubernetes
api
is
the
extension.
Api
is
the
custom
resources
bundle
all
up
back
them
up
and
then,
if
you
want
to
restore
them,
arc,
is
able
to
go
Andrea
reapply.
A
Those
back
to
your
cluster
in
a
very
predictable
way,
so
again,
not
something
that
anybody
in
the
direct
core
kubernetes
team
ever
planned
for,
but
which
people
who
are
knowledgeable
about
kubernetes
in
the
ecosystem,
we're
able
to
apply
the
concepts
and
make
something
that
makes
people's
lives
easier.
The
cube
router
project
simplifies
both
networking
and
the
cube
proxy
to
make
services
much
more
efficient,
and
it
leverages
some
really
cool
ideas
from
other
projects
to
deliver
something
that
feels
very
cube
native.
A
But
it's
also
something
that
we
could
not
have
anticipated
or
would
not
have
designed
ourselves
and
yesterday's
keynote
about
the
cube
medic
controller.
Again.
The
idea
of
making
it
really
easy
to
add
the
next
incremental
thing
to
your
infrastructure,
to
empower
your
developers,
your
operators
and
more
of
these
projects
exist
out
there.
I
could
only
find
4
or
I
could
only
find
4
to
put
on
this
slide,
because,
if
I
keep
talking,
I
could
again
going
for
10
or
15
days.
A
So
there's
a
lot
of
stuff,
but
it's
not
even
a
fraction
of
what's
out
there,
how
do
we?
You
participate,
we're
a
community,
we're
an
ecosystem,
I'm
not
going
to
stand
up
here
and
tell
people
what
to
do.
In
fact,
I
would
much
prefer
to
work
and
you
listen
about
what
people
want.
If
you
build
something,
tell
others
if
you're
using
something
and
it
doesn't
work
for
you
talk
to
them.