►
Description
For more Continuous Delivery Foundation content, check out our blog: https://cd.foundation/blog/
Panel Discussion: Spinnaker Success - CD as a Service using Spinnaker - Balaji Sivasubramanian, OpsMx; AbdulBasit Kabir & Asiya Yunusa, Interswitch Group & Arun Kumar, Cisco
Enterprises are looking to deliver releases faster and safer. Spinnaker is a leading tool for providing CD as a Service/product for scalable multi-cloud deployments among these Enterprises. Spinnaker allows the ability to enable self-service, control (through policy integration), and deploy safely with automated verification. In this session, we will talk to InterSwitch and TBD end-user that leverages Spinnaker to deliver CD as a product to learn about their use cases, best practices, and lessons learned.
A
Hello
welcome
to
Spinnaker
Summit
2022,
and
this
session
is
Spinnaker's
success
stories.
Today
we
have
users
from
Cisco
and
ninja
switch
we're
using
Spinnaker
at
scale
to
provide
CD
as
a
service
to
thousands
of
applications.
Today
they
will
be
sharing
their
use
cases,
best
practices
and
Lessons
Learned
My
name
is
Balaji
Siva
I
am
VP
of
product
at
opsa,
Max
I've
been
as
part
of
the
community
for
over
five
years,
working
with
large
companies
that
are
adopting
Spinnaker
for
various
use
cases.
A
If
you're
not
familiar
with
optimax
Optometry
is
a
provider
of
Enterprise
product
support
and
services.
Around
Spinnaker
Spinnaker
has
been
widely
used
by
many
companies
not
only
deployed
to
Cloud
targets
like
AWS,
Azure
and
Quebec
and
and
and
gcp,
but
also
to
various
kubernetes
targets
as
well.
The
main
benefit
of
Spinnaker
is
the
ability
to
create
end-to-end
pipeline
or
orchestration
workflows
that
allows
people
to
go
from
all
the
way
from
Court
check-in
to
Cloud
deployment
in
a
safe
manner,
including
built-in
Canary
and
blue
green
type
of
deployments
inspected
also
has
a
very
extensible
product.
A
You
can
extend
it
to
add
new
Integrations
new
custom
stages
so
that
you
can
deploy
to
any
type
of
applications
or
use
cases.
For
example,
you
can
use
it
to
deploy
it
to
infrastructure
using
terraform.
You
can
use
it
to
deploy
it
to
VMS
containers
functions
and
anything
and
database
updates
Etc
with
Spinnaker,
and
that
is
why
it
is
quite
popular
among
all
these
large
companies,
as
they
look
at
various
use
cases
that
they
want
to
have
a
unified
platform
to
be
able
to
provide
that.
A
So
with
that,
let
me
introduce
the
panel
quickly
and
before
they
introduce
themselves.
So
we
have.
We
have
three
panelists,
so
Asia
and
Abdul,
both
from
interswitch
into
switch
group
is,
is
a
large
financial
company
based
out
of
Nigeria
Africa.
We
also
have
Arun,
who
is
in
the
U.S
his
base
is,
is
at
Cisco,
so
I
want
each
of
the
panelists
to
introduce
themselves
a
little
more
we'll
start
around
the
table
with
RCF.
First
foreign.
B
D
First
of
all,
thank
you
for
the
invite
biology.
It's
been
a
pleasure
to
be
part
of
the
parent
discussion
along
with
Abdul
and
Asia.
My
name
is
Arun
and
I'm
senior
leader
driving
the
whole
devops
strategy
versus
goit,
apart
from
the
DeVos
portfolio,
I
also
Drive
the
test
automation
for
many
other
key
programs
for
Cisco.
Since
all
my
customers
are
developers,
I
really
obsess
about
the
whole
developer
experience.
The
goal
of
my
team
is
to
basically
provide
frictionless
Fast
Fun
CSD
experience
for
over
8000
Cisco
Developers.
A
A
Let's
talk
about,
maybe
we'll
start
with
you
on
on
on
the
what
is
the
usage
of
where
are
you
with
in
Cisco
in
in
Johannesburg
I
know
you
guys
used
for
a
lot
for
a
while
now,
but
maybe
give
you
a
little
bit
of
a
you
know
for
the
audience
who
are
not
familiar
with
Cisco's
usage
of
you
know,
essentially
the
Spinnaker.
Where
are
you
in
terms
of
adoption?
How
how
big
is
it?
Can
you
just
share
some
details
about
it.
