►
From YouTube: OpenShift at Swiss Railways
Description
Baltisar Oswald from Swiss Federal Railways & Christoph Eberle from Red Hat discuss SBB's 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
I
didn't
have
to
do
those
slides
I,
you
know
so
this
has
been
really
amazing
and
we'll
do
it
again
in
Austin
and
December
at
coupe.
Con
and
I
have
a
funny
feeling
we'll
do
it
again
somewhere
in
between
there
and
then
Austin
is
in
December.
So,
if
you're
interested
in
presenting
your
story
as
a
customer
or
a
service
provider,
let
me
know
the
other
thing
is:
we've
also
were
so
successful.
A
They've
now,
given
me
my
own
mini
pod
in
the
dev
zone,
so
for
two
in
the
next
two
and
a
half
days
after
this
I
will
be
standing
by
a
monitor
somewhere
in
the
dev
zone
by
myself.
So
please
join
me
and
please
join
the
openshift.
Commons
I
can
walk
you
through
the
whole
process
and
get
you
on
boarded
while
I'm
here
and
if
you
bring
me
an
espresso
will
probably
help
to
facilitate
that
process.
A
So
here
we
go,
could
I
get
someone
in
the
back
to
shut
the
door
there
we
go
so
after
lunch
is
always
problematic.
I
don't
want
any
of
you
to
fall
asleep.
That's
why
we
ran
out
of
food,
so
you're
still
hungry
for
more
and
these
guys,
SPB
or
the
Swiss
railways
didn't
bring
any
chocolate
to
dessert,
but
but
but
they
did
bring
their
bulky
cloud.
A
B
So
welcome
to
our
speech:
I
am
Baltazar
Oswald,
my
name
I'm
from
GSB,
and
because
my
colleague
Chris
Weber,
away
from
Red
Hat
rattle
a
man
from
s
to
B,
and
we
essential
for
us
to
be.
We
would
like
to
talk
a
little
bit
about
Balti
cloud.
It's
my
personal
cloud.
I
am
really
proud
of
it.
Now,
what's
behind
the
Baltic
cloud,
it's
open
safe,
and
why
do
we
need
open
shift?
B
What
are
we
now?
So
it's
very
well
I
know:
I,
don't
think
that
you
know
it,
but
it's
a
public
tranportation
company
in
Switzerland,
one
of
the
the
biggest
one
we
have
more
or
less
30k
employees,
1.3
K
ITN
plays
and
more
or
less
400
software
developers,
and
we
do
a
lot
of
software.
We
could
use
a
lot
of
software
and
well,
we
were
really
traditional.
B
We
have
a
traditional
business,
but
we
now
have
to
change
from
a
traditional
business
to
a
business
that
is
a
little
bit
more
faster
with
the
time
to
market
became
an
important
thing
for
us
and
why?
Because
which
railways
want
to
become
the
journey
into
great
Institute
means
that
you
have
we're
going
to
provide
products
for
customers
to
give
from
the
door
to
the
door
journey,
and
we
will
give
all
the
the
services
that
the
customer
needs
to
go
from
A
to
B.
B
What
does
it
mean?
The
utilization
digital
transformation
is
one
of
the
big
things
and
we
thought
with
the
traditional
IT
we
weren't
able
to
deliver
the
products
as
fast
as
a
customer
wants
it,
and
the
second
part
is
always
good
to
culture.
I
mean
we
wear
and
we
are
a
traditional
enterprise
with
a
traditional
enterprise
culture.
So
we
want
to
move
from
a
traditional
to
the
digital
enterprise
with
the
digital
culture.
It's
not
so
easy
to
do
it
and
why
we
started
with
the
technology.
Why?
B
Because
technology
is
a
big
enabler,
and
so
we
started
with
it
with
a
project,
a
little
project
to
set
up
an
environment
for
developers
to
be
able
to
develop
within
hours
within
minutes
in
production,
to
be
able
to
do
it
as
a
self-service
and
so
on
and
little
project.
We
did
it
with
retic
together,
and
that
will
give
everything
like.
C
Say
I'm
from
Red,
Hat
and
I
have
the
pleasure
actually
to
work
with
these
guys
for
approximately
the
last
two
years.
