►
From YouTube: OpenShift at Mass General Hospital/Partners Healthcare
Description
Wayne Raila and Michael Steigman from Mass General Hospital discuss their production deployment of OpenShift at the OpenShift Commons Gathering Boston on May 1, 2017.
Learn more and see the slides here: https://blog.openshift.com/openshift-commons-gathering-at-red-hat-summit-2017-video-recap-with-slides/
A
If
I
flip,
it
short
I
know
that
Diane's
been
promising
beer
all
day
today's
May
first,
my
mother's
birthday
is
May
1st,
her
name
is
May.
That's
why
she's
up
here
demonstrates
showing
cabbage
she
raised
many
years
ago,
I'm
Layne
Riley.
This
is
Michael
Stegman.
We
both
work
for
the
lab
and
computer
science
at
Massachusetts,
General,
Hospital,
Massachusetts,
General
Hospital
was
part
of
a
larger
corporation
called
Partners
HealthCare.
B
A
A
A
Lab
at
the
theater
science,
we
were
founded
in
1964,
which
means
we've
actually
been
around
for
more
than
25%
of
the
hospitals
200
years
more
than
200
year,
history,
so
I'm
pretty
impressed
by
that
and
we
have
from
the
beginning
gonna.
Do
it
yourself
shop.
We
were
some
of
the
people
who
were
building
not
me
personally,
but
building
running
wires,
building
cabinets
around
teletype
machines,
because
the
teletype
machines
made
too
much
noise
for
the
doctor's
offices
and
I
mentioned
this
too.
A
A
The
early
systems
was
a
time-sharing
system
before
this
people
didn't
know
how
to
share
time
on
all
computer.
So
again
that
sort
of
going
back
way
far,
but
we
also
have
medical
education
programs
in
which
we
create
programs
that
teach
subjects
we
have
training
programs
in
which
we
train
for
the
National
Library
of
Medicine
post
graduate
doctors
in
medical
informatics.
A
A
A
That
has,
of
course,
many
of
the
things
you
need
to
have
reliable,
robust
systems,
multiple
devices
for
front
end
multiples
for
back
ends,
database
system,
standby
databases,
storage,
net,
scalars,
source
code
repositories.
These
are
all
the
things
that
we,
as
developers,
ended
up
having
to
manipulate
to
put
our
apps
out
there,
and
we
could
do
it
because
we
had
a
core
mission
in
the
hospital
that
required
us
to
have
this
large
infrastructure
and
to
have
a
team
of
staff
that
could
maintain
this
so
that
when
we
called
new
things,
we
wanted
to
try
out.
C
A
Open
shift
in
about
six
years
ago,
partners,
the
large
corporate
parent
of
all
these
hospitals
decided
to
buy
a
single
medical
record
system,
epic,
the
system
built
on
mumps
developed
here
at
the
lab,
the
computer
science
all
those
years
ago,
which
meant
we
no
longer
had
that
core
relationship
to
all
the
systems
in
the
hospital
that
were
being
built.
We
were
one
of
many
courts,
we
weren't
the
only
court,
but
we
were
one
of
the
many
course
at
the
came
in
and
said
all
of
those
things
get
put
aside
and
replaced
with
epic.
A
So
there's
a
true
investment
in
this
epic
infrastructure
which
we're
not
part
of,
and
we
no
longer
have
the
need
to
support
all
these
different
environments.
We
no
longer
have
a
need
to
have
the
staff
to
support
all
those
environments,
so
we
looked
at.
How
are
we
going
to
put
out
our
own
work
without
giving
ourselves?
And
we
said?
Well,
let's
go
through
the
corporate
provisioning
process.
The
corporation
has
a
way
for
us
to
apply
for
vm's
they'll,
build
us
a
VM,
and
we
can
then
farm
those
and
so
on.
A
I
started
down
that
path
in
August
of
one
year
a
couple
years
ago,
I
applied
the
application
says.
What
do
you
want
to
use
this
VM
for?
Well
we're
we
don't
know
beforehand,
you
don't
know
when
a
new
trainee
comes
in
and
says,
I
want
to
learn
how
to
do
something.
You
don't
know
what
that
training
wants.
A
A
So
when
I
started
down
this
path
of
going
through
the
corporate
provisioning
process,
Michael
and
his
colleagues
were
talking
about
containers
and
when
I
said
I
don't
know
about
that
stuff
I'm
going
to
build
it.
The
way
I
know
because
I
know
I
can
make
it
work.