D
Yeah
sure
I
believe
we've
been
using
spineco
for
almost
like
four
plus
years
now,
if
I'm
not
wrong,
and
it's
definitely
helped
improve
the
overall
development
experience
for
Cisco
developers.
We
got
about
like
six
plus
instance
and
that
just
keeps
growing
every
year
over
year
that
basically
caters
to
all
the
runtime
environments,
that
is
in
both
the
private
Cloud,
as
well
as
the
public
cloud,
and
this
year
we
also
been
looking
through
SAS
platforms
as
well.
D
We
have
about
like
300,
odd
applications
that
basically
does
the
Diplomat
using
Spinnaker
and
close
to
bar,
like
a
thousand
odd
softwares
that
basically
leverages
Spinnaker
in
a
typical
year.
We
we
have
about
like
300
000
departments.
That's
basically
across
all
the
life
cycles
Devastation
product-
and
this
is
just
a
business
Cisco
ID,
Realms
and
Spinnaker
basically
caters
to
maybe
40
of
those
deployments.
A
D
Yeah
so
currently
I
mean,
like
I,
said
that
we
have
like
six
different
instances
of
it
and
each
of
those
instances
basically
got
caters
to
a
particular
vertical,
like
Commerce,
like
sales
right
and
then
again,
we've
been
in
this
transition
of
basically
moving
from
our
Legacy
systems.
So
again,
overall
population
is
close
to
like
40
percent
of
it
is
what
obviously
goes
to
Spinnaker
and
we're
basically
ramping
up
this
Pinnacle
adoption.
A
Great
excellent,
all
right
now,
I
know
that
I
have
done
many
Spinnaker
sessions,
summer
sessions
or
specific
live
sessions
with
the
with
another.
You
know
your
colleague
and
if
people
want
to
know
more
of
her
they
can
they
can
listen
to
it
as
well.
In
the
past
past
years
now,
let's
switch
topic
to
inter
search,
maybe
ask
yeah:
you
can
go
first
and
talk
about
your
current
status
of
Spinnaker
at
the
infrastructure.
B
So
in
terms
of
application
size,
we
have
around
400
applications
for
development
environments
and
all
of
that
is
being
or
deploying
microservices
to
kubernetes
clusters
on
the
public
Cloud.
We
also
have
another
instance
which
deploys
to
production,
which
is
on-prem
so
that
has
around
I
think
a
bit
over
150
applications
deployed.
B
E
B
A
B
Yeah
so
just
to
add
some
I
guess
extra
details
in
adoption,
we've
been
able
to
use
some
features
through
OES,
such
as
the
policies
and
getting
some
pipeline
execution.
Data
from
the
audit
Trail.
So
I
know
that
audit
trail
that
I've
been
Gathering
data
since
the
since
we
started
using
Sri
Lanka,
so
we're
just
trying
to
get
more
information
from
it
so
trying
to
get
kind
of
like
the
dependent
lead
times
and
to
understand
what
benefits
pinata
has
brought
to
our
organization.
A
B
A
A
D
D
Spinnaker
basically
covers
most
of
the
workloads
that
we
basically
deploy
in
the
private
Cloud,
such
as
openshift,
kubernetes,
mulesoft,
infoamerica,
wsg
and
there's
a
new
data
variance
tool
called
a
data,
build
tool
that
we
just
enabled
for
and
a
few
other
runtime
environments
that
we
support
in
the
public
Cloud.
This
was
a
journey
that
we
started
like
a
couple
of
years
back.
We
support
many
of
the
services,
that's
on
AWS
and
gcp.
D
We
haven't
got
to
Azure
yet,
but
in
AWS
like
eks,
Lambda
S3
and
in
Google
we
we
support
Google,
Cloud,
one
gke,
and
then
one
of
the
SAS
platforms
that
we
did
some
level
of
enhancement
is
basically
for
snowflakes.
It
was
not
directly
into
the
slow
fixed
environment,
but
we
may
be
able
to
use
the
custom
stage
with
an
Spinnaker
to
be
able
to
deploy
it
into
an
artifact
that
can
basically
put
it
in
a
snowflakes
got.
A
It
I
know
we
did
I
did
a
session
with
Anil
right
on
the
snowflake
use
case
right
where
you
essentially
used
a
custom
stage
within
a
Spinnaker
I.
Think
this,
your
your.
What
you
just
described
is
is
it's
a
really?
You
know
where
Spinnaker
is
available
to
extend
beyond
what's
available
out
of
the
box
right
and
that's
something.
That's
been
a
great
use
case
for
somebody
like
Cisco
with
so
many
different
groups
that
want
to
do
so.
Many
things
I
think.
