►
From YouTube: OpenShift Commons Briefing #144: Continuous Deployment on OpenShift with Spinnaker (OpsMx)
Description
In this briefing, learn if Spinnaker is the right choice for your Continuous Delivery solution for OpenShift deployments. Your developers have adopted containers and microservices applications. You have implemented continuous integration tools like Jenkins. But your software delivery is still slow due to high ceremony release approvals, manual judgment of releases and myriad of scripts to roll-out and roll-back cloud deployments into OpenShift and other clouds. So in essence, you need to modernize your software delivery to meet the increasing demand for speed of innovation from your developers.
What you will learn:
- OpenShift users top requirements for next-generation Continuous Delivery solutions
- Introduction to Spinnaker enterprise CD solution
- Top features and benefits of Spinnaker
- Spinnaker demo - Deploying to OpenShift
Speakers:
Balaji Sivasubramanian (balaji@opsmx.com)
Gopinath Rebala (gopinath@opsmx.com)
A
Well,
hello,
everybody
and
welcome
again
to
another
openshift
commons
briefing,
this
one
from
an
open
ship
commons
member
up
next,
who
are
amazing
folks
in
the
spinnaker
community
and
they're,
going
to
talk
about
continuous
deployment
on
open
ship
with
spinnaker,
which
I've
been
looking
forward
to
for
quite
some
time.
I'm
gonna!
Let
the
two
of
them
and
open
up
involved
a
introduce
themselves
and
we'll
have
you
can
ask
questions
in
the
chat
and
we'll
have
live
Q&A
at
the
end.
So
take
it
away.
Thank.
B
B
So
we
will
go
through
the
current.
You
know
a
little
bit
of
a
background
of
what
kind
of
delivery
and
deployments,
because
that's
something
we
want
to
get
the
same
language
across
the
board.
So
everybody
is
on
sync
and
then
we'll
talk
about
the
current
version
yesterday
and
how
supposed
to
be
there's
some
drawbacks
or
some
of
the
challenges
there
and
how
spinnaker
could
be
something
you
could
use
to
to
not
only
do
deployments,
but
to
do
full
can
use
celery
with
open
shipped
across
other
hybrid
cloud
or
public
cloud
settings.
B
You
have
environments,
you
have
a
v8
of
device,
you
have
you
know,
native
kubernetes
or
or
DCP,
etc.
So
kudos
maker
is
a
tool
that
allows
you
to
deploy
across
multiple
flowers,
and
this
is
something
a
lot
of
large
customers
have
adopted.
We
will
introduce
the
spinnaker
itself
and
go
through
the
details
and
under
the
various
features,
and
why
I'll
give
a
like
spinner
Lulu
demo
will
have
a
lot
lakh
couple
of
slides
on
what
is
Enterprise.
B
B
So
to
two
level
said:
I
think
the
enterprise's
are,
you
know
they
have
done
the
CI
part
in
all
the
continues
integration
part
fairly.
Well,
they
are
able
to
now
automate
from
cord
to
do
purchase
integration
and
through
Jenkins
and
build
a
particular
build
an
artifact
and
be
able
to
then
deploy
into
test,
deploy
into
production,
staging
and
production,
but
most
of
the
part
of
deploying
into
staging
or
deploying
into
production
is
done
through
manual
scripts,
and
that's
one
part
of
the
problem.
B
Other
part
of
the
problem
is
that
there's
oliveira
fication
that
needs
to
happen,
maybe
for
compliance,
maybe
for
just
a
quality
of
the
release.
Authorities
actually
works.
A
lot
of
these
things
are
done
in
a
manual
fashion
or
some
sort
of
a
scripting,
but
it
is
very
custom.
Scripting
now
is
the
application
change
over
time.
So
it's
difficult
to
keep
track
of
all
of
that.
B
If
you
look
at
very
large
customers
that
we
work
with,
they
are
increasingly
needing
to
go
to
obviously
a
high
velocity
release
process
right
compared
to
the
past
years
of
deploying
every
couple
of
months
or
month
or
even
week,
for
example,
they
want
to
do
more
quicker
than
that,
then
obviously,
that
need
came
about
because
of
micro
services.
The
way
there
are
too
many
now
that
the
applications
are
kind
of
split
into
a
lot
of
micro
services.
So
you
are
actually
doing
updates
much
quicker
and
you
want
to
be
able
to
update
them.
B
You
know
within
our
basically
so
there's
a
need
to
do
faster.
Now,
absolutely
do
you
want
to
do
faster?
You
want
to
automate
it
because
there's
no
way
to
manually
test
and
verify
all
of
these
things,
and
that's
where
the
vinegar
comes
in
as
well.
