►
From YouTube: In the Clouds (S2 E5) | Business value of GitOps
Description
Modern software development relies on concepts such as CI/CD and DevOps, but how does GitOps improve the way applications are developed, deployed, and updated? In this episode, Red Hat's Siamak Sadeghianfar and Cornelia Davis from Weaveworks join the show to talk about the business value of GitOps, modern application development in the cloud and how Kubernetes and GitOps mix.
About In The Clouds:
Red Hat’s senior leadership is having to execute at an ever-increasing pace. That means that today's technology decisions have to balance short-term risk with long-term gains. This unique series provides host Chris Short inviting thoughtful and candid discussions with each guest.
Follow us:
Cornelia Davis https://twitter.com/cdavisafc
Weaveworks https://twitter.com/weaveworks
Siamak Sadeghianfar https://twitter.com/siamaks
Chris Short https://twitter.com/chrisshort
OpenShift https://twitter.com/openshift
Subscribe to Red Hat's YouTube channel: https://www.youtube.com/redhat/?sub_confirmation=1
About OpenShift:
Red Hat OpenShift is an open source container application platform based on the Kubernetes container orchestrator for enterprise application development and deployment.
Openshift.com http://red.ht/3nJ6AKX
OpenShift on github https://github.com/openshift/
OpenShift on redhat.com http://red.ht/34xQoER
#RedHat #OpenShift #HybridCloud
A
Good
morning,
good
afternoon,
good
evening,
wherever
you're
healing
from
welcome
to
another
edition
of
in
the
clouds
here
on
openshift
tv,
I
am
chris
short,
the
host
of
most
shows
I'm
here
on
openshift
tv,
also
the
executive
producer.
I
am
joined
today
by
the
one
and
only
cornelia
davis
of
weaveworks
cornelia.
How
are
you
today.
B
B
A
B
A
That's
awesome.
Well,
thank
you
for
joining
us
today
really
appreciate
it
today.
We're
going
to
be
talking
about
the
business
value
of
get
ops
right,
there's
a
lot
of
buzz
around
the
term
git
ops!
There's
a
lot
of
you
know,
fud
too,
as
well
with
that
buzz
right
like
like
anything
in
tech,
it's
the
inevitable,
but
before
we
dive
into
that,
I
have
a
very
important
question
for
you.
It's.
A
C
Hello,
everyone
I
apologize
for
being
late
a
little
bit
day.
It's
exactly
like
you
said
I
was
stuck
in
a
meeting
was
trying
to
drop
off.
My
name
is
siamak.
I'm
a
product
manager
at
red,
hat
part
of
the
cloud
platforms,
team,
I'll,
look
after
everything,
ci
cd
on
openshift.
That
includes
things
like
argo,
argo,
cd
in
the
github
space,
open
your
pipelines
and
take
time
within
the
like
ci
ecosystem
and
other
related
bits.
A
Awesome,
thank
you
for
joining
us,
so
here
I'll,
ask
the
question
of
both
of
you
and
we
can
discuss.
There's
no
wrong
answer
here.
Cake
or
pie
is
cheesecake
a
cake
or
a
pie.
What
do
you
think
do
you
cmac?
Do
you
even
eat
cheesecake
for
that
matter?
Right,
like
I
feel
like
there
might
be
some
dietary
restrictions,
maybe
for
food.
B
There
is
always
a
way
out
so,
first
of
all,
why
would
you
have
to
choose?
Because
cheesecake
is
so
delightful?
It
is
definitely
one
of
my
faves
and
I
am
a
cook,
and
so
I
can
tell
you
that
I
have
made
cheesecakes
that
have
flour
in
them
and
I
have
made
cheesecakes
without
flour,
and
so
one
you
know,
one
you
might
say
is
a
little
closer
to
cake.
One,
you
might
say,
is
a
little
closer
to
pie.
C
I
think
like
she
gave
the
most
politically
correct
answers.
I
don't
know
how
to
how
to
top
that.
But
to
me,
cheesecake
is
a
is
a
postmodern
pie.
You
know,
okay,
you
it's
it's
a
it's
an
artistic
pie
done
well,
but
regardless
I
love,
cheesecakes
passion.
Fruit
cheesecake
is
definitely
my
favorite,
so
yeah.
A
B
Yeah,
so
so
the
the
like
many
other
kind
of
open
source
tools
and
I'll
talk
a
little
bit
about
how
we
we,
what
that
journey
was
it
started
with
us
doing
our
own
work.
So
we've
works.
Wasn't
a
company
that
started
out
as
the
good
ops
company
it's
been
around
for
quite
a
number
of
years.