Now
it's
really
pleasure
and
basically
what
we
did
is
we
really
had
an
adoption
process
of
open
shipped
as
it
started
very
small.
Those
kind
of
we
call
it
an
import
project
so
something
which
isn't
really
on
the
radar
of
the
management,
and
you
have
some
reasoning
in
it
and
what
we
actually
did
or
what
they
did
is
percentage
said.
Okay,
we
have
to
follow
this
whole
digitalization
friends.
C
We
cannot
do
this
with
the
technology
stack.
We
have
in
place
right
now,
so
the
first
thing
that
it
is,
they
defined
a
new
technology
stack
with
all
the
nice
things
like
spring
boot
and
hazel
calls
and
angular,
and
that
okay,
this
is
the
set
of
technologies.
We
want
to
use
to
build
our
future
our
next
generation
of
applications.
Then
there
is
all
about
the
packaging
format
for
for
these
technologies
and
decided.
C
Okay,
containers
is
the
thing
we
have
to
do
in
the
future
and
if
you
talk
containers,
you
talk
container
orchestration
as
well,
and
they
decided
looked
at
the
market
decided
very
early.
That
kuba
is
the
right
choice
to
do
at
this
point
in
time,
probably
not
because
it
has
the
most
features
at
that
point
in
time,
but
because
of
the
community
behind
kubernetes,
it
was
very
vibrant.
It
has
where
it
has
still
very
good
stakeholders
in
the
community
and
the
roadmap
of
communities
was
really
making
their
minds
again.
C
Is
that
okay,
we
are
not
a
cloud
or
a
platform,
provided
we
don't
want
to
build
it
on
our
own,
so
that
we're
looking
on
the
market.
What
is
out
there?
What
could
really
help
us
or
what
vendors
could
help
us
in
pieces
where
we
fortunately
came
into
the
game
and
first
things
you
do
when
you
want
to
adopt
a
new
technology?
You
do
a
POC,
so
we
said:
okay,
we're
going
to
do
a
POC
a
couple
of
weeks.
Some
guys
from
relative
are
on
site
to
help
them
and
it
was
really
about
looking.
C
How
can
the
technology
fit
into
our
environment
and
what
can
it
help?
What
can
open
fist
help
on
top
of
docker
in
communities?
So
what
well
you
asked?
Can
we
get
out
of
it
and
then,
after
two
weeks,
Baltazar
called
our
sales
rep
and
said
Mr
Graham.
Unfortunately,
we
have
to
stop
the
PLC
and
we
were
all
getting
white
and
asking
why
the
rest.
Actually
we
want
to
do
the
next
step
already
after
two
weeks.
C
All
we
want
to
do
a
pilot
and
the
pilot,
in
definition,
in
terms
of
switch
railways,
is
taking
15
applications,
15
projects
which
are
currently
developed,
move
them
onto
openshift
and
bring
them
into
production
into
terms
of
one
year.
So
this
was
the
pilots
code
that
we
take.
15
applications
put
them
on
open
shift,
bring
them
into
production
and
harden
they're
open,
50
platform.
C
During
this
journey,
then
I
think
5
6
7
months
later
it
was
already
45
projects
who
wanted
to
participate
and
run
on
open
shift
and
the
first
that
actually
we
already
we
want
to
go
live
now.
So
after
six
months
we
had
a
production,
ready,
environment,
50,
applicator
or
50
projects.
Actually
are
wanting
to
go
on
to
open
shift
and
really
where
we
went
into
into
production.
C
We
had
a
lot
of
interactions
in
these
times
and
there
was
still
something
is
lacking
in
the
product,
so
it
was
a
constant
closed
loop
of
iterations,
where
we
worked
together
to
increase
their
operations.
Processes
as
well
I
mean
bringing
something
into
operation
to
making
production
ready,
isn't
always
easy.
It
is
a
lot
of
new
technology,
a
lot
of
stuff
in
it.
So
this
was
really
a
challenge
as
well.
C
If
we
look
at
what
we
have
today
in
terms
of
environment
and
it's
two
clusters,
basically
one
on
premise:
that's
the
left
one
running
on
bare
metal.