Some
of
the
things
that
we
are
we
opted
some
X
is
doing
is
also
comes
in
in
terms
of
verification
of
this
releases,
and
this
is
sort
of
the
problem
that
I
think
that
the
previous
slide
was
talking
about
where
most
of
enterprises
actually
are
complaining
about
that
right.
B
B
The
other
hands
but
I
talked
about
earlier
is
the
multi
cloud.
You
know
you
have
open
shift,
obviously
in
this
audience,
but
you
obviously
have
I'm
pretty
sure
you
have
you
know
either
AWS
or
OpenStack.
You
know
their
other
review,
environment
or
vSphere.
How
do
you
deploy
across
all
of
these
things?
You
don't
want
to
have.
You
know
five
different
tools
to
deploying
to
five
different
town
and
ramens,
and
then
you
want
to
be
able
to
deploy
safely.
We
talked
about
how
fast
you
want
to
do
it,
but
then
fast
is
gonna.
B
B
Now,
if
you
look
at
spinnaker
dangerous
feature,
kantha,
which
is
basically
automated
canada
analysis,
and
we
will
spend
a
little
more
time
on
that
in
the
later
part
of
the
things
and
the
other
part
that
I
think
you
know,
people
who
are
in
this
audience
probably
appreciate
is
the
open
source
mandate.
That's
that's!
People
are
enterprises
are
requiring
for
any
tools,
anything
in
the
IT
industry
right.
B
You
know
I
want
to
be
able
to
have
an
open
source
tool
so
that
I
can
I,
don't
have
to
be
vendor
locked
in
and
that's
something
spinnaker
provides
very
well
so
with
that
I
think
I'll
pass
on
or
Gopi
who's
going
to
explain
to
you.
Obviously
the
open
chef
deployments
strategies
and
what
spinnaker
does
for
you
and
then
we'll
I'll
come
back
in
the
later
part
of
the
session.
B
C
Zoology
gave
a
brief
motivation
for
the
continuous
delivery
and
the
deployments.
Essentially,
you
want
to
have
the
incremental
deliveries
with
feedback
for
all
the
way
from
operations
developers
to
provide
a
continuous
improvement
in
the
processes
and
reduce
the
friction
between
delivering
from
one
stage
to
another,
so
that
that's
kind
of
our
goal
reduce
the
risk
in
delivery
at
high
speed
right,
open
field
is
a
great
platform.
It
is
cloud
native,
a
supports
in
the
communities
natively
it's
multi
tenant.
C
You
can
drive
this
on
multiple
infrastructures,
AWS
GCP
on
prime,
that's
one
of
the
things
that
enterprises
love
about
openshift
and
it
provides
entire
feature
set
of
communities
with
added
benefits
of
few
things
that
are
built
in
on
top
of
it,
including
the
deployments
and
routes
and
things
like
that,
and
they
are
part
of
the
C&C.
Have
any
advances
that
are
going
on
like
there
is
mesh,
integrations
they're
all
being
built
in
so
it's
a
great
platform.
That's
are
very
good
for
micro
services
and
incremental
delivery.
C
So
to
but
people
just
to
refresh
what
what
is
provided
by
open
shift
and
in
terms
of
added
benefits
on
top
of
kubernetes.
Here
you
can
see
that
all
the
blue
ones
are
communities
there
come
as
part
of
in
a
and
the
rest
of
them
are
the
features
that
are
provided
by
open
shipped
environment
and
it
comes
with
a
default.
Sdn
I
think
it's
a
similar
thing
that
comes
from
OpenStack
routes,
which
are
provide.
C
Repository,
that's
essentially,
OpenShift
also
is
providing
additional
features
for
continuous
deployments
right
in
the
delivery,
so
the
deployment
strategies
which,
since
we
are
mostly
focused
on
deployments
today
in
delivery
strategies
generally
of
this
rolling
update,
which
essentially
replaces
one
instance
at
a
time
in
the
target
department
and
once
it's
ready,
then
you
go
to
the
next
one.
This
is
sort
of
what
is
provided
by
kubernetes.
Also,
there
are
readiness
probes
with
additional
pre
and
post
checks
that
can
be
done,
and
there
is
a
blue,
green,
blue
green.
C
Is
the
strategy
in
which
you
bring
up
a
new
cluster,
newer
version
of
the
cluster
and
start
using
it?
If
there
is
something
wrong,
you
have
the
ability
to
switch
back
to
the
previous
version
of
the
cluster,
to
reduce
the
risk
of
the
amount
of
downtime
that
you
would
have
with
the
failure
set
with
the
new
version.
Every
deployment
is
more
for
the
error,
API
compatible
deployments,
where
you
have
both
v1
and
v2
version,
1
and
version
2
running
simultaneously
with
traffic
shaping
this
thing
on.
