►
Description
Keynote: The Road Ahead on the Kubernetes Journey - Craig McLuckie, CEO, Heptio
It has been amazing to watch Kubernetes emerge as a standard operating environment for distributed systems development over the past few years. In a short few years it has become embraced by almost every significant vendor in the ecosystem and is going from strength to strength. It is emerging not only as a way to not only solve hard problems deploying and running applications, but is supporting the development of new approaches to building and running applications that power the world.
During this session, Craig McLuckie, one of the Kubernetes founders and CEO of Heptio will look ahead to the coming years and talk about some important trends in the ecosystem that will continue to support and drive the success of the project. We will focus on the emergence of expert operations and talk about how Kubernetes is starting to change the organizations that build and manage distributed systems. This will touch on how SRE values are starting to find their way into modern development teams, what tools are still needed to drive ops maturity and the overall value of this trend to companies adopting cloud native technologies. We will discuss the value of continued focus on modularity and extensibility in the cloud native ecosystem as a way to foster innovation in the ecosystem, and also discuss the the emerging role Kubernetes is playing in the increasingly heterogeneous world of cloud.
About Craig McLuckie
CEO and founder of Heptio, a company built to propagate cloud native computing technologies. Previous Googler and founder of Kubernetes, an Open Source cluster manager.
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
It's
so
great
to
be
here
how
many
of
you
were
at
odds
Khan
a
couple
years
back
when
we
announced
$1
yeah
I
mean
it's
just
like
amazing
how
this
community's
changed,
and
so
I'm
very
happy
to
be
here
today.
Now
kubernetes
is
just
the
most
amazing
journey,
I've
a
technical
journey
I've
seen
in
my
life
and
I
think
it's
been
as
a
result
of
this
unique
fusion
of
three
things.
What
we've
seen
is
a
community
evolve.
A
This
incredibly
strong,
technically
grounded
smart
community
that
is
invested
in
the
same
set
of
things
and
that's
been
very
powerful.
The
second
thing
we've
seen
is
a
set
of
end
users
start
to
emerge
and
actually
be
part
of
that
community
already
make
the
technology
their
own,
be
a
part
of
it,
build
their
business
around
it.
And
then
the
third
piece
has
been
the
vendor
ecosystem
organizations
that
are
willing
to
make
huge
strategic
bets
in
this
technology.
A
A
These
thing
three
things
are
really
obvious,
but
when
you
start
pulling
them
together,
it
really
tells
a
story
about
how
important
it
is
to
hold
communities
together.
How
important
a
number
of
attributes
of
kubernetes
are
and
I'd
like
to
just
sort
of
tell
the
story
for
you.
The
first
point
here
is
developer
productivity.
A
Some
interesting
challenges
and
saying
I
want
to
talk
through.
So,
let's
start
with
the
first
one
developer
productivity
matters,
we
work
with
some
organizations
that
hire
tens
of
thousands
of
developers,
tens
of
thousands
right
and
some
of
these
developers,
you
know,
spend
their
life
writing
code
and
some
of
them
spend
their
life
waiting
90
days
for
a
virtual
machine
to
be
provisioned.
It's
not
a
great
experience.
Anything
we
can
do
even
a
5
percent
improvement
in
develop
productivity
will
have
disproportional
impact
on
those
organizations.
When
you
think
about
it,
technology
is
changing.
A
The
world
technology
is
driving
every
business.
What's
driving
technology,
it's
developers
and
kubernetes
has
been
a
seed
crystal
to
this
new
way
of
getting
development,
running
forwards,
getting
code
into
production,
and
we
tend
to
think
about
it.
A
lot
from
as
a
test
of
a
developer
framework
and
one
of
the
things
that
I've
really
observed
and
I
want
to
think
about
is,
as
we
move
this
forward
ziz,
it's
not
just
about
development.
It's
really
about
operations.
At
the
end
of
the
day,
the
thing
that's
holding
those
developers
back
is
the
operations
team.
A
It's
the
operational,
drag
it's
the
thing
that
is
constantly
gating
their
ability
to
produce
and
deliver
great
code,
and
so,
when
you
start
thinking
about
you
know
what
has
been
driving
develop
productivity,
particularly
with
that
operations
lens.