D
C
D
A
E
A
That
is
really
flexible
like
that
right
I
mean
there's,
obviously
Jenkins
you
can
sort
of
build
your
own
things,
but
as
well,
but
this
is
actually
a
pretty
modern,
fairly
modern
tool.
That's
available
for
variety
of
use
courses.
Thank
you
for
that.
Let's
maybe
Abdul
is
back.
I
mean
absolutely
want
to
add
anything
to.
A
E
So
because
we're
heavy
on-prem
one
uses
that
we've
been,
that
has
been
in
the
works
and
isn't
finally
there,
but
at
least
we're
making
progress
towards
it
is
deploying
to
on-prem
VMS
and
not
kubernetes,
so
the
whole.
How
do
you
now
take
your
traditional
monolithic
and
Legacy
and
deployment
process
and
how
to
use
and
serialize
them
within
a
Spinnaker
pipeline?
So
that's
something
that
we're
actually
looking
into
and
we're
watching
towards
got.
A
It
and
do
you
have
like
non-tr
like
VMS
containers.
Obviously
we
have
do
you
look
at
Spinnaker
to
like
update
databases
or
do
infrastructure
updates,
like
data
form
or
anything
like
that
and
again,
there's
a
specific
can
be
used
to
do
a
lot
of
things
right.
So
do
you,
do
you
see
any
use
cases
that
are
emerging
that
you
haven't
looked
at
yet.
A
Got
it
I
mean
I,
think
there's
a
yeah,
so
if
you
have
added
stages
that
customer
have
wanted,
for
example,
like
data
farm,
for
example,
allow
you
to
deploy
data
form
as
part
of
your
spinetic
pipeline
or
something
you
can
probably
use
all
your
using
already
great
I
I.
Don't
let's
switch
to
you
and
talk
about
any
use
cases
that
you
haven't
already
I
mean
you.
D
D
We
definitely
plan
to
basically
focus
on
the
more
Cloud
workloads,
the
ones
that
we've
not
enabled
so
far,
because
it's
quite
a
bit
of
people
I
mean
there's
quite
a
bit
of
azure
diplomas
happening
within
Cisco
ID
as
well,
but
it's
just
that
we
never
got
a
bandwidth
to
be
able
to
get
there.
So
we
plan
to
do
some
of
those
and
SAS
platforms.
Right.
D
Sas
is
kind
of
where
everybody's
headed
to
offline,
so
there's
quite
a
bit
of
Salesforce
and
Oracle
Cloud
related
departments-
that's
actually
coming
in,
so
we
enabling
particle
integration.
Clubs
as
we
speak,
and
then
we're
also
looking
into
Spinnaker
to
see
if
we
can
basically
leverage
Pinnacle
through
Salesforce
deployments.
Currently
we
actually
make
use
of
a
tool
called
compatible,
which
basically
is
the
CSE
tool
for
Salesforce,
but
we're
basically
branching
into
that
as
well.
Apart
from
these
capabilities
of
Spinnaker,
we
definitely
want
to
venture
the
whole
Brewing
deployment.
D
So
that's
one
of
those
things
that,
as
a
platform
team,
we
want
to
be
able
to
make
sure
that
we
have
high
availability
internally.
The
same
thing
goes
for
the
application
teams
as
well.
So
that's
one
feature:
we've
never
been
able
to
get
to
largely
because
of
the
size
of
things
that
we
have
to
do
within
Cisco.
So
hopefully
that's
something
that
we'll
be
focusing
the
next
physical
gear,
yeah.
A
You
for
that
I
think
I
mean
I'm,
actually
impress
you
know
looking
at
like
Salesforce
update,
Etc
I
need
to
I
need
to
probably
follow
up
with
you
on
that
on
that
for
another
day,
I
think
that's
pretty
good,
because
there's
actually
companies,
you
know
or
or
different,
like
you,
said,
tools
that
are
specifically
for
that
right
and
and
I.
Think
if
you
can
leverage
clinical
for
that,
you
know
then
then
you're
more
power
to
people
using
connector.
A
D
Yeah
so
currently
it's
kind
of
a
metal
process
as
well,
and
basically
they
do
a
deployment
to
put
Target
instances
and
then
basically
have
the
GSP
changes
done
to
be
able
to
point
out
to
another
infrastructure
right.
So
that's
how
we've
been
doing
it,
but
we
want
to
basically
change
it
and
that's
that's
the
reason
why
we
want
to
basically
understand
and
basically
work
on
that
aspect
of
it
and
every
team
does
it
differently.
