►
Description
Case Study: OpenShift @ GE Digital - Jackie Chute, Timothy Oliver, & Jay Ryan at OpenShift Commons Gathering Seattle 2018
https://commons.openshift.org/gatherings/Seattle_2018.html
A
A
Hi,
as
Diane
said,
we're
we're
from
GE
digital
and
we're
here
to
talk
a
little
bit
about
our
journey
to
openshift
what
we've
experienced
along
the
way
and
kind
of
maybe
some
some
helpful
discourse
about.
Maybe
you
can
help
us
and
we
can
help
you
and
we
can
help
them
make
the
product
better.
So
without
further
ado,
I'll
introduce
Tim,
Oliver
hi.
B
B
C
C
Containers
before
containers
for
containers
and
I
did
automation
as
a
service
for
about
five
years
at
GE,
and
then
I
joined
the
container
cloud
service
and
I
do
with
all
the
Terra.
For
me,
stuff
that
happens
before
we
do
our
open
shift
install.
So,
although
I
am
really
excited
about
for
now,
I
I
don't
have
a
job
anymore.
Well,.
A
We'll
talk
about
that
a
little
bit
later
and
why
we're
excited
about
that
I
think
I
think
there
are
many
goes
all
out
of
jobs
wherever
you
are
waitin.
My
name
is
Jay
Ryan
I'm,
a
staff
infrastructure,
architect,
adji
I'm,
not
sure
what
that
means.
But
you
know
how
titles
go
so
I've
been
working
at
G
for
just
two
years
and
I've
been
working
with
containers
and
kubernetes
since
March,
so
I'm
very
new
to
this
space
and
excited
to
be
a
part
of
it.
A
So
a
little
bit
about
our
team
ge
is
a
complex
organization,
and
so
we
we
sit
on
what
we
call
the
container
cloud
team.
It's
very
small
team,
we're
half
of
the
team
here
today
represented
here
and
we're
dedicated
to
running
kubernetes
as
a
service
at
GE,
and
so
we
sit
in
our
core
tech
division,
which
we'll
talk
about
a
little
bit
later,
but
that's
basically
IT
as
a
service
for
all
of
GE
in
the
GE
business
units,
sims
gonna
talk
a
little
bit
about
what
GE
is
and
what
we
do.
B
One
of
the
things
that's
really
cool
about
GE
and
about
working
for
GE
I've
been
a
GE
since
95.
So
yes,
these
grace
tell
a
story,
but
ge
is
a
huge
company
and
we're
all
over
the
world.
I
mean
take
that
globe
and
stretch
it
out
and
that's
where
we
are
pick
a
country
and
we're
probably
there
have
a
presence.
So
ge
is
the
kind
of
company
that
builds
the
things
that
make
things
like
cities
work.
B
That
is
now
no
longer
a
part
of
us,
but
that's
another
story,
so
we're
all
over
the
world
that
we're
in
a
lot
of
different
businesses
we're
in
healthcare
we're
in
we
do
things
with
the
obviously
the
Department
of
Defense
were
in
aviation.
We
make
engines
and
things
like
that.
So
in
some
of
our
businesses,
that's
really
highly
regulated
and
then
in
other
things,
it's
it's.
What
you
would
think
about
is
IT.
B
So
it's
it's
pretty
cool
the
back
in
probably
about
2010
we're
dealing
with
a
lot
of
things
that
you
probably
have
heard
as
an
enterprise.
Every
developer.
Every
developer
team
thinks
that
their
application
is
special
and
requires
their
own
server,
and
so
you
go
and
buy
that
hardware
it
comes
in.
It
sits
on
the
dock
and
then
it's
underutilized
because
their
app
is
the
only
thing
that's
using
it,
but
then,
at
the
same
time,
they're
complaining
about
compute
costs.
Oh
and
everything
cost
too
much.
It
just
costs
too
much.
B
You
got
to
lower
the
cost,
and
so
I
was
part
of
a
team
that
was
a
part
of
our
initial
entry
into
the
cloud
and
we're
looking
at
these
things
that
we're
trying
to
just
get
the
cost
down.
So
we're
looking
at
all
these
apps
and
we're
these
underutilized
servers
and
all
of
this
and
we're
thinking.
How
can
we
do
this
better?
Well,
along
comes
docker
in
containers
and
it
seemed
like
the
perfect
solution,
but
then
we
have
you
know
we
got
a.
We
had
a.
B
We
had
some
changes
if
you've
been
watching
any
of
the
wah
sri
stuff.
We've
had
some
changes
in
the
way
ge
operates.
We
won't
go
into
that,
but
there's
been
changes
in
management
levels
and
things
like
that
right.