It's
this
idea
of
moving
from
static
code
to
living
services,
and
nothing
embodies
this
more
than
the
cloud
right.
When
you
think
about
the
cloud,
everything
in
the
cloud
is
something
as
a
service,
so
software
is
a
service.
Kubernetes
is
a
service
infrastructure
as
a
service
and
what
that
means
is
I'm
going
to
take
a
technology
whatever.
A
When
you
go
to
Amazon
when
you
go
to
Azure
when
you
go
to
Google
you're,
not
just
buying
basic
infrastructure,
you're,
not
just
buying
access
to
an
open
source
technology,
you're
buying,
sre,
right,
you're
buying
an
Operations
team,
that's
gonna
deal
with
this
for
you
and
there's
nothing
stopping
you
from
bringing
those
practices
into
your
own
organization.
There's
nothing!
A
Stopping
you
from
developing
this
incredibly
rich
way
of
thinking
about
delivering
your
software,
not
just
as
static
dead
things
that
you
throw
over
the
wall
at
the
business
every
90
days,
but
as
a
living
service
with
an
expert
operations
team
that
is
delivered
to
the
entirety
of
your
organization.
That
is
very
powerful.
A
When
you
start
making
this
step
a
lot
of
great
things
start
to
happen,
you
can
start
looking
at
a
world
where
you
know
I
speak
to
so
many
of
these
organizations
today
that
are
living
in
the
world
of
virtual
machines
and
traditional
infrastructure,
and
they
talk
about
resiliency.
It's
like
you
know,
enterprise
is
about
resiliency,
it's
about
how
long
does
it
take
to
these
things
before
they
go
down?
That's
that's!
That's
gonna
impact,
my
business
and
I,
look
at
them
and
I
always
say
like
you're,
focusing
on
the
wrong
thing.
A
You
should
be
focusing
on
recoverability.
How
longs
it
take
you
to
get
it
back
up
when
it
goes
down
right,
there's
a
lot
of
different
ways:
you
can
cut
three
nines.
That
could
be
one
second,
every
thousand
seconds,
where
no
one
even
notices
an
interruption
in
service.
It
could
be
an
entire
business
day
once
every
three
years,
where
everyone
really
notices
it.
That's
the
type
of
thing
that
ready
focusing
on
these
sort
of
operational
practices
will
get
you
the
relentless
focus
on
automation.
A
You
know
removing
toil
out
of
the
day-to-day
life
of
an
organization
getting
to
a
point
where
your
operations
team
is
creating
a
gate,
a
back
pressure
in
your
developers
doing
things
like
SLA
budgeting,
or
this
is
possible
and
kubernetes
really
is
the
secret
of
borders.
It
starts
to
deliver
a
framework
when
you
can
tease
apart
these
various
concerns
right
now.
A
This
here
gives
us
a
way
to
start
thinking
about,
and
here's
just
start
on
the
right
there
with
the
cloud
world.
You
can
look
at
the
stack
where
organizations
have
started
to
create
these
expert
operations,
functions
and
tease
apart
different
services,
so
you
have
one
team,
that's
racking
and
stacking
VMs
and
delivering
it
as
a
service
where
you
can
request
it
with
an
API.
A
It's
called
infrastructure
service,
the
goodness
that
every
major
cloud
provider
out
there
is
now
delivering
to
heddes
in
the
same
way,
and
it's
kind
of
creating
this
Goldilocks
abstraction
where
they
actually
have
a
pretty
harmonized
story.
Above
it,
you
have
the
cloud
services
that
you
know
exist.
You
know
potentially
in
the
future.
Who
knows
maybe
they'll
be
writing
on
kubernetes
I
suspect
there
probably
will
be
in
other
cases,
or
you
can
actually
get
a
fully
packaged
application.
You
know
in
the
form
of
SAS,
so
that's
that's
the
world
of
cloud
now.
A
You
can
do
all
of
this
yourself
as
well
like.
Why?
Wouldn't
you,
if
you're
a
massive
enterprise,
you
could
run
an
infrastructure
operations
team
and
have
a
dedicated
set
of
organizations.