C
So,
to
give
a
quick
example
on
how
the
blue/green
can
be
achieved
through
open,
shipped
environment,
let's
say
we
have
a
version,
one
of
a
service
that
is
running
in
the
target
environment.
You
bring
up
a
version
two
with
a
new
app
example
version
two
and
give
it
a
new
name,
essentially
the
it's
a
label
that
we
are
assigning
into
the
newer
version
that
is
coming
in
and
for
a
service
as
a
selector.
C
Just
to
refresh
for
people
who
are
not
really
clear
on
the
terms.
The
service
is
more
like
a
load
balancer
or
a
firewall,
that's
put
in
front
of
the
pods,
which
can
be
one
or
more
containers
that
expose
certain
features
from
that
part
in
these
services
in
communities
they
work
by
using
selectors.
Each
of
these
pods
can
have
the
labels
that
are
assigned
to
them
and
in
a
service
that
you
specify
that
these
are
the
labels
that
I
want
you
spend
traffic
to,
and
that
is
how
it
works.
C
So
here
what
it's
doing
is
for
the
service.
It's
updating
the
name
of
the
selector
for
which
it
needs
to
send
traffic.
To
to
the
example,
blue
Tilly,
the
previous
version
did
not
have
the
label
example.
You
know
all
the
traffic
for
this
service
that
is
in
coming
into
the
cluster
is
will
be
sent
to
the
example
of
you,
but
the
previous
instance
is
still
running
now.
You
want
to
go
and
clean
up.
That
instance
keep
it
around,
reduce
the
number
of
instances
of
parts
to
zero,
but
have
the
replica
set
on.
C
So
if
something
goes
wrong
with
the
newer
version,
you
can
go
back
and
increase
the
number
of
pores
in
the
previous
version
until
it's
and
you
would
want
to
achieve
this,
Bluegreen
reduced
it
roll
back,
but
this
looks
simple
enough,
but
in
reality
this
requires
a
lot
more
scripting
to
actually
get
it
automated
and
working
in
environment
production,
violence,
for
example,
if
you'd
apply
the
version
two.
This
is
a
declarative
approach
that
you
tell
kubernetes
or
open
share
is
bring
up.
Three
three
parts
put
my
version
two
and
that's
it.
C
The
OSI
comes
back
and
say
success.
Now
you
have
to
wait
until
the
readiness
are
all
of
these
is
confirmed
before
you
switch
over
to
the
newer
versions
this
electrostatic.
So
all
of
that
needs
to
go
in
and
you
have
to
make
sure
all
the
scripting
waiting
is
done
correctly
and
if
there
is
some
failures,
you
don't
want
to
go
forward
with
the
deployment
and
all
of
that
scripting
now
becomes
part
of
the
deployment.
C
So
essentially,
although
there
are
significant
deployment,
things
that
are
provided
by
open
shipped
environment
because
of
the
micro
services,
some
are
much
easier
than
what
it
used
to
be
native
platforms
or
bit
metal
platforms.
You
get
away
from
the
DLL
health
and
incremental
approach
of
having
a
container
that's
immutable,
that's
deployed
that
allows
us
to
rollback
easily,
but
there's
still
the
strategies
require
additional
scripting
that
needs
to
happen
either
you
want
to
do
blue,
green
or
a
B
is
additional
scripting.
That
needs
to
happen
for
those
and
even
with
the
pipeline
integrations.
C
C
C
You
need
to
have
a
consistent
delivery
mechanism,
consistent
view
of
your
application,
which
parts
of
the
system
are
working
in
which
world
or
have
a
high
availability,
the
fallback
mechanisms,
all
of
that
needs
to
come
in
into
single
view,
for
application
developers
to
the
operations
of
those
applications,
and
that
consistent
view
is
critical
for
enterprises
to
manage
the
applications
with
large
number
of
services.
With
the
micro
services
now
they're
exploring
the
each
application
has
20
30
micro
services.
This
management
becomes
much
harder
policy
controls.
C
Is
that
another
one,
so
it
enterprises
how
these
policy
is
saying.
Some
of
these
components
need
to
go
through
thick
Sox
compliance
or
some
study
rules.
Only
at
certain
level
they
can
go
to
Samantha
helpers
in
the
testing.
They
want
to
get
around
these
controls
and
test
indoor
environments
without
having,
through
certain
number
of
checks
that
needs
to
be
done
for
functionality
and
but
use
the
same
kind
of
artifacts
once
they
are
tested
passed
through
the
pipelines
to
go
to
staging
or
test
environments.
