►
From YouTube: Migrating experience from Spinnaker to Keptn
Description
Both CNCF Keptn as well as Spinnaker are popular tools amongst DevOps and SREs to automate delivery.
In this Keptn User Group Jonathan Racine, Sr. Consultant and SRE at SAP, (https://www.linkedin.com/in/jonathan-racine-639aa91/) is walking us through the reasons, migration experience and benefits of moving from Spinnaker to Keptn. In their case it helped them
* Deploy 80% faster than before
* 99% reduction from deployment code: from 16000 LoC to 150!
Session details:
00:00 - Introduction
01:43 - Before and After Migration Architecture
03:42 - Main Benefits of Keptn
06:48 - Live Demo
11:00 - Not everything was smooth
14:00 - What else is planned
16:45 - Thanks back to the Keptn Core Team
Learn more: https://keptn.sh
Get started with tutorials: https://tutorials.keptn.sh
Join us in Slack: https://slack.keptn.sh
Star us on Github: https://github.com/keptn/keptn
Follow us on Twitter: https://twitter.com/keptnProject
Sign up to our newsletter: https://bit.ly/KeptnNews
A
B
B
A
Yeah,
cool,
hey
and
I'm
very
glad
that
you
are
showing
us
your
kind
of
migration
experience
from
spinnaker
to
captain
and
a
lot
of
our
users
are
looking
into
spinnaker
and
other
ways
to
deploy,
and
it's
just
interesting
to
hear
why
you
decided
captain
how
the
whole
migration
process
works.
I
know
you
also
have
some
plans
to
talk
about
the
things
that,
in
the
beginning,
were
not
all
that
great,
because
every
journey
sometimes
has
also
some
stones
and
rocks.
A
B
So
yeah,
the
the
so
first
of
all,
the
the
big
thing
is
we
we
started
with
did,
as
I
said,
with
spinnaker
as
our
our
main
environment.
We've
we've
been
actually
jumping
from
different
technologies,
so
spinnaker
was
the
latest
iteration
and
it
was
mainly
based
on
the
fact
that
it
was
a
service
that
was
offered
or
promoted
within
sap.
B
However,
we
we
had
so
much
a
lot
of
technical
difficulties,
mainly
the
the
complexity
in
managing
the
the
application
code
complexity,
but
we
had
a
lot
of
slowdowns
and
we
also
had
some
integration
issues
that
should
have
would
have
been
solved
by
going
with
the
the
paid
version
from
from
spinnaker,
which
we
were
not
really
ready
to
do
as
we
are
mostly
pushing
towards
open
source
as
well
as
contributing
to
those,
so
captain
became
became
a
something
we.
We
started
to
look
at
so
yeah
on
this
slide.
B
What
you
you
can
see
is
the
this
is
the
deployment
architecture
we
that
is
simplified
and
the
whole
right
hand.
Side
was
mainly
spinnaker,
but
also
all
manage
inside
of
jenkins.
So
it
was
like
push
codes
to
to
get
our
application
running
and
with
replacing
spinnaker
with
captain.
Now.
The
only
thing
we
have
is
one
call
from
jenkins
to
captain
and
everything
else
is
now
fully
automated.
So
it's
a
lot
easier
for
us
to
manage
so
yeah.
I
said:
complexity
has
been
the
the
main
driver
for
for
this.
B
The
we
really
like
the
the
approach
with
the
shipyard
file,
which
is
shared
across
the
project,
and
it
it
it
just
reduce
the
overhead
for
everybody.
For
us.
It's
mainly
16
000
line
of
code
that
we
we
could
remove,
remove
from
the
application
by
going
to
captain,
because
every
every
application
had
a
different
spinnaker
deployment.
That
was
all
customized
and
we
were
never
able
to.
You
know
tackle
the
the
template
properly
so
that
you
know
the
code
would
be
easier.
B
B
A
Very
strong
state
and
it's
a
really
cool
statement,
and
and
thanks
for
sharing
this,
and
especially,
I
think
you
know
I
always
try.
We
always
try
to
show
the
community
what
great
things
captain
can
do
for
people
like
you,
but
I
think
this.