You
could
introduce
kubernetes
as
this
Goldilocks
abstraction
that
creates
a
very
clean
abstraction
above
it
you
could
take
a
lot
of
your
services
and
deliver
it
in
the
same
model.
The
same
metaphor
as
a
cloud
providers
are
delivering
and
you
could
create.
A
You
know
entire
applications
that
are
run
that
same
way.
That
separation
of
concerns
is
very
powerful.
You
get
the
same
efficiency
to
the
cloud
providers
do
or
you
can
mix
and
match.
You
can
say
hey.
You
know
what
I'm
just
gonna
use
the
cloud
provider
Scriven
as
a
service,
but
I'm
still
gonna
start
thinking
about
creating
my
own
service
Operations
teams.
This
becomes
very
powerful,
but
to
really
light
it
up.
You
start
to
need
to
have
a
set
of
common
policies.
A
That's
span
across
these
areas
right
if
you're,
starting
to
mix
and
match
I
want
to
use
my
infrastructure
at
cloud
providers
infrastructure
we
use
communities
as
abstractions
on
I
still
probably
have
security
and
governance
policy
that
needs
to
exist
need
to
think
about
that,
and
in
many
ways
that's
why
kubernetes
is
so
powerful,
like
you
could
draw
that
line
across
that
cluster
ops
area,
and
you
can
get
a
perfectly
harmonized
solution
that
works.
What
about
those
environments?
And
then
you
know,
create
commonality
above
it,
so
you
can
flexibly
pick
where
things
run.
A
We
really
need
strong
conformance
right,
so
I
have
to
take
a
moment
to
congratulate
every
one
of
the
major
cloud
providers
out
there
who
have
introduced
kubernetes
as
a
service
offerings
right.
It's
not
just
kubernetes
as
a
service
offerings,
they've
introduced
upstream
friendly
kubernetes
as
a
service
offerings.
That's
a
very
powerful
thing
and
part
of
the
reason
that
we
can
feel
good
about
that
is
that
we're
getting
into
a
situation
where
it
doesn't
matter
who
is
delivering
the
communities
it
doesn't
matter
how
it
is
operated.
A
It
doesn't
matter
how
it
is
provisioned
the
thing
that
matters
is
how
it
runs
and
for
us
to
get
to
a
point
where
we
have
high
levels
of
assurance.
That
kubernetes
is
working
exactly
as
it
should
conformance
really
matters.
We
need
to
have
the
ability
to
a
test
that
these
clusters
are
semantically,
consistent,
that
they
have
the
same
behavior,
and
you
know
for
that.
A
This
is
why
the
CNC
F
certified
kubernetes
program
is
so
important
when
you
see
that
kubernetes
logo
on
someone's
offering
you
can
have
a
relatively
high
degree
of
assurance
that
it's
working
as
it
should,
and
this
leads
us
to
some
really
good
places.
It
makes
hybrid
cloud
wheel
right
now.
You
can
say:
hey
I'm
gonna
run
something
on
Prem.
It
could
be
something
I
construct
myself.
It
could
be
something
from
one
of
the
districts
and
I'm
gonna
have
something
that's
running
in
my
public
cloud.
A
That's
delivered
by
my
favorite
container
as
a
service
or
kubernetes
as
a
service
provider.
Now,
how
would
I
make
sure
that
these
things
are
semantically
consistent?
How
would
I
be
able
to
say
that
hey
we're
I'm
using
my
development
tools,
they're
gonna,
be
able
to
focus
on
with
these
environments.
The
answer
is:
you
need
to
run
the
same
tests
in
both
places
right,
you
need
to
go
to
certify
your
on-prem
piece.
So
this
is
why
we've
been
working
with
the
community
to
create
a
technology
like
cirnoboys.
A
You
can
actually
package
up
and
run
these
technologies
and
make
sure
that
you're
getting
that
semantic
consistency
across
both
of
these
environments.
No.
How
do
we
take
that
one
step
further
and
get
multi-cloud
to
be
real?
Well,
multi
cloud
is
interesting,
like
I,
don't
think
about
multi
cloud
is
being
hey.
I
want
to
take
this
application
and
I'll
literally
want
to
run
into
clouds.