C
C
So
we
let's
say
we
are
deploying
version
two,
you
you
what
the
strategy
is
to
deploy
by
five
percent
of
the
traffic
or
ten
percent
of
the
traffic,
the
newer
version
and
make
sure
the
performance
is
consistent
with
the
existing
versions
are
better
and
only
then
you
go
ahead,
and
you
do
this
in
an
automated
way,
and
this
is
commonly
done
in
manual
in
the
sense
when
you
deploy
a
new
version.
The
operations
team
would
go
and
check
if
the
latencies
are
good.
Memory.
Utilization
is
within
the
range
and
then
they
move
forward.
C
So
this
spinnaker
provides
this
automated
Kennedy
that
you
can
use
it.
Out-Of-The-Box
DevOps
processes
shift
left
is
one
of
the
critical
things
really
to
improve
the
process
continuously
and
with
all
these
declarative
approach,
amell's
and
you
have
the
too
many
animals
so
each
of
the
service
you
have
one
ammo
you're
managing
for
this
applications
and
different
environment
that
you
are
deploying
for
the
development
or
in
production.
You
construct
a
different
animal
because
you
want
to
go
to
different
checks.
C
All
of
that
is
exploding,
and
even
with
the
templates
Helen
charts,
there
is
still
a
lot
of
scripting
that
goes
around.
All
of
that
is
better
managed
if
you
have
a
system
to
have
pipelines
end-to-end
than
that
support
it,
and
also
it
cannot
be
completely
declarative
because
of
the
policy
checks
that
are
involved.
So
you
need
to
be
able
to
combine
this
imperative
and
declarative
sections
to
provide
it.
So,
essentially,
you
want
to
have
a
simpler
process
to
deploy
code
to
production.
C
At
the
same
time,
you
want
to
be
able
to
manage
these
pipelines
in
an
enterprise
with
the
flexibility
of
self-service
of
the
developers
without
having
to
do
lot
of
training.
So
that's
essentially
what
pinnacle
brings
to
the
table.
It
is
built
ground-up
with
the
DevOps
principles
in
mind,
and
it
supports
the
scale.
C
Netflix
has
been
using
it
for
a
few
years
now
it
supports
the
multi-cloud
deployments
and
it
it
has
built-in
automated
safe
releases
ability
to
rollback
this
canary
blue
green,
rolling
updates,
they're
all
built
in
and
they're
consistent
across
multiple
cloud
environments.
If
you're
deploying
to
open
shipped
environment
the
terminology
and
how
it
construct
the
pipelines
is
similar
to
how
you
would
deploy
for
an
AWS.
C
C
Essentially,
the
community,
for
example,
target,
has
implemented
OpenStack
driver
and
look
how
it
is
implemented
easy,
as
for
Amazon,
it's
a
very
active
community
in
vibrant,
so
just
to
touch
on
core
principles
of
spinnaker
with
the
micro
services,
and
the
primary
thing
that
we
are
trying
to
do
is
increase
the
velocity
of
delivering
the
software
to
end-users.
You
do
that
by
incremental
updates
and
at
the
same
time
you
want
to
use
the
immutable
infrastructure.
C
Essentially,
the
component
that
is
built
that
is
tested
is
the
same
component
that
goes
to
the
production,
so
that
gives
us
a
consistency
in
deploying
this
includes
configuration,
essentially
the
configuration
approach
that
you
would
take.
You
don't
want
to
rebuild
the
component
that
you
are
deploying
in
production
that
is
tested
differently
and
because
of
this
incremental
changes
in
immutable
infrastructure,
you
can
roll
back
to
the
previous
version.
C
So
when
you
do
roll
back,
you
want
to
be
rolling
back,
not
just
the
bits
binaries,
but
also
the
configuration
that
goes
with
that,
and
so
you
build
the
deployment
strategies
into
it
and
the
primary
goal
purpose
goal
is
also
to
automate
all
of
this.
So
there
are
no
errors
that
are
introduced
on
a
day
to
day
basis,
so
this
repeatability
and
tracking
allows
for
lower
risk
of
deployments
and
last
but
not
least,
is
operational
integration.
C
So
the
operations
team
and
the
development
team
should
have
visibility
into
what
is
going
in
to
the
operations
when
it's
going.
However,
how
is
what's
changed
and
what
are
the
checks
that
are
done
before
it
go?
Isn't
it
a
brief
view,
worried
of
spinnaker
architecture?
The
spinnaker
itself
is
built
with
the
micro-services
architecture
and
each
of
these
components
scales
horizontally,
and
you
could
use
only
subset
of
the
components
if
need
be,
for
certain
applications.
C
And
it
is
supported
with
zero
downtime
releases,
so
here
at
the
components,
some
of
these
components
are
horizontally
scaled
and
you
can
size
them
vertically
also
for
efficiency,
depending
on
the
deployments
they're
both
distributed
in
local
instance.