So
some
of
our
initial
efforts
got
sorted,
but
we
started
down
the
docker
road
with
docker
swarm
and
we
had
some
successes
with
that
and
then
we
started
with
some
of
the
other
technologies
and
we
had
some
successes.
We
found
ourselves
actually
developing
the
pieces
around
kubernetes
and
scoober
Nettie's,
particularly
in
the
earlier
days.
B
Kubernetes
is
the
bear
to
install.
It
was
just
huge
and
it
was
unwieldly
and-
and
it
really
took
something
to
install
that
thing
and
to
keep
it
running
right,
but
we
were
building
services
around
that
and
then
we
saw
OpenShift
an
open
shift.
Everything
that
we
were
trying
to
build
open
ship
had.
So
what
we
were
doing
didn't
make
any
sense
anymore.
So
then
it
was
like
well,
okay,
how
do
we
just
if
they're,
if
they
already
have
it,
then
why
are
we
building
this?
B
We
can
just
get
it
from
them
and
then
of
course,
Red
Hat
does
all
of
the
great
stuff
that
as
an
enterprise,
you
need
with
the
support
and
everything
like
that.
So
we
got
down
the
road
of
containers.
We
had
a
lot
of
false
starts,
but
we
had
a
lot
of
opportunities
for
containers
and
then
openshift
and
kubernetes
happened,
and
we
got
started
with
that.
So
so.
A
So
yeah,
so
if
we
take
a
step
back
a
little
bit
and
talk
to
you
a
little
bit
about
core
tech
about
IT
as
a
service
at
G
and
how
that
works,
it
will
kind
of
make
sense
about
why
we
approached
openshift
the
way
that
we
did
so
half
a
million
people.
Core
tech
is
supporting
in
over
170
countries,
7,000
enterprise
applications
and,
as
Tim
mentioned
migration
to
the
cloud
app
monetization
300,000.
A
So
thousand-plus,
that's
I,
think
that's
old
data,
probably
thousand
plus
applications
migrated
so,
as
Tim
said
we're
around
the
globe.
So
one
of
the
things
that
makes
cortex
special
is
that
we're
really
a
business
partner,
we're
partnering
with
our
business
units
together
to
help
to
help
them
migrate
to
the
cloud
to
help
them
optimize
their
applications.
And
we
really
do
that
through
like
a
product
centric
approach.
So
we
are
a
service
provider
for
GE,
and
so,
if
we're
not
successful,
GE
is
not
successful.
A
B
One
of
the
things
that
that
I'm
doing
actually
it's
taking
OpenShift
and
writing
a
layer
around
it.
That
is
specific
to
a
self-service
kind
of
a
model
that
we
want.
Our
users
to
use
and
I'm,
using
calling
making
cause
to
the
open,
shipped
API.
In
order
to
do
things
like
create
accounts
and
assign
groups
and
roles
and
all
that
to
users
and
groups,
and
things
like
that.
We're
handling
that
metadata
side
of
it
outside
and
that's
the
piece
that
I'm
riding
around
around
open
shift.
A
So
open
shift
at
GE,
so
we
took
like
I,
said:
core
tech
takes
a
product,
centric
approach,
so
we
kind
of
took
a
product,
centric
approach
when
approaching
open
shift,
and
so,
as
you
heard
already
this
morning,
the
reasons
why
open
shift
open
shift
checked
all
of
those
boxes
out
of
out
of
the
gate.
For
us
kubernetes.
A
The
hard
way
is
a
hard
way
right
so
and
now
open
shift
what
we
find
out
that
in
open
shift
3
that
was
open
shift
the
hard
way
and
as
as
Clayton
as
a
Clayton
to
Derrick,
stated
right,
they're,
making
things
a
whole
lot
easier
for
us
going
forward.
So
they're
checking
those
boxes
they're
a
step
in
front
of
us
the
whole
way
and
that
that
just
excites
us.
A
So
when
we
have
conversations
with
our
security
teams,
we
can,
we
can
tell
them
that
you
know
we're
bringing
secure
platform
out-of-the-box
to
start
with
a
little
bit
about
kind
of
the
background
that
Tim
mentioned
10,000.
When
we
did
a
small
survey,
a
small
sampling
of
what
our
customers
were
doing
with
regards
to
containers
and
orchestration,
we
found
some
moderately
surprising
numbers,
but
we
found
tens
of
thousands
of
containers
running.
A
We
found
thousands
of
docker
Damons
and
around
a
hundred
or
so
orchestration
engines,
and
so
that
was
like
I
said
just
a
small
sampling
that
was
maybe
a
20
or
30
percent
sampling
of
the
environment
that
we
had
access
to
dive
into.
