►
From YouTube: OpenShift Commons Gathering 2019 Santa Clara The Road to Unified Hybrid Cloud Reza Shafi RedHat
Description
OpenShift Commons Gathering
2019 Santa Clara
The Road to Unified Hybrid Cloud
Reza Shafi Red Hat
A
Good
morning,
everyone,
it's
good
to
see
the
room
fool.
It's
actually
been
great
to
see.
Diane's
work,
I
come
from
core
OS
I
was
running
the
product
team
at
core
OS
and
for
the
last
year,
I've
been
helping
here
with
the
next-generation
openshift
platform,
which
we
call
open
ship
for
which
you're
going
to
hear
quite
a
bit
about.
It's
really
good
to
see
the
room
fou
here,
and
you
know
this
is
the
second.
Actually
it's
the
third
Commons
event
that
I've
been
at
the
the
first
one
was
Buenos
Aires,
where
we
had
over
600
people.
A
It
was
amazing
to
see
that
that
room,
full
and
cube
con,
which
had
to
think
over
400
people.
So
these
events
continue
to
impress
me
and
I
wanted
to
before
getting
started
with
my
presentation.
I
wanted
to
thank
Diane,
she's
she's
been
doing
amazing,
work
running
these
events
and
bringing
this
community
together
and
by
the
way
she
won.
The
community
award
at
the
last
cube
con
all
right.
So,
let's
get
started
now
before
you
know.
Diane
talked
about
the
the.
A
How
and
the
what
of
the
community
were
trying
to
build
here,
but
we
I
wanted
to
step
back
and
talk
about
the.
Why
and
these
one
important
reasons
behind
the.
Why
and
to
me
that
starts
a
while
back
back
in
2003,
where
there
was
an
article
that
came
out
called
IT,
doesn't
matter
now,
I
wanna,
see
who
show
of
hands?
Who
remembers
this
article,
okay,
a
couple
of
show
of
hands
and
you
do
another
show
of
hands
who
was
actually
in
the
industry
in
2003?
A
Okay,
that's
pretty
good
about
half
of
the
room
if
I,
if
I,
just
put
back
my
mind
to
that
time,
I
was
I
was
a
young
consultant
just
starting
my
career
as
a
computer
scientist,
and
this
article
caused
a
lot
of
steer.
People
were
going
out
there
saying.
Well,
you
know
what
what
we
have
you
know
the
industry
we're
in
is
not
going
to
go
anywhere,
IT
doesn't
matter
and
I
see,
is
going
to
commoditize
overtime
and
the
end.
A
Now,
if
I
fast
forward
to
now
and
I
look
just
at
the
last
week
and
somebody
just
day-to-day
things,
I've
been
doing
right,
I
use
uber
on
a
regular
basis,
although
today
I
I
took
two
hours
ride
from
Marin
to
here,
which
you
know,
I
should
have
probably
used
uber.
So,
but
you
know
that
innovation
was
not
there.
That
has
been
fundamentally
disruptive
to
the
to
the
basic
commute
industry.
A
If
I
look
at
concur,
anyone
from
concur
here
by
the
way
well
I
think
some
of
the
folks
are
coming
and
they
run
on
tectonic,
which
is
some
of
the
technologies
that
we're
putting
into
open
ship
for
now.
I
use.
Concur
all
the
time
and
I
remember
back
in
2003,
using
the
expense
systems
that
were
around
was
you
know,
night
and
day
to
using
concur
it's
just
so
much
easier
and
they
completely
disrupted
that
industry.
A
So
so
who
here
thinks
then
that,
based
on
what
I
just
said,
this
article
nicholas
carr
was
right.
Anyone
obviously
I'm
presenting
a
biased
view
on
this,
but
no
well
I
actually
think
he
was
somewhat
right.
So,
let's,
let's
let
me
explain
why,
because
I
think
that
last
example
I
used,
does
give
you
some
commoditization
of
IT
right
if
I
want
to
go,
get
compute
resources
again,
it's
like
now
like
a
utility
and
I'm
actually
in
fact
charged
on
it
like
a
utility,
sometimes
I.
A
You
know
very
often
I
end
up
getting
bills
from
Amazon,
where
I
think
I've
closed
everything
up,
but
I
haven't
and
I
still
get
that
50
cents
for
that
elastic
cashiering
whatever
it
is.
You
know,
and
that's
that
feels
like
a
utility.