They
work
with
openshift
environments
aw
as
when
two
environments
or
distributed
in
Ubuntu,
and
they
support
zero
downtime
upgrades
with
tool
called
halyard
in
multiple
environments.
C
It
it
has
a
first-class
citizen
support
for
kubernetes,
it's
their
two
methods.
It
supports
the
the
manifest
files
are
natively
supported
it,
treats
them
as
templates
and
with
the
triggers
it
has.
This
ability
to
understand
the
changed,
artifacts
and
construct
is
manifest
files
with
the
change
artifacts
and
apply
them
and
make
sure
the
levels
are
reached
within
the
community
openshift
environments
before
it
moves
to
the
next
stage.
C
It
supports
all
the
native
kubernetes
or
OpenShift
deployment
strategies
for
the
developers
who
are
starting
off
with
containers
and
not
familiar
with
openshift
that
they
could
also
use
a
simpler
deployment
pipeline
stages
where
they
can
just
specify
what
container
ID
they
want
to
use,
and
it
said
the
service
porch
that
need
name
to
open
and
use
that
to
just
deploy,
help
chumps
tower
support
is
already
there
in
spinnaker
and
for
people
who
don't
know.
This.
Here
is
a
service
mesh
initiative
with
the
cloud
native
foundation
and
that's
a
network
layer
service
distribution
service
mesh.
C
Basically,
it
supports
both
l,
4
and
l
7,
a
load
balancing
for
the
services.
It
has
intelligent
controls
on,
let's
say
incoming
traffic.
We
only
want
Chrome
browsers
to
go
to
the
newer
version
and
the
rest
of
it
all
goes
to
the
older
work,
and
so
you
can
set
selectors
in
similar
ways.
So
this
will
be
integrated
in
the
deployment
strategies,
the
one
that
we
were
talking
about
in
terms
of
Kennedy.
C
C
It
is
also
very
flexible
stage:
executions,
the
you
can
conditionally
test
a
particular
say,
run
a
particular
test
or
not.
There
are
some
things
called
pipeline
expressions
based
on
the
execution
context.
You
have
these
variables
that
are
available
to
you.
You
can
use
those
variables
set
conditions
and
allow
a
particular
stage
to
run
or
not.
C
There
are
deployment.
Windows
is
essentially
if
you're
going
to
production-
and
you
can
say
I
only
want
to
deploy
between
nine
and
11
a.m.
on
Monday
morning.
You
will
do
that.
The
reason
people
tend
to
do
deployments
at
their
high
traffic
time
is
it's
easier
to
test
the
canary
at
the
time
and
make
sure
nothing
goes
wrong
in
those
periods
for
5%
of
the
traffic
and
control
back
quickly,
instead
of
deploying
on
Sunday
night
when
there's
little
traffic
and
get
into
trouble
on
my
live
on,
so
it
also
supports
multiple
cloud
environments.
As
we
discussed.
B
Quick
comment:
I
want
to
make
on
the
previous
slide,
which
is
support.
The
deployment
I,
guess
the
video
deployment
environments
and
we
actually
enabled
we
are
the
first
render
or
first
people
in
the
community
actually
to
enable
openshift
spinnaker.
If
you
go
to
the
spinnaker
and
that
I
owe
you
what
you
would
not
see,
you
punch
it
in
there
they're
trying
to
get
it
added
there,
but
essentially
we
have
enabled
you
will
use
it
component
is
AP.
C
Deployment
strategies
and
then
type
as
a
code
in
spinnaker
these
the
api
is
each
one
of
these
stages.
It's
the
pipeline
itself
is
codified,
so
it
is
possible
for
us
to
create
a
pipeline
structure
outside
and
apply
it.
So
in
an
enterprise,
if
there
are
breaks
best
practices
with
its
templates,
they
can
be
applied
from
outside,
as
well
as
you
create
visually
a
pipeline
and
take
that
and
construct
a
template
or
rocket
out
of
it.
C
There
are
multiple
integrations
and
they're,
always
growing,
that
triggers
from
CI
systems,
docker
registry
and
it's
also
possible
to
trigger
one
pipeline
from
another.
This
is
useful
when
you
have
the
dependent
services
which
are
not
API
compatible.
That
say,
if
you
have
a
certain
version.
V
3.2
only
works
with
the
1.0
and
above
another
service.
You
can
have
a
pipeline
dependencies
that
goes
and
checks
those
dependencies
and
or
deploys
them.
C
C
This
is
one
of
those
popular
ones
with
Netflix
they
have
the
simian
army,
which
allows
for
testing
of
resiliency
of
in
the
production
environment.