So
we
saw
our
customers
with
a
need
and
again
with
the
product,
centric
approach
right,
we're
gonna
build
a
product
to
help
them
solve
that
need
because
kubernetes
is
hard
to
operate
when
we
did
when
we
surveyed
some
of
these
customers
the
what
was,
what
was
the?
What
was
the?
What
was?
B
And
all
of
them,
without
without
fail
just
about
everyone
that
we
talked
to
so
here,
just
to
back
up
a
little
bit,
they
started
doing
kubernetes
or
docker
or
whatever
the
container
strategy
was,
they
started
doing
it
on
their
own
because
we
didn't
have
a
corporate
service
offering
for
them.
So
it
started
happening
at
a
lower
level,
but
when
they
heard
that
we
were
doing
open
shipped,
not
every
one
of
them
said
great,
because
I
don't
want
to
run
this
I,
don't
want
to
run
open
ship
I
want
somebody
else
to
do
it.
B
A
We
don't
have
pretty
pictures
to
show
you
about
all
of
our
cool
orchestra
architecture,
because
it's
open
shifts,
architecture
and
again,
one
of
the
things
that
made
this
such
a
great
choice
was
they.
You
know
they
lay
out
reference
architecture.
They
tell
you
how
this
should
work
in
production,
and
so
we
followed
their
model
with.
You
know:
high
available
masters
at
C,
D
and
M
for
infra
we're
using
four
node
OCS
clusters
for
Gluster.
B
Of
the
things
about
this
is
that
really
I
mean
orchestration
is
not
sexy.
It's
just
running,
it's
just
running
containers,
but
that's
exactly
what
you
want.
You
don't
want
it
to
be
out
there
visible,
saying:
hey,
Here,
I
am
running
containers,
you
just
want
it
back,
they're
doing
the
job
and
that's
what
we've
seen
so
far.
It's
been
up
and
our
users
are
using
it
and
it
just
the
thing
just
goes
yep
so.
A
We
started
with
the
internal
registry,
which
is
fitting
most
of
our
needs.
Actually,
we
also
have
some
registries
that
are
part
of
some
of
the
other
work
that
that
some
teams
were
doing
that
we're
plugging
those
in
as
well.
We've
set
a
standard
that
that
says
for
a
registry
to
be
you
know,
white
listed
and
in
our
environment.
It's
gonna
have
to
check
these
security
boxes
scanning
and
the
like.
B
A
A
Sorry,
the
OpenShift
CTO
came
up
here
and
talked
earlier
about
you
know
either
building
mega
clusters
or
building
lots
of
little
clusters,
and
if
we
would
have
had
a
conversation
with
him
about
six
months
ago,
we
probably
would
have
went
lots
of
little
clusters
right
way,
which
we're
kind
of
going
down
that
path.
Now
it
turned
out
a
lot
of
our
customers
have
workloads
that
require
that
require
their
own.
Their
own
clusters,
as
Tim
mentioned
lots
of
lots
of
government
lots
of
regulation.
If
there's
a
three-letter
acronym,
it
exists
at
GE,
so
so
yeah.
A
So
we
have
three
types.
We
have
shared
clusters,
basically
for
test
and
dev
type
workloads.
Where
you
can
come
and
play
you
can
learn
OpenShift.
You
can
bring
your
test
workloads
to
get
it
working
and
then
we're
offering
dedicated
nodes
in
that
environment
so
for
customers
that
maybe
need
that
next
step.
That
say:
hey,
don't
I,
don't
want
I,
don't
want
customer
Y
from
business
unit,
X
being
able
to
affect
my
availability.
A
They
can
run
their
own
nodes
on
that
same
platform
and
then
dedicated
clusters
is
ultimately
the
the
kind
of
route
that
we're
doing
so.
The
way
that
the
way
that
we're
really
doing
this
is
through
automating
the
lifecycle
of
the
clusters
so
from
soup
to
nuts
we
can
stand
up
clusters,
install
them
patch
them
in
in
the
way
that
that
Clayton
and
Derek
we're
talking
about
for
the
for
the
next
version.
So
it's
great
to
hear
that
because
thank
you,
Clayton
and
Derek
for
automating
us
out
of
jobs.
A
I
really
appreciate
that
and
I
actually
say
that
very
sincerely,
because
what
we
need
to
focus
on
and
what
Clayton
I
think
mentioned
several
times
is.
We
need
to
focus
on
like
driving
business
value
right
and
the
less
time
we're
spending
getting
the
cluster
running
and
and
and
automating
the
cluster.
The
more
time
that
we
get
to
spend
solving
business
problems.
The.