It
kind
of
started
out
as
network
networking
for
container
networking,
and
then
it
moved
into
kind
of
you
know
more
squarely
into
the
kubernetes
space
and
started
dealing
with
observability.
B
So
we
had
a
sas
offering
that
we
call
weave
cloud
which
offered
essentially
prometheus
as
a
service.
So
you
could
register
your
cluster
and
and
then
you
could
go
and
get
dashboards
and
all
of
that
stuff
and
see
what
was
going
on
in
your
kubernetes
clusters
and
we're
running
that
sas
service
and
we
had
an
outage
and
we
were
able
to
like
a
complete
outage
like
a
total
meltdown.
That's.
A
A
B
Yeah
the
whole
thing,
the
whole
sas
system
top
to
bottom
and
our
ceo
alexis
richardson,
who
actually
coined
the
term
git
ops
said:
how
did
you
do
that
and
the
engineers
the
sres
described,
that
what
they
had
done
and
the
whole
system
was
built
on
kubernetes
and
and
those
principles
that
they
said?
Well,
we
have
everything
we
have
all
of
our
declarative
configurations
stored
in
a
git
repository
and
all
we
needed
to
do
was
we
needed
to
point.
B
What
is
the
tooling
that
supports
these
practices
that
we
just
did,
and
that
was
the
the
genesis
of
creating
the
flux
open
source
project
which
provides
some
set
of
that
convergent
automation,
not
all
of
it,
but
some
set
of
that
convergent
automation.
It
basically
handles
delivering
things
from
github
into
your
fcd
and
kubernetes,
that's
to
a
large
extent
with
what
flux
is
focused
on.
So
that's
the
genesis.
A
That's
pretty
crazy
and
also
a
very
common
pattern
too,
at
the
same
time,
but
like
the
story
of
hey
everything's
down
and
just
blow
it
away.
You
know:
that's
amazing,
absolutely
amazing!
The
and
here
we
go.
We
have
some
folks
on
the
line
here,
satyam
and
so
forth,
so
it
out,
if
feel,
free,
to
ask
questions.
Folks,
if
you
have
them
about,
you,
know,
get
ops
and
how
to
implement
it
and
so
forth
so
on.
B
Yeah,
so
I'm
happy
to
jump
in
again
as
well.
So
I
think
the
first
thing
that
I
would
say
is
that
I
I'm
not
a
big
fan
of
ci
cd
as
one
word
and
I'm
not
suggesting
that
you
used
it
as
one
word,
you
very
likely
used
it
as
too,
but
ci
is
something
that
we
as
an
industry
have
have
really
kind
of
mastered.
We've
been
working
on
it,
for
I
don't
know,
15
20
years
or
something
like
that.
B
The
tooling
has
matured
a
lot.
The
understanding
of
the
practices
have
matured
a
lot,
and
so
we
we've
been
doing
ci
and
then
we
kind
of
extended
that
a
bit
and
said:
okay,
well,
we'll
extend
that
ci
out
to
delivery,
and
so
that's
why
we
often
hear
cicd
is
because
cd
is
kind
of
it.
Really.
It
is
representative
of
the
way
that
a
lot
of
folks
do
it
as
they
kind
of
tack
on
cd
as
the
last
step
in
ci.
B
What
we
do
with
get
ops
is
actually
we
start
with
cd.
We
start
with
the
continuous
delivery,
we're
not
trying
to
disrupt
those
mature
ci
processes,
but
we
start
with
cd,
but
we
take
it
further
than
cd.
So
I
like
to
say
that
get
ops
is
continuous
delivery,
so
cd
and
co
continuous
operations
so
get
ops.
Remember
it's
get
ops,
it's
not
get
delivery,
it's
get
ops,
and
so
it's
all
about
linking
what
you're
doing
in
git
and
using
git
as
the
interface
not
for
delivery,
but
using
git
as
the
interface
for
operations.
B
So
we
want
to
link
all
the
way
out
to
hey
this
thing
has
deployed.
How
is
it
working?
Is
there?
Are
there
any
adjustments
that
need
to
be
made,
etc,
etc?
So
so
that's
that's.
I
think
one
of
the
the
first
fundamental
differences
is
that
we
see
get
ops
as
cd
and
then
beyond,
and
then
I
already
hinted
at
some
other
things,
which
is
convergence.
B
So
we
are
pretty
adamant
that
gitops
is
modern
cloud
native
operations
and
the
only
way
that
you're
going
to
be
able
to
do
that
is
if
you
have
an
eventually
consistent
system
and
that
your
automation
is
built
in
such
a
way
that
it's
convergent
and
I
think
we'll
talk
a
lot
a
lot
more
about
what
that
means
as
we
go
along.