The
spinnaker
is
tightly
integrated
with
chaos.
Monkey
the
cash
monkey
is
a
tool
which
would
randomly
go,
terminate
certain
instances
or
clusters
in
the
target
environment,
and
you
can
then
attach
a
automated
can
any
kind
of
analysis
with
it.
Whenever
this
is
instance,
it
goes
down.
You
do
the
analysis
and
verify
your
system
is
up
and
as
expected,
and
you
can
also
find
the
differences
in
performance
when
those
things
happen.
C
You
can
also
have
a
flexible
configuration
here
to
say
in
one
day
do
not
delete
more
than
one
instance
in
this
particular
cluster
or
exclude
certain
clusters.
All
of
this
is
configurable
through
the
spinnaker
UI,
that's
integrated
with
cache
monkey,
and
this
is
a
configuration
of
a
pipeline
that
you
can
see
in
spinnaker
for
resiliency
or
high
availability.
An
application
is
deployed,
let's
in
multiple
regions.
C
Typically,
you
would
deploy
to
region
one
after
everything
is
up
and
running
with
the
newer
version
you
go,
deploy
to
region
two
and
cascade
to
multiple
locations
for
that
application.
This
is
the
spinnaker
very
easily
supports
those
kind
of
deployments
visually
and
you
can
see
how
the
progression
of
deployment
is
going
on
across
different
regions
to.
C
There
is
a
trigger
configuration
that
is
coming
from
bitbucket,
which
builds
a
project
and
then
deploys
to
staging
here.
One
of
the
interesting
things
here
is,
if
the
staging
itself,
this
is
a
pipeline
stage.
Essentially
it
is
saying
after
this
is
built,
I
want
to
deploy
it
to
the
staging
using
a
different
pipeline,
so
this
actually
goes
and
instantiate
this
this
pipeline.
After
that
there
is
a
manual
judgment
and
then
it
deploys
the
production.
C
C
Doing
a
recreate
strategy,
because
it's
a
testing
environment
and
applying
the
artifact
see
here,
you
can
see
that
the
version
is
not
specified
after
the
build
is
done
when
the
artifact
ID
is
specified
in
the
ruthenians
properties
file.
This
pinnacle
has
this
ability
to
pick
up
the
new
artifacts
that
are
there
created
and
apply
them
to
the
pipelines
and
deploy
to
OpenShift
and
mana.
C
C
C
C
C
So
these
are
the
ones
that
are
deployed
through
manifest
and
in
the
production
environment.
This
is
essentially
what
we
saw
in
the
one
here.
You
can
see
the
number
of
parts
depending
what
the
cluster
is
in
its
operations
that
we
can
do.
The
services
also
are
displayed
here
in
this
cluster.
This
is
the
cluster
that's
attached
to
this
load,
balancer.
C
C
C
C
We
will
see
that
there
is
a
new
replica
set
that
gets
created
and
once
this
is
done
will
see
that
the
transom.
So
this
is
again-
and
we
said
this
stage
itself-
is
a
pipeline
stage,
so
we
can
see
that
this
stage
is
running.
It's
actually
triggered
this
pipeline
and
when
the
pipeline
is
complete,
you
can
have
this
manual
page,
which
I
continue
with,
because
we
like
the
deployment
and
it
continues.
C
C
C
We
can
specify
what
is
the
baseline
version
that
we
want
to
check
with
and
every
time
the
new
version
is
defied.
It
verifies
the
cannery
of
that
system,
and
this
is
the
report
that
you
would
get
the
deccan
of
the
analysis.
You
can
specify
thresholds
for
the
score
where
it
allows
for
moving
the
production
or
not
good,
close
running
out
of
time.
So
Balaji
you
wanna
continue.
B
Yeah
and
so
I
think
there
was
also
a
question
actually
there
that
I
was
answering,
which
is
obviously
spinnaker,
is
open
source,
and
so
there's
need,
or
enterprises
to
to
have
some
sort
of
and
support
that
helps
them
to
deploy
to
production
in
case
there's
issues
that
pop
up
or
or
generally
you
want
to
get
a
bug
and
you
want
to
drive
to
the
community,
etc,
etc.
So,
there's
also
aspect
to
how
to
onboard
applications
are
more
users,
so
obscene
mix.
B
This
is
sort
of
the
advertisement
for
oh,
yes,
this
presentation,
you
know
we
actually
provide
Enterprise
spinnaker
for
users,
so
we
obviously
support
open
source
panicker.
So
we
have
a
community
version
where
we
support
people
who
want
to
use
open
source,
but
most
enterprises
natori
by
7
support
or
enterprise
support,
and
so
they
want
to
also.
We
also
provide
additional
features
on
top
of
Springer.