They
use
cluster
4
for
storage,
on
both
environments
on
the
public
cloud
as
well
as
on-premise.
It's
around
80
nodes,
bare
metal,
big
nodes,
a
lot
of
CPUs
900
course,
the
internal
environment,
around
9
terabytes
of
RAM,
and
it
stretched
across
to
2
sides,
basically
and
as
well,
an
environment
on
AWS.
C
C
These
two
clusters
state
today
or
yesterday,
probably
today,
even
more
it's
around
2000
containers
which
are
really
running
and
operated
on
on
these
open
drift
environments,
and
if
you
look
at
most
of
them
or
on-premise,
because
there
you
have
the
big
capacity
there,
all
the
heavy
lifting
is
done
and
on
public
cloud
it's
around
10%
of
these
containers,
which
run
there
Baltazar
we'll
talk
in
out
through
a
couple
of
the
use
cases
that
you
really
see
what
kind
of
applications
it's
not
all
just
making
our
stuff.
So
it's
really
importantly,
application
as
well.
B
Only
a
few
words
because
the
other
guys
they
have
a
demonstration
and
I
think
that's
more
interesting
tortoise
what
I
have
to
say,
but
overall
we
have
more
or
less
100k
requests
per
sector
per
minute,
and
so,
with
this
implication,
there's
one
of
the
most
popular
applications,
which
would
five
million
downloads.
A
lot
of
people
are
using
it
and
you
have
here's
kindling
when
you
can
go
from
two
and
you
have
you
can
buy
tickets
and
so
on.
So
we
have
more
on
peak
times.
B
B
B
B
B
And
a
third
one:
it's
this
one
here,
wait
for
a
lot
of
displays
and
this
this
place
are
in
future
in
two
or
three
weeks
or
two
months
delivered
by
open
fixes
or
the
data
is
delivered
by
a
project
and
to
did
the
the
well
D
central
applications
or
phase
one
that
the
computer
suits
that
delivers
to
the
display
of
these
slave
ports
or
docker
containers,
not
in
Oakland
safe
inside,
because
the
docker
delivered
by
satellite
6.3.
And
now
we
are
ready
to
software
engineers.
B
D
Things
are
running
on
the
cluster,
pretty
important
things.
If
it
goes
down,
we
have
millions
of
people
not
able
to
use
our
services,
and
we
are
in
all
the
newspapers
the
next
day
or
the
same
day.
So
I'm
part
of
the
team
that
operates
the
clusters
and
thus
engineering
and
and
upgrade
Tobias,
is
in
the
same
team.
He
will
talk
about
the
CI,
CD
integration
and
the
developer.
Experience.
I
will
talk
about
some
of
our
things.
D
We
do,
especially
because
we
heard
this
morning
already
what
you
guys
do
with
objects
and
that's
pretty
similar
with
us
most
of
the
time.
We
do
the
same
things
as
you.
We
do
edge
hae
aware
of
the
clusters.
We
also
do
cluster
SS,
for
example,
so
I'm
going
to
talk
about
a
few
things
that
we
do,
especially
that
are
invented
where
we
are,
or
we
did
some
things
that
helped
us,
and
we
want
to
show
you
these.
So
the
first
thing
is
monitoring
is
crucial.
Of
course
it
is.
D
But
if
you
operate
in
this
size
in
this
short
time,
you
need
to
ask
good
monitoring.
So
we
will
show
you
something
about
this.
Then
the
most
critical
part
are
updates
and
we
heard
about
the
a
B
deployment
rolling
update
of
ABS.
But
what
about
open
gist
that
for
me
the
most
important
part?
How
do
you
upgrade
the
cluster
without
breaking
it,
because
this
is
the
biggest
risk
you
could
crash
everything
with
just
one
command
and
that's
what
we
heard
this
just
before
lunch,
the
Federation,
for
example.
D
D
We
can't
do
they
request
manually
because
we
would
not
do
anything
else
so
update
I
like
to
send
them
weather
does
not
like
it
because
because
I
always
say
hey
guys,
I,
don't
trust
your
PlayBook,
you
all
the
time
you
have
things
in
there
like
restored
all
the
et
de
t,
CDs
at
the
same
time
or
restart
to
master
api's
on
all
the
monsters.