But
I
want
to
give
cm
mac
a
chance.
C
That's
piece
exactly:
I
wanted
to
say
like
that.
That
piece
is
extremely
important,
because,
like
one
of
some
of
the
comments
that
I
get
when
I
talk
to,
like
customers
and,
unfortunately
not
much
conferences
to
go
these
days
to
get
connected
with
the
community
in
better
ways,
but
in
a
conversation
that
come
up
is
that
well,
this
is
we
have.
We
always
have
had
our
stuff
in
git.
C
So
this
this
is
nothing
new
or
infrastructure,
as
code
is
really
is
one
of
the
devops
principle
that
has
been
around
for
a
long
time.
So
it's
like.
Why
do
you
treat
git
ups
as
as
a
new
new
thing,
but
that's
exactly
where
this
flavor
of
ci
cd,
that
we're
talking
about
is
is
is
different
from
how
traditionally
people
have
been
using
git
and
declarative,
configs
and
infrastructure
as
code
that
is,
is
not
really
sufficient
that
we
have
declared
everything
in
a
negatory
poll
and
push
that
to
a
certain
environment.
Forget
about
it.
That's
really.
C
The
main
differentiator
of
when
we
talk
about
git
ops,
compared
to
just
using
git
and
declarative,
config
and
isc,
is
that
the
work
is
not
done
by
fire
and
forget
that
you're
constantly
in
this
loop
of
monitoring
the
state
of
their
applicant,
the
goal
is
that
application
keeps
running,
and
it's
in
the
state
that
you
you
expect
and
not
just
application
like.
This
is
something
that
I
sometimes
struggle
with,
that
get
ups
quite
often
becomes
reduced
to
just
being
a
part
of
the
cicd
flow
or
application
delivery,
but
it's
really
a
way
of
operation
right.
C
It
applies
to
specifically
kubernetes
operation,
a
lot
and,
in
my
experience,
a
major
great
majority
of
teams,
organizations
that
start
adopting
git
ups
practices.
They
first
actually
do
it
for
cluster
operations,
cluster
configurations
rather
than
delivering
applications.
It's
the
platform
owners
that
start
with
it
and
then
when,
when
they
feel
comfortable
using
that
and
see
the
value
they
start.
Thinking
of
how
could
I
offer
this
to
my
dev
teams
now
app
teams
that
to
use
the
same
workflow
for
for
the
application,
so
that
that
monitoring
decline
that
continuous
operation?
C
That,
currently
I
was
talking
about
in
my
mind-
that's
really
a
huge
differentiator
compared
to
yes,
these
are
the
practices
that
exist
that,
but
the
reason
that
the
collection
of
these
practices
have
have
a
new
name
is
that
it's
a
different
angle
perspective
in
in
the
way
that
these
practices
are
used
together
with
an
extension
of
the
way
that
we
used
to
work.
So
it's
a
so
that's
really
a
core
of
it.
For
me,
at
least.
A
That's
awesome,
so
it's
it
is
kind
of
a
paradigm
shift
right,
like
you
have
to
think
of
things
as
well.
No
wait,
don't
touch
it.
It'll
fix
itself
right
like
right.
That
is
something
that
people
like
you
get
a
page
and
you
just
kind
of
wait
until
reconciliation
happens
right
like
you,
don't
necessarily
have
to
engage
the
system
other
than
to
make
sure
the
system
is
doing
what
it's
supposed
to
do.
C
Yeah
absolutely
this
is
this
is
actually
a
little
like
funny
anecdote
that
this
before
this,
like
this
mind,
shift
sometime
comes
as
as
a
funny
moment
to
to
a
lot
of
the
teams
and
and
and
customers
that,
like
you,
need
to
get
to
this
understanding
of
I'm
not
support
to
supposed
to
manually
doing
anything
on
the
cluster
anymore
right.
So
this
is
not.
This
doesn't
come
naturally,
and
it
happens
often
with
teams,
starting
that
they
do
something
that
they
see
it.
Doesn't
it
doesn't
work
like
they?
They
apply.
They
do
this.
C
C
So
this
this
mindset
of
like
you're
there's
someone
in
the
middle
that
makes
sure
that
there
is
no
one
manipulating
in
the
middle
unless
that
thing
is
in
in
the
git
repos
right,
so
that
that
mindset
definitely
is
something
that
is
new
and
even
for
teams
that
are
actually
actively
like
adopting
these
practices.
It
it
does
take
a
little
bit
of
time
to
come
to
this
mindset
of
yes,
I
I'm
not
supposed
to
touch
the
cluster
directly.