This
especially
is
great
also
for
the
captain
cool
team
that
has
initially
started
captain
and
now,
obviously,
captain
sees
a
lot
of
contribution
from
around
the
globe,
but
this
is
thanks
for
sharing
that
detail.
This
is
very
cool,
yeah.
B
B
So
this
is
a
small
screenshot
of
this
application
and
this
is
basically
a
a
dashboard
of
an
aggregate.
It's
an
aggregate
of
all
of
our
monitoring
solutions,
and
this
is
used
by
the
internal
monitoring
team,
especially
during
black
friday,
to
make
sure
that
all
of
our
customers
are
working.
You
know
at
a
hundred
percent
and
that
there's
no
issue.
B
So
it's
a
very
mission,
critical
system-
and
there
is,
we
need
to
be
really
agile
if
there
is
any
any
changes
that
occur,
and
we
need
to
be
sure
that
it
doesn't
impact
this
monitoring
system.
Otherwise
you
know
it.
It
could
have
major
issues
and
we
can
say
so
far
that,
with
with
our
captain
deployment,
we
never
had
any
issues.
The
deployment
always
went.
Fine,
so
again
that's
up
to
the
development
team.
It's
it's
working.
B
B
I
don't
I
need.
I
need
to
be
careful
with
all
types,
so
we
managed
to
integrate
the
istio
installation
within
captain
itself,
so
our
all
of
our
cluster,
the
sdo
system
and
ingress,
is
managed
right
into
kitten.
This
also
includes
the
upgrades
for
same
with
the
vault.
We
used,
the
sidecar
injector
and
the
ci
slide
driver
for
our
secrets
and
we
managed
to
get
all
of
that
integrated.
So
now
it's
just
a
one.
Click
deployment
or
a
github
commit
on
our
end.
Actually,
and
everything
goes
well.
B
B
So
yeah-
and
this
this
actually
is
one
of
the
the
main
part
that
was
missing
for
us
in
spinnaker.
Was
this
evaluation
and
that's
another
selling
point
for
for
captain,
for
us
is
a
being
able
to
monitor
actively
the
system
doing
rollbacks
or
having
you
know
any
idea
of
how
our
micro
services
is
doing.
It
really
is
it's
it's
a
life-saving
feature
for
us.
A
So
just
to
understand
this
correctly,
maybe
for
people
that
are
not
familiar
with
dynatrace,
sorry
with
captain.
We
can
see
here
that
the
build
id
the
build
number,
the
job
name.
This
is
all
stuff
from
jenkins,
so
jenkins
does
the
build
and
then
it's
triggering
a
captain
sequence
and
you
even
have
to
link
back
to
to
jenkins.
I
guess
that's
also.
B
A
B
A
B
A
Could
you
show
me,
can
you
go
back
to
the
sequence
details
if
you
click
on
go
back
here,
I
think
you're.
That
means,
as
you
mentioned
before,
right
you
have
about
180
lines
of
of
shipyard
that
actually
define
the
delivery
you're
doing
monaco.
You
first
configure
the
your
monitoring
environment,
then
doing
a
deployment
and
then
you're
doing
the
evaluation
from
a
deployment
perspective.
Do
you
use
just
the?
What
do
you
use
the
job
execute
or
the
helm
service?
What
do.
B
So
since
we've
been
with
the
project
since
the
beginning,
wasn't
you
know
smooth
sailing
all
all
the
time?
We
had
very,
very
big
issues
in
the
fut
in
the
beginning,
with
the
configuration
service,
because
we
were
doing
too
many
commits
to
the
code
with
the
five
deployments
per
per
minutes,
and
it's
also
because
we're
using
a
github
server,
that's
internal,
so
we
were
overloading
that
github
server
and
then
you
know
the
captain
was
not
able
to
to
fetch
the
data
as
fast
as
it
could.
B
The
new
resource
service
actually
fixed
everything
that,
for
us,
we
just
had
to
edit
a
bit
the
the
resource
usage
on
it,
increasing
memory
limits
and
cpu,
and
now
we
never
heard
of
it
anymore.