This
other
way,
and
three
months
later,
I
had
two
virtual
machines
with
nothing
on
them,
but
Red
Hat
Linux.
A
It
took
three
months
to
get
those
two
machines:
lots
of
meetings,
lots
of
discussion
so
by
the
time
I
had
finished,
building
out
these
couple
of
machines
with
ansible
and
salt
and
all
these
other
tools
for
provisioning.
Michael
said:
oh
hey.
By
the
way
we've
got
this
so
turn
over
to
Michael.
To
tell
more
about
you.
B
B
It
wasn't
quite
so
didn't
quite
happen
in
that
order,
but
we
were
following
I
think
there
was
a
student
in
the
small
group
who
was
focusing
on
what
would
be
the
next
generation
of
infrastructure
at
the
lab,
at
least
what
LCS
we
were
looking
at
docker
and
we're
following
some
of
the
smaller
pad
systems.
We
hadn't
looked
at
OpenShift
the
earlier
versions
of
OpenShift
but
kubernetes
came
out
and
that
started
a
discussion
internally
we
heard
about
the
release
of
kubernetes
and
again
I
I
went
to
Wayne
and
I
said
you
know.
B
You
know
he
gave
us
the
you
know
the
push
off
when
we
talked
about
containers.
I
said:
look
at
kubernetes
and
again:
Kent
Lane
was
like
yeah
kind
of
talk
to
me
when
you're
actually
running
something
on
it,
and
shortly
after
that,
we
discovered
that
Red
Hat
was
was
investing
in
in
kubernetes
and
building
openshift
3
on
top
of
kubernetes,
so
I
went
back
and
I
told
Wayne
and
he
said
yeah
come
back
to
me.
You
know
when
you
got
something
running,
but
that's
really
I'd
like
to
say
that
we
we
did
this.
B
You
very
thorough
analysis
of
the
of
all
the
products
that
are
out
there,
but
the
reality
was:
is
it
is
kind
of
from
up
from
a
developer
standpoint,
that's
kind
of
like
a
dream.
We
saw
kubernetes
and
openshift.
We
thought
this
is
this,
gives
us
pretty
much
exactly
what
we
if
we
could
draw
it
up
in
our
head?
B
So
this
would
enable
us
to
do
this
would
be
like
a
developer's
dream.
Basically,
I,
almost
we've
all
seen.
Demos
is
basically
a
turnkey.
You
know
you
get
your
Django
application
or
you
know
an
application,
and
so
we
started
taking
with
that
and
so
I
think
the
key
pieces
of
the
aside
from
the
the
appealing
nature
of
the
platform
itself,
the
key
pieces
of
why
we
were
looking
at
open
shipped,
was
one
it
offered.
B
An
on-premise
container,
orchestration
solution
and
on-premise
is
important
for
us,
like
Wayne,
said
now
that
there's
this
tremendous
investment
in
epic,
this
is
the
the
central
clinical
data
repository
at
the
organization
there
now
walls
up
around
that
it's
very
difficult
to
get.
There's,
there's
a
process,
there's
a
governance
process
or
a
data
use
data
access
for
the
smaller
labs
that
are
trying
to
do
innovative
work
in
informatics.
B
It's
difficult
to
get
access
to
that
data
and
although
running
services
in
the
cloud
seems
like
be
natural
for
us
to
get
out
of
the
infrastructure
business,
it
actually
would
put
up
a
big
obstacle.
It'd
be
very
difficult
for
us
to
actually
run
spin
up
these.
These
services
accessing
epic
from
the
cloud
so
I'm.
C
A
B
B
But
partners
decided
for
better
or
for
worse
that
they
wanted
to
go
the
the
old-school
route
and
you
know
big
iron,
and
so
they
may
be
the
huge
investment
in
ethic,
which
is
what
a
lot,
which
is
what
most
big
medical
centers
were
doing,
so
that
that
made
it
a
little
bit
more
difficult
for
the
smaller
groups
within
partners
that
typically
had
working
relationships
with
the
partners,
data,
different
partners,
data
sources.
So
there's
now
there's
a
new
data
governance
process
and
it's
a
lot
of
the
people
who
spend
their
time
doing.
B
Informatics
and
partners
spend
most
of
their
trying
trying
to
get
access
to
to
health
data
to
do
their
job.
It's
it's
kind
of
a
kind
of
part
of
the
I
guess
it
comes
with
the
territory
so
and
the
other
big
piece
of
it
was
that
partners
was
already
there.
They
had
a
pretty
big
Enterprise
Linux
footprint.