B
B
B
B
So
it's
that
convergent
automation
in
order
to
allow
you
to
make
git
be
the
interface
for
operations,
because
you're
just
chain
I
mean
just
because
something
went
down,
doesn't
mean
you
need
to
change
your
declarative
configuration.
Actually,
my
declarative
configuration
was
right.
I
just
need
my
automation
to
run
again.
B
Exactly
but
but
I'm
going
to
take
your
scenario
and
I'm
going
to
extend
it
just
slightly,
so
you
get
a
page,
something
goes
wrong
and,
and
you
investigate
and
you
actually
find
that
my
application,
which
is
running
in
aws,
u.s
west
one,
is
down
because
u.s
west
one
is
down,
and
so
you
it
you
could
wait
for
a
while
and
it
won't
come
back,
and
so
if
your
operations
operational,
you
know
your
run
book
now
says.
Well.
If
one
region
goes
down
point
to
a
different
region.
B
What
is
your
your
operational
task
now
you've
been
paged.
Your
operational
task
is
what
well
remember.
Git
is
the
interface
for
operations.
What's
my
operational
task,
I'm
going
to
go
into
github
and
I'm
going
to
say:
oh
yeah,
I
was
running
in
u.s
west
one.
I
now
want
to
run
in
u.s
east
2.,
so
you
make
a
change
in
git.
That
says.
Actually
I
want
to
run
this
in
u.s,
east
2.
and
then
the
convergent
automation
says:
oh,
oh
okay,
you
don't
want
to
be
in
west
one
anymore.
B
A
That
right,
there
is
like
a
selling
point
for
me
right
like
right.
I
can
do
dr
just
by
changing.
You
know,
east
to
west
right
yep,
or
vice
versa.
That's
pretty
powerful
stuff
like,
and
the
next
question
I
have
is
like
what
makes
get
ops
a
better
practice-
and
I
say
practice
intentionally,
but
that
is
right
there.
One
of
those
reasons
why
it
all
it
takes
is
just
one
commit
and
your
applica,
your
application
and
infrastructure
is
getting
deployed
elsewhere
off.
C
A
A
B
Of
either
one
either
one,
I
was
giving
making
sure
that
I
wasn't
always
the
first
one
to
go,
but
I'll
I'll
go
again.
So
so
we've
been
talking
a
lot
about
this
convergent
this.
This
convergence,
this
concept
of
convergence
and
how
central
it
is
to
get
ops,
and
so
my
recommendation
to
people
who
are
starting
and
this
doesn't
it
doesn't
preclude
just
about
anybody,
because
I
think
we
found
that
90
of
all
enterprises
are
already
somewhere
on.
B
B
Configuration
they've
probably
created
some
scripts
to
apply
that
yaml
or
they
might
be
applying
that
yaml
from
within
the
ci
system,
for
example,
or
they're
doing
a
cube,
cuddle,
apply
and
so
linking
together
that
now
taking
that
declarative
configuration
that
they
already
have
and
now
using
git
ops
practices
to
manage
it
in
a
git
repository
to
manage
that
your
folder
structure
to
decide
who
your
access?
What's
the
access
control
that
you're
going
to
layer
around
it?
Who
are
the
people?
What
are
the
kind
of
approval
cycles
that
you're
going
to
put
in
to
get
layering?
B
All
of
that
around
this
declarative
configuration
and
then
adding
a
you
know,
a
convergent
delivery
mechanism
like
flux
or
argo
cd,
to
link
that
git
repository
over
into
the
kubernetes
cluster
is
a
really
there's
very
little
that
you
have
to
do.
You
basically
have
to
start
source
code,
controlling
your
yaml
configurations
and
installing
some
a
utility
or
two,
and
then
you
can
start
to
get
some
practice
at
get
up.
B
C
C
They
there
are
still
like
huge
law
for
python
scripts
and
batch
scripts,
and
there
are
different
types
of
scripts
or
even
and
using
declarative
tools
in
a
non-declarative
manner
when
it
comes
to
ansible
and
and
popping,
and
things
like
that,
so
there
is
there's,
there's
a
lot
of
what
tendencies
to
that
way,
you're
working,
which
is
which
makes
it
more
difficult
to
start
with
github,
because
you
really
have
to
redo
all
of
that
and
figure
out
a
way
to
that.
You
can
declare
these
kind
of
things
from
a
different
perspective.
C
C
We
authorized
to
make
these
changes
through
git
who
is
gonna
approve?
They
are
the
ones
that
proving
it.