I
mean
there's
quite
a
bit
of
teams
within
Cisco,
as
you
can
imagine,
so
everybody
does
it
differently.
A
Yeah,
there's
actually
quite
a
bit
of
activity
that
I'm
personally
trying
to
kind
of
you
know
create
solutions
for
because
there's
a
lot
of
customers
doing
different
things:
progress
to
delivery.
Obviously
right
you
wanna,
that's
all
canopy,
which
is
sort
of
obviously
thing
to
do,
and
also
Progressive
Terror
race,
which
is
kind
of
stepping
through
different.
You
know
loads
and
also
something
that
I
recently
started
looking
into
it
is
people
are
deploying
to
multiple
regions
or
multiple
clusters.
The
government
of
us
use
us
West
same
same
application,
but
you're.
A
You
know
deploying
to
different
servers
in
different
regions,
different
countries
or
different,
whatever
the
you
know,
where
you
do
it
and
then
now,
how
do
you
do
Progressive
delivery
of
that?
Maybe
you
deployed
to
air
cluster
or
a
region
whatever,
and
if
it
all
works,
fine,
then
you
sort
of
okay,
then
roll
out
of
the
remaining
clusters.
You
don't
want
to
like
spray
everybody
at
the
same
time,
so
there's
a
lot
of
variations
of
this
sort
of
so-called
Progressive
deliveries
or
safe
deliveries.
E
So
one
thing
we've
done
is
to
first
of
all
roll
out
to
a
different
name
space
and
just
to
ensure
and
make
sure
that
everything
is
good
and
fine
without
really
giving
giving
the
bulk
of
production
traffic.
Once
that
seems
fine,
then
we
actually
roll
out
to
the
darkshop
production
in
spaceware
production
traffic
hit
six.
So
that's
the
way
we
generally
generally
take
a
about
the
production
at
the
progressive
or
safety
delivery.
A
And
I
want
to
go
back
to
other
use
cases,
maybe
Abdullah
you.
You
talked
about
your
answer.
Yeah
is
around
using
policies
as
part
of
the
pipeline,
Abdullah
I
think
you.
You
had
presented
a
talk
in
the
last
city
con
or
last
minute
or
seven
or
something
around
policies.
How
are
you
coming
along
in
the
policy
space
around
automatic
automatic
policy
through
Opa
integration
within
Spinnaker
foreign.
B
Policies
we've
been
able
to
kind
of
limit
or
set
some
checks
for
pipelines
both
in
development
and
production,
so
something
as
simple
as
checking
the
checking
the
name
of
the
pipeline.
So
in
case
we
put
checks
on
development
pipelines,
we're
able
to
evaluate
all
development
pipelines
in
the
Smacker
instance.
So
something
like
making
sure
that
security
checks
are
made
on
applications
before
they
are
deployed
to
development,
environment
or
uat,
and
also
making
sure
the
quality
checks
are
done.
B
A
A
C
A
D
Yeah
we
actually
do
today,
so
the
thing
is,
as
you
can
imagine
like,
the
whole
complex
is
all
polishability.
All
of
those
controls
are
baked
in
the
Spinnaker.
We
don't
use
oppa,
but
we
do
have
a
custom
application
that
basically
does
all
the
policy
check
for
us.
It's
basically
exact
month,
based
implementation
that
we
currently
have,
and
we
try
to
automate
some
of
the
approvals
as
well
like
the
peer
review
right.
We
basically
try
to
understand
at
this
particular
software
as
peer
reviewed.
D
D
A
Really
great
one
thing
that
we
have
I
think
it
was
Abdullah
actually
who
gave
me
the
idea,
which
was
that
I
think
this
is
what
I
say
you
talked
about
in
solar
queue,
I
mean
one
of
the
common
use
cases.
If
you
look
at
the
Spinnaker,
the
other
manual
judgment,
which
is
just
sort
of
like
a
manually
like
you
know,
you
just
click,
yes
or
no
or
whatever
the
thing.
B
A
Then
you
know,
but
there
are
many
times
you
need
to
check
different
rules
whatever,
and
you
know
so.
One
of
the
some
of
the
rules
are
pretty
straightforward,
meaning
you,
you
know,
you
will
not.
You
know,
release
a
bad
release
or
bad
code
or
you
will
not.
You
know
approve
something.
You
know,
there's
a
very
strict
rules,
and,
and
essentially
you
can
enable
automated
checks
on
that,
and
that's
really
what
many
people
love
the
ability
to
do
that.
That's
great
now,
I
know.