So
what's
going
on
here,
we
have,
you
know
we're
in
renovated
we're
continuing
to
innovate,
but
yet
there
is
the
commoditization
happening
and
now
would
argue
that
it
all
depends
at
what
layer
you're.
A
Looking
at
this,
if
you're,
looking
at
it
from
a
compute
infrastructure
perspective,
I
would
argue
that
is
well
on
its
way
to
you
know
to
commoditization.
In
fact,
the
utility
companies
for
that
are
becoming
fewer
and
fewer.
We
know
them
all.
But
if
you
look
at
the
layers
above
that,
if
you
look
just
one
layer
above
at
the
type
of
services
that
are
available
for
application
developers,
application
owners
to
build
their
application,
innovation
is
still
thriving.
A
I
mean
one
just
has
to
look
at
the
number
of
projects
that
are
being
donated
to
the
CN
CF,
just
to
see
a
little
bit
of
that
right.
The
type
of
projects
that
are
being
there
like
Vitas
and
like
harbor
and
other
projects
that
are
being
donated
there,
the
innovation
is
still
happening.
So
then
you
go
one
level
above
that
and
to
argue
that
enterprises
do
not
need
to
build
applications
in
order
to
disrupt
their
industry
and
continue
plowing
ahead.
A
On
top
of
this
stack,
you
know,
obviously
it
would
not
make
any
sense,
so
innovation
is
happening
at
the
top
layers,
but
the
infrastructure
level.
It
is
on
its
way
to
come,
a
decision,
I
would
argue-
and
it's
in
fact,
going
back
to
Nicholas
Carr's
argument.
The
same
thing
would
apply
to
electricity.
You
know
not
many
folks
are
innovating
at
the
to
go
infrastructure
level,
but
yet
the
electrical
services
level,
you
would
say
envision-
is
still
probably
happening.
A
You
know
innovation
in
the
battery
power
innovation
in
just
going
to
Starbucks
and
putting
your
cell
phone
on
a
table
and
getting
it
charged
automatically
wirelessly.
These
are
innovations
at
the
electrical
services
level
and,
of
course,
at
the
appliance
level.
Well,
you
know
we're
continue
to
innovate
in
terms
of
electrical
appliances.
A
Obviously,
there
is
one
important
difference
between
the
IT
world
and
the
electrical
world
that
worries
me,
and
that
goes
to
the
root
of
why
I
think
this
community
of
people
really
matter
and
what
we
are
doing,
we
all
are
doing
in
the
open-source
community
and
across
all
the
projects
were
involved
in
matter,
and
that
is
that
with
electricity,
you
actually
have
the
option
of
moving
your
electrical
appliance
from
one
electrical
provider
to
the
other.
Without
much
effort,
and
let's
take
a
look
at
how
this
happened.
A
It
almost
was
by
chance,
but
essentially
when
electricity
was
introduced
back
in
the
days
of
Edison
and
Tesla,
it
was
mainly
you
first
introduced
for
lighting
and
light
bulbs
were
created,
the
light
bulb
interface
was
created,
but
then
folks
realized
well.
I
need
to
create
other
appliances
and
I
need
to
plug
that
in
so
you
can
see
it
as
a
toaster
there.
A
In
the
big
picture
there
is
a
hair
straightener
and
there's
a
foot
warmer
and
all
of
them
the
food
bomber
actually
goes
up
into
the
ceiling,
have
the
plug
that
goes
into
the
electrical
plug,
and
that's
because
they
leverage
that
interface
as
a
standard
interface
for
plugging
into
the
electrical
service.
The
infrastructure
service.
Now
imagine
if
this
hadn't
developed
that
way.
That
would
mean
that
if
I
wanted
to
move
to
San,
Diego
and
I
wanted
to
move
my
toaster.
A
A
How
does
one
fight
that?
How
does
one
work
with
that
environment
and
figure
out
a
way
where
you
get
the
flexibility
but
also,
at
the
same
time,
get
the
portability?
That's
one
big
wide
that
I
think
this
type
of
community
is
really
working
in
solving,
so
I
wanted
to
talk
about
how,
from
our
perspective
from
Red
Hat's
perspective,
it's
all
about.
Having
that
true,
hybrid
cloud
experience,
if
you
go
to
the
Amazon
console,
if
you
go
to
the
Google
console,
you
get
this
amazing
cloud
experience
where
you
can
see
all
of
your
compute
fabrics.
A
You
can
see
all
the
various
different
types
of
services
and
you
can
deploy
applications
that
leverage
your
services
and
the
whole
thing
kind
of
runs
right.