You
know
they
are
the
one
owning
it,
so
they
have
less
of
a
concern
on
access
control
and
how
they
have
to
fit
that
into
the
git
ops
model
does
do
they
have
to
rely
on
the
git
provider
access
model,
or
can
they
how?
How
should
the
git
repost
structure
look
like
if
they
want
to
reconcile
the
app
teams
with
the
cluster
config?
C
And
things
like
that,
so
there
are
other
questions
become
come
into
the
picture
so
depending
on
which
side
of
this
is
stronger
for
a
particular
organization.
I
think
they
they
might
be
more
inclined
to
in
one
direction
or
other
right.
If
a
platform
team
has
been
extremely
imperative
like
doesn't
matter,
they
don't
have
to
think
about
them
all
the
tenancy.
C
It
becomes
a
huge
hurdle
for
them
to
be
able
to
get
on
the
get
on
this
workflow,
but
on
the
other
hand,
if
they
have
been
on
a
kubernetes
destro
or
they
have
been
already
like
thinking
themselves,
even
declaratively,
it
becomes
much
simpler
because
they
have
less
of
a
hurdle
on
their
multi-tenancy
part,
or
maybe
there
are
now
we
are
seeing
this.
I
I
think.
C
They
they're
having
more
and
more
of
it
getting
more
comfortable
with
it,
so
they
are
they're
coming
to
the
day
two
challenges
of
having
kubernetes
that
they
they
weren't
aware
of
these
challenges.
Yet
three
years
ago,
four
years
ago,
in
the
beginning
of
their
their
journey,
so
they're
there,
there
are
organizations
that
get
to
a
level
that
they
are
giving
a
cluster
to
each
application
team
and
and
for
those
ones,
the
model
that
cornell
described
fits
perfectly,
because
there
is
no
multi-tenancy
concern
at
all
you.
C
C
B
I
I
wanted
to
circle
back
because
you,
you
talked
a
little
bit
about
you
briefly
mentioned
before
we
got
onto
this
latest
thread
about
you
know
what
makes
git
ops
better.
You
know
some
of
these
kind
of
business
values
and
that's
kind
of
the
title
of
our
of
our
session.
Here
I
wanted
to
highlight
another
one
that
I
think
is
really
interesting,
that
really
kind
of
dovetails
two
things
together,
and
that
is
you
know.
B
I'm
a
big
fan
you've
seen
me
do
talks
before
I'm
a
big
fan
of
the
the
work
that
dora
has
done
in
the
dora
metrics,
and
so
what
we
talked
about
before
with
disaster
recovery.
Well,
the
dorametric
that's
tied
to
that
is
mean
time
to
recovery.
So
how
quickly
can
you
recover
a
system
but
there's
another
one
that
I'm
a
big
fan
of,
which
is
very
squarely
in
kind
of
the
operation
space,
which
is
the
change
failure
rate.
A
B
And
one
of
the
things
that's
super
interesting
about
get
ops
is
that
it
allows
you
you
know
again.
Going
back
to
this
notion
of
git
is
the
interface
for
operations,
is
that
before
we
had
a
collaborative
a
really
like
a
tool
that
was
designed
around
collaboration
around
change
control,
the
way
that
change
control
typically
worked
was
we
did
all
of
this
work
and
then
kind
of
the
last
step
before
we
pushed
things
out
into
production.
We
had
this
elaborate
ceremony
where.
A
A
B
So
what
get
ops
allows
you
to
do?
Is
it
allows
you
to
shift
that
control
that
change
control
left?
So
now
you
are
doing
these
checks
on
your
you're.
Storing,
now
configuration
changes
for
your
production
system,
you're,
storing
those
in
git,
and
so
now
you
can
have
all
of
the
people
have
the
best
context
as
approvers
in
that
git
repository
and
you
can
set
it
up
so
that
you
have
to
have
three
approvers.
You
have
to
have
this
approver
as
required
in
these
other
two,
two
of
the
other
five.
B
You
know
that
type
of
thing,
and
so
there's
this
notion
of
collaboration
so
get
get
up,
supports
collaboration
around
change
control,
so
broad
visibility,
lots
of
different
eyeballs.
It's
not
so
checklist
based
it's
in
context,
which
I
think
is
really
important
right,
but
then
there's
the
other
part
which
also
lends
to
this
lowering
the
change
failure
rate,
which
is
okay.
B
B
I
recognize
oh
and
I
might
even
have
an
automation
like
flagger
or
something
that
is
helping
me
with
this
progressive
delivery.
At
the
first
sign
of
trouble,
we
go
oh,
never
mind.