It's
fully.
The
VW
no
sticker
is
open-source
and
we
provide
this
additional
things
on
top
of
it
with
like
uneasy
onboarding.
B
You
know,
like
the
automated
verification
that
Capri
talked
about
their
being
able
to
use
alongs,
APM,
metrics
and
other
kinds
of
metric
that
you
have
like
eight
at
all,
you
really
Splunk,
exeter,
etc,
and
to
be
able
to
do
a
smart
deploy.
So
you
need
to
ploy
it
all
out,
and
you
know
today
is
just
a
basic
health
check
that
you
do.
On
top
of,
when
you
million
that,
when
communities
also
an
application,
you
don't
really
test
the
actual
application
itself.
B
Whether
it's
actually
is
good
or
not,
you
know
it
doesn't
have
any
regressions
and
does
it
still
meet.
The
SLA
is
as
except
track
cetera,
so
you
you
want
to
be
able
to
people.
Do
it
today
manually
they
to
check
the
few
api's
and
see
if
it's
okay,
what
we've
done
is
we've
create
an
automated
way
to
look
at
it
and
we
also
use
machine
learning
so
that
the
what
it
checks
is
automatically
updated
all
the
time.
So
this
is
some
of
the
value
at
v2
as
part
of
the
enterprise.
B
B
Spinnaker,
as
you
talked
about
yes
under,
was
an
API
is
for
all
the
events
that
is
such
a
string
in
terms
of
who
deployed
what
they
deployed
in
when
it
was
deployed,
who
will
provide
at
Exeter
Exeter,
but
there's
no
sort
of
an
easy
way
to
sort
of
a
productionize
that
right
in
terms
of
you
know,
if
you're
have
an
auditor
who
comes
in
and
says,
show
me
this
you,
you
need
to
write
your
own
script,
for
example,
to
find
that
information.
So
some
of
the
things
we
do
and
compliance
is
another
thing.
B
People
have
rules
around
compliance,
you
know
you,
you
shall
not
attempt,
shall
not
deployed
production
or-
or
this
needs
to
be
in
place
to
deploy
this
application
into
production
in
Romans,
the
Congo
or
AWS.
They
can
go
to
Europe
zone,
whatever
X,
etc,
and
as
a
developer,
you
may
make
a
mistake
or
somebody
who's
sitting
at
the
pipeline.
You
may
make
a
mistake,
but
these
things
will
will
essentially
audit
that
and
verify
that
those
things
are
there.
So
we
have
integration
into
various
data
sources.
B
Obviously,
the
logs
metrics
in
a
VM
data
sources,
but
also
data
sources,
can
go
back
to
web
sorry
data
sources
around
you
know,
security
verifications,
ask
the
in
task
tools,
etc
as
well.
I
don't
do
security
in
verifications,
the
other
part
or
on
the
plan
in
Romans
we
talked
about
already
that
panic
obviously
provides
a
set
of
deployment
options,
like
called
the
cloud
drivers.
You
know
the
obviously
the
the
things
around
80
of
the
best
communities,
etc,
etc,
and
in
me,
what
we
do
is
we
enable
to
open
ships?
Basically,
a
customer
requirement.
B
You
know
asking
us
to
say:
open
ship
is
not
there.
How
can
we
enable
spinnaker
deployed
openshift,
and
you
know,
I-
have
a
multi-tenancy
I
have
these
these
requirements,
Enterprise
requirements
that
we
have
on
a
rocket
ship
cluster
I
want
to
make
sure
that
spinnaker
actually
you'd
be
able
to.
You
know
work
within
that
in
environments
and
that's
something
we
did
that's
a
lot
of
effort
that
we
put
in
to
make
that
work
and
we're
happy
to
share
with
the
community
and
to
other
customers,
and
there
will
also
be
other
environments
like
OpenStack,
for
example.
B
Even
though
it's
supported
that
target
actually
created
that
features
actually
contribute
its
accord,
there's
still
some
gaps
in
the
way
it
is
implemented
for
somebody
who's
not
using
exactly
the
same
way
targeting,
for
example,
or
you
have
a
newer
version.
It's
something
changed,
for
example,
those
things
where
we
come
in
and
really
do
a
good
job
of
helping
customers.
B
You
know
I,
think
AWS
is
fully
baked
again.
I
would
say
the
spinnaker
the
community
is
pretty
close
to
being
fully
baked,
but
there
are
other
provided
in
Romans
where
there's
a
lot
of
gaps,
and
we
definitely
put
into
that.
We
have
customers
asking
is
for
V
spear
and
V
spear
I
want
to
party
be
square,
so
these
are
some
of
things
we
provide
as
part
of
enterprise
spinnaker
and-
and
we
have
customers,
for
example,
want
to
use
vinegar
to
deploy
to
multiple
multiple
instances
of
spinnaker
itself.