It
updates
by
itself.
You
don't
have
to
worry
about
how
it's
built
it
just
works.
Well,
imagine
if
you
could
get
the
same
benefits,
but
in
the
in
a
truly
hybrid
model,
where
you
could
deploy
those
fabrics
on
any
infrastructure
provider,
and
you
could
leverage
the
services
that
are
beyond
the
services
that
those
cloud
providers
provide
and
tap
into
the
diversity
of
services
that
desert
these.
A
You
know
a
software
vendor
ecosystem
is
building
out
there
and
and
actually
consume
those
services
in
the
same
way
so
self-service
or
to
provisioned,
metered,
also
automatically
updated
and
so
on.
So
that's
what
we
we
internally
call
the
unified
hybrid
cloud
and
we
are
working
with
our
community
to
actually
rally
behind
and
built
now
the
unified
hybrid
cloud.
You
know
the
to
me.
The
three
pillars
behind
it
are
are
as
follows:
first,
cube
earnings;
second,
it's
all
about
automated
operations
and
third,
is
all
about
actual
hybrid
services.
A
A
It's
good
from
my
perspective
that
Red
Hat
took
a
chance
on
kubernetes
back
in
2015,
along
with
core
OS,
and
it's
good
to
see
all
the
other
vendors
jumping
in
as
well.
Now
now
what
that
gives
Red
Hat
is
an
advantage
that
I
think
is
important
in
this
Y,
which
is
that
we're
one
of
the
biggest
contributors
as
she
saw
with
some
diane's
slides,
cuber
Nettie's.
A
But
if
you
look
at
another
axis,
if
you
look
about
infrastructure
neutrality
and
who
is
well-positioned
to
actually
drive
this
initiatives
that
allow
for
portability,
Red
Hat
also
stands
out
now
automated
operations.
This
was
actually
at
the
core
of
core
OS,
which
is
all
about.
How
do
you
make
everything
more
automated,
because
we're
outnumbered
out
there
Bharani
I
run
across
this
quote,
which
I
think
summarizes
that
the
philosophy
of
Korres
and
some
of
the
early
Korres
folks
here
might
disagree
with
me.
A
A
If
you
look
at
that
stack,
it
started
from
the
operating
system
with
container
Linux,
which
was
our
operating
system
that
operated
on
an
automatic
basis.
We
had
over
200
instances
of
container
Linux
running
out
there
listening
into
different
channels
and
whenever
we
would
push
that
button
for
updates
to
be
sent
every
two
weeks.
A
100,000
plus
instances
of
operating
system
would
get
updated
and
we
brought
that
up
to
the
kubernetes
layer
with
over
6,000
clusters
being
updated
every
month,
and
so
that
stack
is
being
placed
at
the
core
of
the
next
generation.
Openshift
stack
and
all
of
the
openshift
capabilities
that
the
overshift
customers
and
community
are
used
to
is
being
maintained.
On
top
of
that
stack,
that's
the
green
stuff!
That's
that's!
Where
we're
going
from
an
other
bringing
in
automated
operations
to
open
shift
now
installation.
Of
course,
matters
too.
A
We
want
to
make
the
installation
of
your
infrastructure
as
simple
and
automated
as
possible,
we're
not
going
to
demo
it
to
here,
but
internally,
using
the
open
script
for
snack
clayton
coleman
who's,
one
of
their
leader
Vangelis
out.
There
has
created
a
tool
on
slack
where
I
can
create
a
an
opener
shift
instance
using
a
slack
command
called
launch
in
less
than
20
minutes
right.
I
just
do
a
launch
I
get
back.
A
The
the
updates
I'm
just
going
to
talk
about
this
a
little
bit,
but
essentially
the
way
updates
work
is
that
clusters
can
be
registered
to
this
notion
of
channels
and
they
can
receive
updates
on
these
channels.
In
the
tectonic
world
we
had
a
pre-production
channel
and
a
production
channel.
We
first
push
updates
to
a
pre-production
Channel.
You
would
receive
a
note
as
an
administrator
that
an
update
is
available.
You
can
then
decide
to
apply
the
update
a
cluster
of
over
a
hundred
notes.
A
Moving
on
to
oK,
we've
got
cuban
Eddie's.
That
gives
you
some
level
of
sort
of
neutralization
layer
to
the
infrastructures
that
are
out
there.
Then
you've
got
automated
operations
on
Cuban
Eddie's
and
the
operating
system
doesn't
matter
where
it
runs.