A
I
mean
there
are
some
applications
where
you
can,
if
you're
in
a
bank
and
you're
running
these
big
old
Markovian
processes
that
are
largely
stateless
and
just
have
long,
you
know
compute
run
time
cycles.
Yes,
it
looks
great
if
your
Bitcoin
mining
works
great,
if
you're
doing
anything
that
has
massive
amounts
of
data
like
nothing,
creates
gravity
or
inertia
like
data
right,
and
then
we
start
to
pull
in
service
dependencies.
A
That's
also
gonna
kind
of
hold
you
back
see
you
have
to
be
judicious
about
that,
like
multi-cloud
to
me,
means
having
the
flexibility
to
pick
which
cloud
I
want
to
deploy
my
new
service
into
it
means
being
able
to
say,
hey.
I
would
like
to
run
this
in
different
regions
and
I.
Have
my
data
shot
it
out
by
those
regions
and
I
can
pick
different
cloud
providers
to
make
those
sense
of
those
regions
but
to
really
get
there?
A
We
need
to
get
to
a
point
where
we
have
not
just
you
know,
there's
a
tested,
stable
narrative
where
everything
is
working
all
these
other
in
these
areas,
but
you
can
also
start
to
extend
your
governance,
risk
management,
compliance
policy
and
security
enforcement
across
all
of
those
with
a
common
abstraction
at
across
the
top
of
them,
and
that's
why
extensibility
is
going
to
become
so
important
and
I'll.
Give
you
an
example
of
this.
A
A
Backup,
restores
hard
right,
like
let's
make
sure
that
we
have
a
great
solution
for
that,
and
so
they've
been
working
with
us
to
make
sure
that
the
arc
technology
provides
a
great
backup
recovery
solution
for
them,
but
I
want
to
take
your
plus.
The
state
actually
gives
you
a
line
on
multi
cloud
that
I
think
is
going
to
be
really
exciting,
as
we
look
to
the
future
like
this
technology
enables
you
just
do
not
just
you
know,
take
your
cluster
stayed
and
back
it
up
and
restore
there,
but
potentially
restore
it
somewhere
else.
A
Now
we're
not
there
yet
on
the
persistent
volumes,
there's
a
whole
bunch
of
work
that
will
need
to
be
done
to
make
sure
that
that's
real,
but
you
can
see
the
promise
of
this
as
a
way
to
start
moving
things
around
inside
clouds.
You
know
directly
so
inside
something
like
Microsoft's
cloud
will,
potentially
between
clouds
or
from
on
prenta
cloud
and
kubernetes
is
the
key
to
that
whole
thing.
A
Now,
Ken
did
a
great
job
of
talking
about
this
earlier
and
I.
Think
there's
a
couple
of
reasons
why
extensibility
is
so
important
right
Enterprise
is
complicated
when
you
start
working
with
a
really
big
organization.
You
know
from
our
perspective-
and
this
is
just
a
message
to
the
community
for
folks
that
are
surveilled
in
the
journey.
I'm
sure
that
people
have
a
late
in
the
journey
already
have
experienced
this
in
in
buckets.
There's
a
lot
of
unique
needs
right.
If
you're
deploying
a
kubernetes
cluster
into
an
air-gapped
environment,
there
are
a
lot
of
unique
needs.
A
If
you're
working
with
a
regulated
environment,
there
are
a
lot
of
unique
needs
and
it's
difficult
for
one
organization,
one
vendor
one
person
to
meet
all
of
those
needs
so
for
us
to
succeed
community.
We
need
to
really
think
through.
Where
are
those
key
extensibility
points
you
know
like?
How
do
you
make
sure
that
kubernetes
remains
extensible
and
open
and
there's
been
so
much
great
work?
That's
been
delivered
by
the
community
over
the
last
couple
of
months,
even
in
terms
of
extensibility,
but
there's
a
lot
more
work
that
still
needs
to
be
done.
A
You
know
we
need
to
start
thinking
about
how
to
formalize
a
lot
of
the
interfaces
that
exist
and
how
to
make
sure
that
our
conformance
frameworks
allow
for
these
lines
of
extensibility,
because
when
we
get
to
that
world,
when
we
start
thinking
about
you
know
where
this
is
all
going,
you
know
how
to
start
reasoning
about
kubernetes
across
the
multi
cloud
verse.