Let
me
go
back
to
the
previous
version
and
so
that
those
two
things
in
combination,
collaboration,
git
based
collaboration
and
the
fact
that
I
can
just
snap
back
to
an
earlier
part
in
my
version,
history
and
I
have
a
complete
representation
of
what
that
looked
like.
So
I
can,
you
know,
go
back
to
that
state.
B
That
is
a
game
changer
for
change
failure
rate,
reducing
change
failure
rate.
So
that's
another.
Just
perfect
example
of
how
git
ops
changes
this
whole
devops
scene.
I
mean
the
state
of
the
devops
report
has
been
around
for
longer
than
the
term
git
ops
has
been,
but
it
really
get.
Ups
really
supports
the
whole
devops
agenda
in
a
really
particularly
compelling
way.
A
C
That
that's
the
actually
what
I
usually
point
out
when
I
talk
to
customers
and
they
they
have
this
question
of
like
they
haven't
been
too
involved
in
in
this
space
if
it
is
get
offs
like
what
comes
after
devops
or
what?
How
is
that
like
related,
but
this
is
this-
is
exactly
like
the
the
relationship
is
so
the
core
of
devops
is
almost
a
cliche
now
that
that
is
a
collaboration
right,
you've
seen
hundreds
of
presentations.
That
start
with
this
slide
of
collaboration,
which
is
like
useful
information
to
have.
C
C
One
of
them
in
every
team,
or
how
am
I
gonna?
What?
How
are
we
gonna
collaborate
more
like
give
me
the
recipe
right?
This
is
the
the
question
like.
I
think
that
that
has
been
one
of
the
like
challenges
of
devops,
that
I
see
that
it
from
the
value
to
the
practice
was
a
there's,
a
big
gap
of
how
do
I
realize
that
right,
it
was
left
a
lot
to
imagination
and
interpretations
and
git
ops
really
like
gives
gives
that
comes
with
a
prescriptive
process
for
that
that
realizes
those
values
right.
C
When
was
the
last
time
you
talked
to
a
team
that
is
not
using
git,
so
a
large
portion
of
these
workflows
are
known
and
it
connects
certain
dots
for
them.
That
was
not
this
disvisible
so
that
that's
how
I
is
that's
one
of
the
big
values
that
I
see
out
of
like
really
putting
this
together
and
like
going
back
to
like
the
the
term.
C
I
do
think
that,
even
though
these
practices
are
not
new,
I
do
see
a
lot
of
benefit
in
in
in
the
term,
thanks
to
vivo
works
for
for
coding,
the
term,
because
it
brings
attention
to
the
packaging
of
these
practices
in
a
new
way
and
how
they
relate
to
each
other,
how
they're
used
together.
It
helps
a
lot
with.
With
that
perspective,
in
my
opinion,.
A
A
C
I
I
can't
start
with
that
one,
so
it
is,
there
are
like
varying
degrees
of
that
that
they
can
see.
I
like
frankly,
I
can't
say
I've
seen
a
like
total
shift
of
culture
across
the
entire
organization,
but
I
do
see
especially
like
talking
with
the
same
teams
over
like
different
quarters.
You
could
it
changes
and
the
the
level
of
collaboration
how
it
increases,
especially
like
I
said
like
in
within
the
enterprise
context
and
customers,
the
type
of
customer
the
red
hat
has,
and
we
we
talk
to
very
often.
C
Let
go
of
some
of
the
controls
that
a
lot
of
these
organizations
have
at
the
lower
levels
of
infrastructure,
and
it
it's
difficult
for
me
to
judge
if
git
ops
is
the
reason
for
that,
or
that
was
the
reason
they
adopted
get
us,
but
they
they
go
hand
in
hand
a
lot
in.
In
many
of
the
conversations
we
have
with
these
people,.
A
B
Yeah,
and-
and
I
would
concur
with
that-
I
I
I'm
not
sure
that
git
ops
is-
has
been
the
forcing
function
and
I'm
I'm
right
there
with
you,
chris
that
get
ops
again.
It
serves
the
devops
agenda,
particularly
well,
but
I
think
that
the
devops
agenda
is
the
thing
that
has
been
forcing
the
cultural
change
more
than
more
than
get
offs
as
a
practice
in
specific.
B
It's
become
rarer
and
rarer.
I
mean
back
in
my
earlier
pivotal
days.
It
most
organizations
still
had
a
very
kind
of
traditional
I.t
structure
where
they
had
the
app
developers
were
super
siloed
from
and
they
threw
things
over
the
wall,
and
there
was
an
operations
team
and
the
operations
team
wasn't
operating.
B
You
know
just
infrastructure,
they
were
operating
infrastructure
and
the
applications,
despite
the
fact
that
they
didn't
have
the
context
of
the
applications
and
so
that
bifurcation
of
developers
just
do
development.