B
Less
time
we
spend
with
with
development
teams
app
development
is
talking
about
infrastructure
and
servers,
and
things
like
that,
when
really
all
they're
trying
to
get
to
is
I
want
to
run
my
application
and
I
want
to
run
it
with
all
the
capacity
that
I
need
and
I
wanted
to
scale
and
I
wanted
to
be
up
all
the
time
we
can
have
that
conversation
and
we
obfuscate
the
infrastructure,
then
we're
winning.
So.
A
A
Can
have
their
we
have
their
time
back
yeah.
So
our
back
one
of
the
things
that
G
is.
We
have
really
complex
teams,
teams
that
span
different
organizations.
You
know
a
dotted
line
teams,
and
so
when
we
want
to
build
when
our
customers
want
to
build
our
back
models
and
our
back
controls
inside
their
projects,
it
can
get
a
little
messy.
So
one
of
the
things
that
we
want
to
do
is
build
out
custom.
Our
back
rolls
that
can
take
away.
You
know
certain
certain
access
and
just
enable
our
teams
to
work
better
and
I.
A
Those
are
things
that
I
think
are
might
help
the
community
as
well.
There
are
default,
our
back
rolls
that
come
with
the
cluster
and,
of
course
you
know
you
can
change
our
back
to
your
heart's
content,
but
that's
something
that
I
think
you
know
it
may
be
a
set
of
practices
around
that
that
we're
looking
to
to
help
with
I
guess,
identity
as
well
so
open
ID
is
what
we
use
in
the
cluster,
and
we.
A
The
user
info
endpoints
an
open,
ID,
aren't
necessarily
supported
yet
an
open
shift,
and
so
there's
lots
of
balls
in
the
air
to
get
all
of
our
our
identity,
matched
up
and
synced
up
throughout
throughout
G
in
the
cluster.
So
that's
one
of
the
things
that
that
we're
working
on
too
and
that's
something
that
Tim's
working
on
as
a
part
of
as
a
part
of
his
his
front
end
yeah
a
tendency
is
a
thing
that
is
starting
to
get
talked
about,
I
guess
in
the
community.
A
Let's
move
we're
really
interested
in
in
the
working
group
multi-tenancy
that
the
kubernetes
community
has
we're
building
business
constructs
and
tenant
constructs
outside
of
the
platform
today
and
using
the
product.
The
project
tenancy
that
that
exists
today,
but
I.
We
think
there's
a
bigger
story
around
that
that
we're
following
as
well
in.
B
Each
of
our
business
that
they're
all
pretty
discrete
businesses,
the
thing
that
they
have
in
common
is
that
they
are
GE,
but
then
it's
pretty
much
stops
there,
so
that
the
shared
hardware
and
shared
platforms
and
things
like
that,
we
really
need
the
isolation,
because
nobody
wants
to
deal
with
or
even
they
don't
want
to
deal
with
anybody
else.
They
want
to
be
the
only
ones.
A
Ingress
again,
we
have
application
teams
that
have
complex
needs,
and
so
OpenShift
has
routes
and
kubernetes
has
ingress
and
those
things
are
coming
together.
So
we're
interested
in
that
story
and
how
that
how
that
evolves
and
how
we
can
support.
You
know
complex
ingress
policies
in
the
cluster,
so
the
future
for
us
is
basically
just
listen
to
what
the
guys
talked
about
this
morning,
because
that's
the
stuff
that
we're
excited
about
we're
rivet.
A
A
Ephemeral
build
environment,
so
one
of
the
things
around
build
is
it
requires
privilege.
So
we
want
to
build.
We've
talked
to
the
open
ship
dedicated
team.
A
little
bit
and
they've
talked
to
us
about
how
they're
doing
that,
and
it's
it's
very
interesting.
So
they
build
physical,
Hardware
and
tear
down
every
time
they
do
build.
So
I,
don't
think
we're
gonna
get
quite
there,
but
an
ephemeral,
build
story
is
definitely
in
the
cards,
so
core
OS
in
cryo,
no
further
explanation
needed
there.
Why
we're
going
in
that
direction?
A
And
the
biggest
thing
for
us
is
our
customers
are
innovating
GE
is
innovating.
Ge
has
always
been
innovating,
so
the
customers
that
we
have
in
the
environment
today
are
teaching
us
about
kubernetes
and
teaching
us
about
OpenShift
and
asking
about
operators
and
asking
about
helm
and
and
wanting
to
know
how
they
can
get
in
at
the
ground
floor
and
build
their
applications
and
their
platforms
on
top
of
OpenShift
and
integrate
and
build
them
in
a
cloud
native
way.
So
that's
that's
one
of
the
most
exciting
things
for
me.