When
we
start
thinking
about
hey,
we
have
different
requirements
for
different
workloads.
A
A
I
might
have
a
CI
CD
framework
that
I've
built
out
that
you
know
works
across
potentially
other
environments
to
not
just
kubernetes
and
then
I
might
need
some
security
capabilities,
like
the
security
is
a
great
example
like
look
at
what
kubernetes
is
doing
for
security
and
think
about
the
power
that
that
brings
right.
Like
you,
look
at
the
work
that
aqua
and
stack,
rocks
and
twistlock
are
doing
in
the
security
space.
When
you
think
about
that
increased
line
of
visibility
that
containers
provide.
There's
some
really
cool
stuff.
A
What's
talking
what
you
can
have
a
very
strong
sense
of
observed,
security
exploits
rather
than
you
know,
having
to
kind
of
harden
at
the
image
level
and
when
you
think
about
kubernetes
as
a
control
plane,
you
have
an
immediate
source
of
remediation
where
you
can
start
to
just
cut
things
off
and
shut
them
down.
There's
an
amazing
amount
of
innovation
that
we're
going
to
start
to
see
and
in
a
place
just
like
the
security
demand.
A
I
like
and
I
feel
safe,
saying
this
I
have
no
horse
in
the
security
race
like
as
a
event
I'm
just
delighted
to
see
the
work
that's
being
done
there,
but
for
those
organizations
to
be
able
to
bring
value
to
the
community,
it's
critical
that
they
can
plug
into
the
environment,
that
users
are
providing
not
just
that
on-prem
environment
or
that
you
know
the
spoke
environment
or
that
environment.
That
organizations
are
running
themselves
but
into
a
KS
into
e
KS
into
g
ke,
and
so
the
formalization
of
these
sorts
of
interfaces.
A
The
formalization
of
extensibility,
the
introduction
of
that
into
the
conformance
framework
itself,
is
going
to
enable
you
the
end-user
to
get
access
to
all
kinds
of
awesome,
wonderful
things
that
are
being
delivered
in
the
surfing
community
and
it's
going
to
enable
these
organizations
to
innovate
and
and
not
be
tied
into
single
provider
and
I.
Think
you
know
when
I
look
to
the
kind
of
you
know
the
future
this
coming
year.
This
to
me
stands
out
as
one
of
the
key
things
and.
A
A
You
know
it's
very
much
in
its
infancy,
we're
still
at
a
point
where
we
haven't
felt
the
full
potential
of
it,
not
just
as
a
way
to
solve
some
long-standing
issues
around
the
deployment
and
management
of
software,
but
as
a
way
to
build
new
classes
of
distributed
systems
where
kubernetes
is
the
core
development
environment.
Where
effectively
emerges
as
the
kernel
for
this
new
for
this
new
world,
and
to
do
that,
you
know
we
really
need
to
solve
the
operations
problems
like
it's,
not
about
zero
ops.
A
Anyone
who
says
your
office
is
starting
something
it's
about
ops
power
tools
like
of
specialization
getting
developers
are
the
operational
game,
getting
the
S
or
ease
into
it
right.
It's
about
conformance
like
we
currently
have
less
than
30
percent
of
the
formal
api's
stable
API
is
covered
by
the
kubernetes
conformance
test
right.
If
we
want
this
vision
to
be
real,
we
as
a
community
need
to
put
a
whole
bunch
more
resources
into
formalizing
that
not
just
the
api's
that
you
use
on
a
day-to-day
basis
to
run
your
applications.
A
Okay,
so
Jimmy's
paradox
is
this
kind
of
economic
theory
where,
when
you
massively
decrease
the
cost
of
something
when
you
massively
increase
the
efficiency
of
something
you
expect
to
see
the
Dom
things
are
going
to
consume.
Less
of
it
right.
You
increase
the
efficiency
of
Coal
engines,
people
expected
to
see
coal
consumption
go
down,
it
actually
went
completely
the
other
way.
The
increased
efficiency
created
new
uses
of
it.
It
created
new
ways
to
consume
it.