They
don't
do
operations
and
operations
just
as
operations,
they
don't
do
development
and
by
the
way
they
do
operations
of
the
entire
stack.
That
was
still
quite
prevalent.
I
would
say
six
seven
years
ago,
what
I'm
finding
now
is
this
this
this
realization
of
a
platform
team
and
cmac
earlier
used
platform
owner
this
notion
of
platform
and
platform
team
has
become
pretty
darn
common.
B
A
B
Versions
that
they
update
the
software,
you
know
the
version
of
prometheus
that's
running
on
there,
and
so
what
they've
done
is
they've
self-served
infrastructure,
essentially
by
giving
a
full
kubernetes
cluster
with
you
talked
about
cmc
no
multi-tenancy
there,
so
you
can
do
anything
you
want
on
it
right,
meaning
you
could
do
good
things.
You
could
do
bad
things
with.
B
A
A
B
So
I
mean
we
can
start
with
the
high
level
marketing
buzz
around
get
ops
and
auditing.
We
go.
Oh
it's
easy.
So
this
is
the
the
marketing
it's
easy
part
which
says:
hey,
everything's
in
get
and
everything's
recorded
and
get
it's
the
only
interface
to
operations.
So
every
single
change
made
to
your
your
environment
is,
you
can
see
who
made
it
who
approved
it
and
it's
in
in
a
git
repository
somewhere.
B
We
just
talked
about
how
the
platform
team
has
responsibilities
for
part.
Some
layers
in
the
system
and
the
application
teams
have
responsibility
for
some
layers
in
the
system,
and
so
this
is
an
area
where,
yes,
the
data
storage.
Is
there
it's
the
storage.
All
of
this
stuff
is
recorded
in
git.
This
is
something
that
early
get
ups,
adopters
and
practitioners
have
then
hit
squarely
and
said:
okay.
Well,
it's
all
recorded
in
there.
How
on
earth
do
my
auditors
navigate
that?
And
this
is
an
opportunity.
B
This
is
an
opportunity
for
all
of
us
to
build
kind
of
a
certain
set
of
tooling
that
allows
that
serves
the
auditor.
Well,
so
that
allows
them
to,
for
example,
see
that.
Well,
if
something
went
wrong
here,
let's
trace
or
if
something
went
right,
let's
trace
what
were
all
of
the
different
you
know
commits
that
came
that
played
a
role
in
the
way
that
this
runtime
environment
was
operating.
In
the
end,
it
doesn't
come
from
it's
not
that
the
entire
thing
from
your
cluster
configuration
or
your
cluster
definition,
cluster
configuration
infrastructure
definition.
B
A
B
A
I
mean
steven
nemo
here
in
chat
just
mentioned,
like
socks,
separation,
responsibility
right
like
that
separation
can
still
exist,
but
it
can
converge
in
a
git
repo
into
a
pull
request.
Kind
of
thing
right
like
that's.
Definitely
the
christian
hernandez
is
in
chat
and
says
someone
needs
to
write
a
tool
that
turns
git
log
into
a
report
for
auditors.
That's
a
good!
That's
a
good.
C
There
are
like
other
sources
of
influence,
but,
like
I
like,
oh,
I
wouldn't
say
that
I
would
say
this
that
this
is
definitely
like,
provides
easier
access
to
audit
than
not
having
a
a
a
trail
of
changes
in
git
like
if
you
don't
have
that
you.
You
have
much
bigger
problem
right,
frank
right,
but
there
is
a
gap,
a
grid.
There
is
a
gap
that
so
there
like
git
is
a
great
place
because
multiple
repos
perhaps
is
distributed
you
you
also
have
we
talked
about
prometheus.
You
also
have
perhaps
alerts
in
in
alert
manager.
C
There
do
you
have
kubernetes
events
of
the
cluster
that
has
been
the
target
of
these
deployments
and
there
there
is
usually
a
git
ops
tool
also
in
place.
That
has
also
a
trail
of
changes
that
is
applied.
So
there
is
definitely
an
opportunity
there
for
like
new
new,
tooling,
really
to
appear
to
close
the
scale
of
correlation
of
data.
How
do
you
correlate
all
these
different
events
that
I
have
in
different
systems?
C
So
everything
is
documented,
but
how
do
I?
How
can
I
look
at
a
window
of
time
and
put
all
this
on
on
a
timeline
that
can
help
me
not
not
just
for
audit
purposes
to
be
honest,
but
even
for
like
rudka's
analysis
and
reducing
the
like
failure
time,
the
metrics
we
were
talking
about
it.