From
the
same
time,
and
all
the
time
I
say,
hey
guys,
we
need
to
do
in
place
upgrade
this
break
our
production.
So
what
did
we
do?
D
We
created
a
little
tool
where
we
that
we
run
on
the
loading
cloth.
For
example,
we
had
a
little
problems
with
display,
so
maybe
it
laggy
just
a
little
tool
that
checks
everything
inside
the
cluster
and
it
creates
quite
a
heavy
load.
So
we
can
simulate
our
production
environment
on
the
def
cluster
in
text,
for
example,
our
masters
available
is
ET
cv,
ok
is
DNS.
Ok,
we
do
HTTP
check
from
all
the
levels,
so
we
do
it
in
the
pot
on
the
master
or
node.
D
D
D
So
you
just
die
in
the
background.
I
got
error,
see
you
when
I'm
getting
more
errors,
lawyers,
whatever
I'm
down
here,
I'm,
getting
a
list
of
what
errors.
What
did
crash
so
at
the
moment,
I
just
broke
everything,
so
it
crashes
on
every
point,
but
this
little
tool
helps
us
to
check
if
the
playbooks
will
have
an
impact
on
our
production
or
not.
So
that's
just
one
little
thing
and,
as
you
saw
in
the
presentation,
it's
open
source
so
just
check
it
out.
D
The
second
thing
I
talked
about
was
the
self
service.
As
I
said,
we
like,
we
don't
want
to
manually,
handle
all
the
requests
from
our
projects.
For
example
quota
we
have.
If
we
began,
we
define
the
quota
set,
that
with
default
and
most
of
the
projects
reach
limits
very
fast,
so
they
want
to
increase
it,
and
we
did
this
manually.
We
logged
into
the
server
and
update
this,
and
we
didn't
like
this,
but
we
don't
want
to
give
them
permission
to
just
change
it
to
anything.
So
what
did
we
do?
D
We
created
the
self-service
portal
that
has
certain
rule
sets
and
boundaries,
but
the
people
can
still
change
the
things
by
themselves,
but
we
know
what
they
did
when
they
did
it,
and
we
can
add
a
boundary
street
same
here.
I
will
show
you
a
quick
example
if
I
can
get
to
the
screen
yeah
you
did
so.
This
is
the
tell
stories
portal.
It's
in
German
story
about
that
you
can
create
projects,
for
example,
that
we
have
additional
fields
for
the
building,
for
example.
D
So
we
have
it's
mandatory
that
you
add
the
billing
number,
otherwise
they
would
just
create
a
new
project
and
who
did
it?
Why?
Where
do
you
have
to
do
it?
So
I
will
now
to
say
submit
17,
for
example,
click
on
create
and
instantly
on
the
open
v.
Gui
I
will
see
this
project,
so
this
is
use
pretty
pretty
often
another
use
case
is
this
quota
thing,
so
that's
default
2,
CPUs,
4,
gig
memory.
D
That
will
be
reached
very
fast,
so
we
just
say:
I
want
to
change
my
quota,
which
is
type
in
the
project
name,
say
maybe
3
and
10.
Do
it
and
pretty
much
the
same
here
immediately
you
get
the
new
value.
So
that's
just
a
few
use
cases.
You're
still
increasing
this
thing
and
putting
features
in
we
have
stories
a
concretions,
for
example.
In
there
we
will
add
monitoring
and
setup,
and
things
like
this.
So
my
last
point
is
the
monitoring.
What
do
we
do
here?
D
We
use
a
new
relic
as
a
sub
solution
to
monitor
our
infrastructure
or
service,
for
example,
just
what
they
do.
Then
we
added
a
custom
layer
that
does
more
or
less
the
same
text.
I
showed
you
before
just
on
all
our
clusters
and
we
use
the
big
drops
tool
to
do
pay
to
duty.
That's
just
a
little
web
tool
that
you
can
add
calendars
and
themes,
and
it
calls
you
if
something
goes
wrong.
D
E
Okay,
so
we
have
mainly,
after
the
same
setup
of
all
our
guides
here
first
step,
we
have
a
jenkins
or
a
git
repository
where
our
developers
commit
their
code
first
step.