So
super
super
great
on
that,
and
the
only
there's
two
two
small
pain
points
where
we're
still
looking
at
is
since
we're
using
the
helm
service.
B
When
you
do
a
deployment
and
then
there's
an
issue,
if
there's
a
an
error
within
hell
and
a
rollback
occurs,
we
do
lose
some
of
the
the
the
logging,
so
we
have
to
to
do,
go
to
the
cluster
itself
and
and
look
at
it.
You
know
actively
and
that's
one
of
the
reason
why
we
also
are
looking
at
the
job
executor
and
have
a
bit
more
control.
A
B
A
Also,
this
is
for,
for
people
that
are
maybe
new
to
captain.
Here
is
the
first
time
we
traditionally
with
captain
also
ship
a
so-called
helm
service
that
can
do
basic
helm
deployments,
but
the
job
executor
is
a
service
that
allows
you
to
deploy
any
type
of
container
as
a
kubernetes
job,
to
do
basically
any
type
of
task
right,
and
it
gives
you
a
lot
of
flexibility.
A
Like
doing
you
know
more
logging
or
doing
actually
with
as
part
of
that
deployment
itself
doing
some
troubleshooting
already,
and
I
think
we
see
a
lot
of
people
now
leveraging
the
job
executor
more
and
more.
We
also
have
some
thoughts
to
really.
You
know
really
make
people
aware
that
the
helm
service
from
captain
is
great
for
demos
and
for
initial
pocs,
but
we
know
it
has
limitations
and
therefore
the
job
executor
is
definitely
the
way
to
go.
Yeah.
B
Yeah
I'd
say
the:
as
you
said,
the
helm
service
has
been
for
us.
It's
been
a
way
to
rapidly
migrate
from
spinnaker
to
captain,
have
all
of
our
service
working
and
and
stabilized
so
that
we
get
at
least
the
skeleton
in
there,
and
then
we
can
start
to
to.
You
know,
get
some
something
even
more
more
customized
in
the
in
the
future.
So
that's
that's.
What
we're
we're
looking
at
right
now.
B
So
yeah,
so
in
the
future,
what
we
we
are
planning
again
the
job
executor,
is
one
we
want
to
deploy
the
dyna
trace
operator
within
captain
itself.
So
that's,
that's
a
you
know,
it's
a
a
a
challenge
in
itself,
since
we
want
to
use
the
sli
in
slo
dashboard.
So
it's
a
chicken
and
the
egg
problem,
but
we
we
still
want
to
at
least
be
able
to
to
manage
or
update
the
dynares
operator
within
captain
in
the
future.
B
Sli
and
slo
is
still
a
a.
This
is
like
a
ongoing
topic
and
it's
always
going
to
be
there.
We
are
always
improving
him,
trying
to
get
better
dashboards
and
so
on
and
the
the
big
biggest
project
which
is
going
to
be
the
for
us.
It
would
be
the
last
big
step
would
be
to
be
able
to
deploy
captain
with
captain,
so
we
have
two
instances,
one
for
our
development
environment,
one
for
production
and
we
want
to
have
them
interlinked
with
each
other
so
that
we
can
do
everything
as
code.
A
B
A
It
yeah,
hey
jonathan,
thank
you
so
much,
I
think,
just
a
great
story
and
a
great
journey.
Thank
you,
especially
for
sticking
with
us
from
the
beginning,
because,
and
we
know,
captain
came
a
long
way
and
the
path
is
not
yet
over.
So
there's
more
stuff
coming
and
yeah.
Thank
you.
So
much
also
for
sharing
the
story
here,
I'm
pretty
sure
if
people
are
kind
of
in
the
same
shoes
as
you
are
not
really
sure,
if
maybe
spinach
has
something
for
them.
A
It's
great
that,
with
this
you've
shown
them
that
captain
is
definitely
helping
you,
I
think
you
said
something
from
instead
of
sixteen
thousand
lines,
you're
down
to
a
couple
of
lines
for
the
sheep
file,
eighty
percent
increase
in
in
in
speed
I
mean
this
is
just
great
testimonials.
Thank
you
so
much
yep.