But
we
haven't
talked
about
those
compute
services
right.
How
do
you
make
sure
now
that
the
applications
that
you
build
don't
end
up
using
a
proprietary
service?
A
So
the
thing
is:
if
software
vendors
leverage
these
tools,
the
idea
is,
then
the
consumers
of
those
software
on
cue,
bernardi's
environments
can
benefit
from
the
same
benefits
as
any
service
that
a
cloud
provider
provides
yet
have
also
the
benefits
of
portability
and
more
and
a
good
example
of
that
to
me
is
the
crunchy
operator,
so
the
crunchy
operator,
if
you
compare
it
to
the
alternative,
exemplifies
what
I
just
said
so
before.
If
you
just
wanted
to
run
a
database
well,
RDS
gives
you
a
database,
on-demand
self-service,
auto
provision
automatically
updated
automatically
replicated.
A
But
if
I
run
the
crunchy
operator
well,
I
can
do
a
cube
kind
of
creative
that
it's
automatically
hooked
up
for
updates.
It's
automatically
doing
the
replication
and
by
the
way,
can
run
on
any
kubernetes
and
by
the
way,
it's
open-source
and
has
advanced
observer
ability,
capabilities
that
RDS
doesn't
so
now.
A
What
was
hard,
though,
was
to
find
the
operators
that
met
certain
criteria
in
terms
of
quality
of
documentation
in
terms
of
how
much
activity
they
have
by
the
community,
that's
behind
them,
and
so
the
goal
of
operator
hub
dot
IO
is
to
create
that
single
place
where
you
can
easily
find
all
of
these
automated
operation
driven
services
that
are
out
there
that
are
also
portable
and
that
can
run
anywhere.
So
if
you
go
there,
you'll
you'll
see
how
to
contribute
to
it.
A
The
goal
of
the
community
or
hope
for
the
community
is
that
this
continues
to
thrive.
I
know:
we've
got
a
couple
of
other
ones
that
are
coming
up
that
have
been
submitted
just
as
of
last
week,
and
we
hope
that
many
of
the
hundred
plus
operators
that
we
know
are
out
there
are
going
to
make
it
here,
but
we're
putting
that
first
bar
of
curation
to
make
sure
that
certain
qualities
are
met
before
they
get
in
and
we're
doing
that
in
collaboration.
A
You
know,
I
always
get
this
question
of
what
about
serverless.
You've
talked
about
all
this
stuff,
but
what
about
serverless
and
to
me
that
question
is
interesting,
because
several
us
is,
you
know,
depends
what
you
mean,
but
the
way
I
like
to
think
about
it
is
from
two
aspects.
Is
a
question
of
scaling
up
the
workload
to
zero
when
there's
no
workload
coming
in
and
scaling
back
up
based
on
an
event-driven
model
and
the
other
aspect
which
is
somewhat
orthogonal,
is
a
question
of
function
as
a
service
building
functions
versus
applications
in
any
other
way.
A
T
native
to
us
was
a
project
that
was
very
promising
because
it
allows
us
to
have
all
the
constructs
to
bring
those
benefits
to
the
containerized
infrastructure
world
in
a
way
that
allows
to
us
to
meet
the
mandate.
I
was
just
talking
about
which
is
full
Portability
and
therefore
hybrid,
and
so
with
the
work,
we're
doing
and
we're
gonna
get
into
a
lot
more
detail
on
that.
With
presentation,
red
beard
is
going
to
be
giving
on
K
native
and
openshift.
A
What
we're
doing
there
is
enabling
server
less
on
any
workload
on
any
event,
so
you're
going
to
have
a
starting
point
set
of
events
that
we're
going
to
bring
in
through
cam.
Ok.
So
if
a
kafka
topic
has
certain
event
on
it,
then
please
start
up
this
application
by
the
way
notice.
I
said
application
not
function,
because
you
can
start
up
a
function
on
the
server
spaces
or
your
spring
boot
app
on
a
server
list
basis.
A
So
so
that's
that's
the
premise
by
which
we're
driving
our
server
strategy
and
I'm
really
excited
about
that
in
general.
Stepping
back
we're
working
on
when
we
say
platform
Services
is
providing
a
set
of
services
that
create
a
platform
that
allowed
the
air
courses
out
there
to
build
portable
services
on
top
of
OpenShift
so
that
we
can
achieve
that
goal
of
portability
and
cloud-like
experience.