Is
we
build
and
test
on
jenkins
and
produce
a
jar
or
whatever
and
place
it
into
nexus
and
from
then
on?
We
have
special
generic
chops
and
that
are
responsible
for
creating
or
they're
creating
the
docker
image
on
openshift.
E
So
what
we
basically
do
is
we
configure
Jenkins
shop
to
trigger
the
open,
shipped
API
and
create
image
stream,
build
config
and
then
instantly
triggered
to
build
itself
on
operative
to
produce
the
image
stream
or
docker
image,
and
for
that
we
provide
some
some
generic
jobs.
For
example,
we
have
one
for
simple
spring
good
application.
If
you
have
a
spring
boot
application,
you
can
add
this
generic
job
as
post,
build
to
your
existing,
continuous
or
nightly
job,
and
you
automatically
will
create
a
image
stream
on
openshift
that
you
can
use
to
deploy.
E
After
you
talk
about,
you
can
use
the
image
change
trigger
to
deploy
and
there
we
provide
different
resets,
for
example,
the
normal
automatic
deploy.
If
the
image
change,
the
image
will
change,
you
do
automatic
redeploy,
but
our
other
projects
also
have
or
use
more
manual
deployment
strategies,
for
example,
use
manual
steps
and
Jenkins
to
deploy
or
use
reassignment
of
tags
of
the
of
the
image
streams
to
deploy
to
production,
for
example.
E
E
So
when
project
wants
to
use
a
hazel
Cass
cluster,
we
have
a
open
check
template
where
you
can
specify
how
many
node
you
want
to
have
some
some
other
configurations,
click
on
create
and
you
instantly
have
a
hazel
cost
close
to
with
five
nodes.
For
example,
as
I
said
before,
all
these
technologies
are
also
monitored.
So
you
have
completely
monitoring
of
your
cluster
technologies.
E
What
you
also
provide
with
these
images
are
functionality
for
backups,
so
we
have
automatic
backups
for
maybe
Postgres
or
cassandra,
and
these
are
all
functionalities
that
most
of
the
time
you
have
to
build
it
on
a
on
a
existing
docker
hub
image.
So
we
take,
we
take
an
existing
image
from
from
docker
hub
and
modify
it,
and
then
our
projects,
our
internal
projects,
can
use
that.
E
Here
example:
I
talked
talked
of
it,
the
hazel
cost
template,
so
you
can
choose.
If
you
want
to
have
it
in
memory
or
persistent,
you
can
select
names,
name
space
of
projects.
You
can
choose
different
version
of
Facebook.
Ask,
for
example,
if
you
want
to
use
3.7
or
3.6,
you
can
add
the
EDUCAUSE
Management
Center,
as
well
as
some
other
stuff
and
click
on
create
and
your
hazel
cloth
clusters
is
ready
in
in
about
30
seconds.
E
So
I
will
give
you
a
sneak
preview
into
New
Relic
and
your
allocates
our
solution
for
application
monitoring,
as
well
as
monitoring
of
server
infrastructure.
What
we
see
here
is
a
APM
application
monitoring
for
the
next.
That's
the
the
software,
the
runs
in
the
background
to
serve
all
request
coming
from
the
mobile
application
and
on
the
clients
when,
when
our
passengers
want
to
want
to
know
where
to
go
next,.
E
D
D
Lots
of
them
that's
why
my
sentence
was
dead,
you
mean
in
if
we
did
it,
then
in
production
right.
No,
we
were
able
to
get
to
the
issues
in
the
development
stage,
then
fix
them
together
with
Red
Hat,
sometimes
modify
the
playbook,
so
they
don't
occur
and
we
were
able
to
do
production
upgrades
without
issues
related
to
the
playbooks
I
would
say
which
I.
A
D
There,
in
the
base
image,
we
have
the
whole
chain
built
in,
so
we
just
have
imported
the
CH
and
intermediate
certificate
inside
the
java.security
path,
for
example,
and
so
then
they
trust
the
prob.
The
certificates
are
on
the
mosque
and
H
a
proxy.
For
example.
That's
what
we
mean
by
building
yeah,
welcome
cool.