A
E
A
C
A
The
ability
to
audit
everything
that's
going
on
right,
everything
is
is
stored
and
you
can
actually
then
mine
the
data
to
look
at
what's
happening
and
that's
really
powerful
or
you
know,
for
security,
audit
or
compliance
audit
and
those
kind
of
models.
I
think
you
guys
work
in
a
financial
sector,
so
you
have
more
potentially
control
I.
Don't
anything
any
comment
on
audit
or
other
things
you
you
are
using
at
Cisco.
A
A
D
Sure
thing
one
thing
is:
we
definitely
make
sure
that
all
our
configurations
of
Spinnaker
are
centralized.
We
actually
make
use
of
conjure
for
basically
showing
all
our
secrets,
so
it's
all
centralized.
That's
the
best
practice
that
I
think
might
be
a
good
thing
for
everybody
to
do.
Plan
for
frequent
advocates
for
each
of
the
microservices
part
of
the
pipeline.
D
Automation
for
upgrades
helps
avoid
all
the
human
errors.
So
today
we
basically
can
spin
up
like
a
Spinnaker
and
since
it's
just
a
minute
right,
so
we
have
all
the
automations
done
already
to
be
able
to
do
that
and
then
test
automation
for
all
the
upgrades
is
something
that's
that's
that's
key
for
us,
as
we
are
trying
to
basically
work
on
done
quite
a
bit
of
progress
and
we're
still
making
progress
on
that
with
respects.
To
Lessons
Learned
I
mean
the
custom
stage.
D
Option
is
like
beautiful,
so
that's
basically
helps
us,
basically
support
all
kinds
of
Target
environments,
so
we
actually
use
that
all
the
more
and
then
the
optimal
image
right
with
fix
for
both
Cloud
driver
as
well
as
VR
right.
The
cloud
report
startup
basically
used
to
take
30
minutes
right
now.
It's
like
three
minutes
or
less
than
three
minutes
and
also
the
field
service
memory.
Consumption
has
come
down
with
the
increase
in
like
service
accounts
and
pipelines.
It's
gonna
come
down,
so
that's
kind
of
things
that
we
actually
did
and
like
I
said.
D
A
Got
it
I
mean
I,
think
we
we
are
open,
sourcing
or
close
to
open
sourcing
project
called
stock
driver,
which
is
basically
it's
basically,
it's
a
distributed,
Cloud
driver,
so
that
allows
you
to
scale
the
accounts
in
for
kubernetes
or
AWS,
or
anything
like
that.
So
it
is
basically
sharding
I
mean
basically,
if
you
have
one
large
Cloud
driver,
you
know
if
you
have
to
restart
it
or
if
the
new
account
due
to
levels
happening
slow.
A
E
C
E
Significantly
changed
changed
how
we
use
how
your
stabilize
Spinnaker
it
has
been
and
again
making
sure
you
have
a
dedicated.
You
have
dedicated
resources
in
terms
of
whether
it's
a
dedicated
cluster
or,
what's
not
I,
see
I
mentioned.
That's
also
significantly
improves
the
stability
and
how
well
you
do
your
adoption
of
Spinnaker
is
also
touching
on
something
Aaron
mentioned:
custom
stages.
E
A
That's
a
good
one,
so
I
think
yeah.
You
guys
have
a
session
later
today,
talking
about
your
custom,
States
or
customer
integration,
which
is
great
I,
think
you
know
people
should.
You
know,
check
it
out
as
well.
I
think
that's
that's
a
good
one!
I
think
we
have,
for
example,
like
the
optimax.
We
have
done
like
70
plus
integration,
because
everybody
wanted
something
you.
A
View
wanted
some
things
that
we
did
some
Integrations
and,
and
somebody
else
wanted
something
I
think
it's
so
easy
to
do
it
right.
It
just
takes
for
us
a
couple
of
days
and
I'm
sure
like
I,
don't
and
your
team
seems
to
be
doing
on
your
own
I
mean
this
is
really
great
right.
Essentially,
you
don't
really
need
anything
in
that
sense
to,
and
it
will
extend
this
by
the
way.
Thank
you
very
much
for
all
of
you
guys
to
be
there
and
all
ICN.
A
We
will
share
the
contact
information,
I
guess
we
can
all
find
you
on
the
LinkedIn
or
or
Twitter,
and
hopefully
we
can
reach
out
to
you
guys,
and
if
you
have
any
questions
about
the
session,
we
will
be
in
the
Q
a
panel.
So
please
feel
free
to
ask
any
questions
and
we'll
be
answering
the
questions
along
the
way.
Thank
you
very
much
and
enjoy
the
rest
of
the
Spanish.
Thank
you
all
of
you.