So
the
group
that
we
were
part
of
our
infrastructure
was
actually
Linux
and
we
were
working
with
the
Linux
guys
at
partners
pretty
closely.
So
that
connection
was
pretty.
You
know
it's
pretty
much
instantaneous
Wow.
B
This
is
good
as
a
cool
product,
Red
Hat
somebody
were
working
with
so
based
on
that
we
reached
out
to
partners
and
they
were
actually
interested
in
containers.
At
the
same
they
were,
they
were
thinking
of
ways
to
use
containers
to
reduce
their
their
DM
spa.
They
were
dealing
I
mean
we
heavily.
You
know,
researchers
firing
up
my
sequel,
databases
all
over
the
place
and
those
are
those
were
proving
to
be
difficult
to
to
maintain
for
four
partners.
B
I
have
so
they
were
thinking
about
containers
to
help
with
with
that
part
of
their
that
part
of
their
work,
but
they
didn't
really
have
a
concrete
use
case.
So
when
we
came
to
them,
we
started
talking
with
a
strategy
group
and
they
were
really
excited
that
it
was
really
was
like
for
us.
It
was
a
dream
come
true,
because
this
is
something
you
gave
us
pretty
much
everything
we
wanted.
You
know
and
partners.
I
asked
was
was
very
receptive.
They
were
really
excited
to
try
this
out
and
the
project.
B
You
know
we
approached
them
in
2015
and
the
project,
basically
the
product
and
the
project
sold
itself.
We
went
we
presented
to
to
the
partner
CIO
twice
once
for
the
for
the
proof-of-concept
and
then
to
get
the
final
approval
back
in
January
and
it's
you
know
all
along
the
way
it
was.
You
know
this
sounds
really
cool,
so
it
was
and.
A
That
fits
them
that
the
these
institutions
view
themselves
as
leaders
in
research
leaders
in
technology,
so
to
be
able
to
support
different
groups,
disparate
groups,
with
different
abilities
of
experience
with
infrastructure
and
bringing
up
systems.
This
promises
a
way
for
them
to
continue
to
support
that
and
maintain
that
reputation.
B
So
these
are
just
some
of
those
basic
kind
of
boring
questions
that
we
had
to
deal
with
before
proof
of
proof
of
concept
for
us,
but
the
lab.
It
was
exciting
to
get
to
this
point
where
we
could
do
a
proof
of
concept,
but
at
the
same
point
there
was
a
question
of
who's
going
to
who's,
going
to
take
on
the
risk
and
we
need
to
buy
licenses.
We
needed
some
consulting
help
from
Red
Hat.
We
needed
hardware
and
again,
luckily
for
us,
the
partners
leadership
decided
that
they
would.
B
You
know
they
would
take
on
that
risk
and
then
just
bet
you
know,
I
mean
the
common
things
you
need
to
do.
These
are
things
that
we
had
nothing
to
do
with
setting
up
the
storage
setting
up
networking
load
balancers
all
that
stuff.
You
know
that
was
pretty
much
all
out
of
our
we
participated
in.
It
was
a
collaborative,
but
we
weren't
doing
the
work
that
was
partners
is
or
was
doing,
the
work.
So
we
really
didn't
face
meant
much
in
terms
of
challenges.
B
B
So
that's
that's
a
big
thing
and
we're
working
through
these
details.
Now
we
have
a
governance
group.
That's
that's
that's
meeting
weekly
to
go
over
these
issues
and
the
next
is
like
Wayne
was
talking
about,
allowing
that
the
types
of
tinkering
that
that
we've
done
for
so
many
years
in
this
new
environment-
and
it's
you
know
we
don't
you
don't
want
to
you,
don't
want
to
introduce
a
new
framework
and
then
have
the
same
type
of
provisioning.
You
know
bottleneck
that
word
that
we're
dealing
with
now
of
virtual
machines.
That
really
wouldn't
be
a
step.
A
B
This
promise
is
a
you
know,
a
way
to
maybe
get
in
on
the
process
and
stay
on
top
of
applications
that
are
running
whereas
right
now,
it's
sort
of
like
you
know,
can
sometimes
be
the
wild
wild
west,
because
you
provision
your
application,
you
say
what
you're
going
to
use
it
for
and
then
you're
up
and
running.
You
can't
get
your!
You
can't
expose
your
stuff
outside
the
firewall,
but
you
know
still:
it
can
be
a
little
bit
scary
when
you're
you're
you're
charged
with
securing
the
huge
clinical
data
store.