B
Like
you
know,
I
want
to
be
able
to
apply
to
multiple
groups
or
I
want
to
have
multiple
spinnaker
deployments,
and
so
these
are
things
like
things.
Okay,
I
think
we
talked
about
this
a
little
bit
already.
I
ability
to
apply
a
verification.
Basically,
using
the
automated
notification
feature,
you
want
to
do
smart
deploys
right,
it
will
to
roll
back
the
invasion,
to
be
able
to
control
your
rollouts
and
using
data
and
using
machine
learning
and
data,
and
compliance
is
again
is
a
something.
An
enterprise
level
feature
looks
like
I
think
you
already
covered
it.
C
So
in
the
verification
we
you
can
specify
what
do
you
want
to
refine
in,
so
it
can
be
the
logs
or
APM
metrics.
So,
for
example,
in
this
particular
case,
the
score
is
no.
Usually
we
have
a
threshold
in
somewhere
like
around
8
85,
above
which
it's
a
successful
comparison
between
the
versions
and
below
is
a
failure.
Now,
in
this
particular
case,
it's
identified
going
through
the
lords,
we
use
NLP
with
RNN
processing
and
it
identified
the
differences,
and
there
are
some
error
topics
that
were
generated
in
the
newer
one.
C
That
did
not
exist
in
the
previous
version.
So
if
there
are
topics
that
have
consistent
warnings
or
errors
in
the
version
1
and
in
the
newer
version,
then
it
would
not
hi
like
that
here
they
were
significant
reconnects.
That
was
happening
in
the
second
version
that
they
picked
out
and
shown
that
so
these
are
the
API
metrics
based
on
the
SLA
S&S
loz
wii.
You
can
specify
what
you
want
to
check
and
it
automatically
goes
and
verifies
each
one
of
those
groups
and
identifies
which
one
odds
are
important
for
this
particular
deployment
or
long
tear.
C
So
it
learns
essentially
to
identify
if
you're
collecting
for
the
infrastructure
metrics,
as
well
as
the
service
level
metrics
it
identifies,
which
ones
are
the
ones
that
are
important,
which
ones
are
not
and
and
waste
them
automatically
for
you.
So
it
identifies
that
these
are
the
dimensions
in
which
the
system
has
differed
and
identifies,
if
that's
a
failure,
condition
or
not
so.
B
To
reiterate
right,
this
is
a
report
that
gets
auto-generated
as
part
of
the
pipeline.
So
when
I
think
opee's
showed
in
the
in
the
in
the
spinnaker
UI
as
part
of
the
stage
you
go
back
to
the
run
report
go
to
the
wrong
side.
I'm
sorry,
so
you
basically
in
this
stage,
is,
is
automatically
set
up
as
part
of
every
pipeline.
Everybody
been
one,
for
example.
B
If
the
score
is
below
a
certain
threshold
and
we
can
basically
program
it
to
stop
and
because
usually
you
don't
want
to
have
a
batteries
go
out
and
and
then
the
users
are
able
to
basically
figure
it
out
from
the
okay.
While
it
is
Early's
fail,
and
then
you
are
able
to
be
able
to
go
to
that
report
and
figure
out
what
what
went
on
okay,
that's
good
for
now.
Okay,
I
think
we
are
out
of
time.
So,
let's
just
go
back
into
there.
B
What
does
I
think
yeah
one
last
slide,
yeah,
so
there's
a
something
called
spinnaker,
summon
that
is
Netflix's
is
a
sponsor
and
they've
set
up.
They
did
a
one
last
year
and
this
year
it's
happening
in
Seattle,
October,
8th
through
9th
I,
think
for
the
40
plus
sessions,
or
you
know,
on
road
maps,
case
studies,
best
practices,
etc,
etc
and
I
think
the
expectation
is
what
400
to
500
attendees.
We
have
a
discount
coupon
this
Concord,
we
don't
know
the
sponsors
of
the
spinnaker
summit.
B
So
if
you
use
the
ops,
MX
20
you
get
about
20%
off,
so
please
feel
free
to
use
it
and
enjoy
the
spinnaker
summit
next
slide.
Please
I
think
we
are
done
their
thing.
You
know
if
you
have
more
questions
reach
out
to
us
at
info
at
ops
of
XCOM.
Obviously
you
have
the
spinnaker
slack
channel
that
you
can
use
to
ask
questions
so
the
community.
If
there's
any
questions,
we
can
happy
to
take
I.
Think
I'm.
Reading
talking
to
one
tourist
a
little
bit
you
do.
Email
me
is
all
go
ahead.