B
B
They
also
do
a
lot
of
work
at
MGH
supporting
research
enterprise
their
it.
So
these
will
be
the
three
initial
groups
there,
so
this
won't
have
any
impact
on
on
epic
on
any
of
the
clinical
systems.
Yet
those
are
really
tightly
prescribed.
Hardware,
hardware,
recommendations
and
all
big
organizations
like
epic
or
sorry
like
partners
follow
those
they
like
to
follows.
You
like
to
stay
close
to
the
you
know,
tried
and
tested
configurations,
and
so,
although
partners
does
have
their
I
think
there
are
75
people,
something
like
that.
B
B
A
B
B
A
Using
I
mean
it's
so
much
fun
to
use
it
it's
you
know
you
want
to
do
something.
You
create
a
project.
You
bring
up
all
these
different
pods
that
you
already
know
how
to
bring
up
but
you're,
just
wiring
them
in
different
ways,
and
if
you
don't
like
it,
you
toss
it
away
and
all
those
resources
are
reclaimed
for
some
other
use.
A
So
as
a
developer,
I
didn't
like
having
to
do
load,
balancing
and
hardware
setups
and
make
a
DNS
entry
for
all
of
the
systems
we're
using
and
the
routing
and
all
those
things
you
have
to
take
care
of
I
had
to
do
it
to
make
my
apps
work,
but
it
wasn't
where
my
interest
was.
This
allows
me
to
focus
on
what
is
the
application
doing.
D
F
B
That's
how
we're
getting
you
know
so
we're
synchronizing,
LDAP
groups
or
partners,
I
acted
or
as
Active
Directory
groups
into
into
open
ship.
We're
trying
to
tie
that
in
with
partners.
Partners
has
its
own
homegrown
app
to
manage
directories
which
would
give
us
a
little
bit
of
the
audit
trail
piece.
So
we're
trying
to
tie
that
in.
But
it's
probably
not
going
to
be
part
of
our
initial
implementation.
But
it
is
a
that
is
a
concern
and
ongoing
concern.
Essentially.
B
There
is
there's
a
there's,
a
tool:
that's
part
of
open
shipped.
It
will
will
synchronize
the
people
from
Active
Directory.
So
that's
what
we're
using
but
there's
a
tool
that
we
have
that
allows
a
Department
administrator,
allows
you
to
grant
a
person
admin
privileges
over
their
department
and
then
allows
that
person,
then
to
add
people
today
to
that
group.
B
C
E
So
our
lab
has
a
lot
of
genetic
data
and
we've
been
sequencing
human
genomes
there
for
quite
some
time
and
so
we're
presenting
that
is
a
set
of
micro
services
for
diagnostic
applications
through
our
partners
and
integrating
that
into
the
EHR.
But
it's
it's
developers
that
are
doing
this
at
the
guidance
of
researchers
right.
E
So
the
researchers
are
giving
us
the
requirements
of
what
it
is
they
want
and
the
developers
are
actually
implementing
that
and
we
are
seeing
that
it's
a
much
quicker
development
time
that
it
does
support
the
agile
process
and
it's
also
much
more
efficient
use
of
resources
in
terms
of
BMS
and
computing.
So
yeah
we're
totally
on
board
with
OpenShift
and
this
model.
B
B
Hundred
and
two
hundred
and
thirty
million
dollars,
or
something
like
that
between
Brigham
and
Brigham
and
Women's
and
Master
animals,
so
there's
there's
a
lot
of
research
money
and
there
are
a
lot
of
people
doing
small
little
projects.
They
have.
You
know
two
developers
that
are
you
know
that
are
firing
up.
You
know
local
docker
images
to
do
this
or
to
do
that.
So
that's
you
know
those
also
that
that
will
also
be
the
audience
or
potential
users
of
the
system.
They're.
Not
yet
right.
Now,
it's
yet
into
the
lab.
B
G
C
G
We
have
a
lot
of
young
guys
coming
to
us
working
in
the
area
of
integrating
endpoint
devices
in
the
healthcare
industry.
Your
patients
rolling
in
you
have
heart
monitors.
You
have
blood
pressure
monitors
all
these
RI
OTS
containers
is
a
great
way
of
gathering
data
from
them
and
working
are
you
actually
working
in
the
space
at
all?
Thinking
about
does
open
you
cater
to
that
market
within
the
mg8
ecosystem
and
the
reason
I'm
asking
is
obviously
you
set
a
precedence
for
the
rest
of
the
healthcare
industry.
Yeah.