So
there
are.
The
current
situation
is
better
than
like
without
relying
on
gets,
but
there
are
improvements
in
my
opinion
that
that
can
be
made
still.
A
Yeah
absolutely
so,
I
wanted
to
talk
about
the
event
youtube
just
keynoted
for
a
little
bit
get
ops
con
was
day
zero
or
day
one
whatever.
We
call
it
now
event
at
kubecon
eu
a
couple
weeks
ago.
What
were
your
big
takeaways
from
that
event
right,
like
there's
a
lot
of
interest,
and
we
had
a
lot
of
people
speak
both
of
you
included.
Well,
did
you
pick
up
anything
along
the
way
during
that
conference
that
you'd
like
to
share
with
anyone
else.
C
From
my
perspective,
I
go
first
on
this
one
from
my
perspective,
like
the
very
fact
that
it
was
so
well
received,
even
though
there
are
like
this
year.
There
are
multiple
git
ops
events,
really
in
a
very
short
time
frame,
close
to
each
other
and
and
the
fact
that
it
was
so
well
received
such
a
high
level
of
participation
and
at
kubecon
of
of
any
event,
which
is
there
are
a
lot,
a
lot
of
different
events
that
are
really
competing
for
attention.
C
As
a
a
possible
workflow
that
could
help
them
address
these
challenges,
so
it
has
been.
It
was
great
to
see,
but
also
shows
like
how
much
more
aware
we
need
to
be
across
these
different
type
of
use
cases
and
how
much
more
conversations
like
we
need
to
have
with
this,
with
these
teams
organization
to
understand
how
their
cases
are
different,
how
their
requirements
or
challenges
might
be
different
and
other
other
flavors
of
this
process
or
other
changes
that
might
be
unique
to
a
particular
vertical,
for
example.
B
And
and
to
put
to
put
a
metric
against
some
of
the
things
that
you
just
said,
we
had
900
people
over
the
course
of
the
day
join,
get
ops
con.
It
was
amazing
phenomenal
when
we
first,
you
know,
put
put
the
event
on
and
it
was
an
event
that
was
co-sponsored
by
weaveworks
and
red
hat.
When
we
first
came
up
with
this
idea
of
hey,
do
we
do
a
zero
day
event
around
get
ops
and
all
of
that
stuff?
I
think
I
don't
know
about
you
chris,
but
I
was
thinking
man.
A
B
B
B
B
B
B
A
C
B
Your
listeners
that
didn't
have
a
chance
to
attend
there.
Of
course,
all
the
get
ups
con
videos
are
are
up
on
and-
and
chris
was
the
executive
producer
for
that.
So
thank
you,
chris,
for
providing
all
of
your
expertise
in
in
doing
that.
So
chris
and
his
team
have
made
all
of
those
things
available
on
youtube.
There
were
a
couple
of
talks
that
I
found
really
great,
I'm
always
really
drawn
to
the
end
user,
so
the
success
stories,
so
one
in
particular,
was
generally
switzerland
or
switzerland.
Generally.
B
I
think
it
was
generally
switzerland
and
there
was
another-
and
I
can't
remember
off
the
top-
and
I
had
the
name
of
the
company,
but
I
would
encourage
you.
It
was
a
single
track
conference,
so
there's
only
about
nine
nine
sessions
or
something
up
there,
so
you'll
be
able
to
easily
find
the
ones
that
are
are
connected
to
the
r,
the
practitioners,
the
people
who
are
putting
these
practices
in
place
and
who
are
inventing
those
along
with
all
of
us
in
the
industry,
because
it
is
in
the
early
days.
B
B
A
C
For
me
from
from
from
the
the
process
perspective
and
the
other
side
of
it
is
really
the
kubernetes
space
is
extremely
exciting.
Like
we,
I
care
passionately
about
developers
about
development
process.
We
talked
about
get
ops
and
a
lot
of
that
relies
on
that
declarative
ways
of
working
that
we
talked
about
and
kubernetes
as
an
ecosystem.
Is
it's
such
a
such
a
live
thing
and
the
direction
that
it
that
is
going?
C
But
as
we
grow
that
part
of
the
ecosystem,
I
I
get
like
really
excited
where
this
this
will
be
in
a
little
while
and
the
type
of
conversation
we
will
be
having
with
with
with
users
with
kubernetes,
like
teams
that
are
adopting
kubernetes
of
the
the
degrees
of
like
work,
ways
of
working,
the
git
ups
type
of
operation
that
can
apply
and
how
this
can
expand
to
things.
That
is,
is
not
really
imaginable
today.
B
And
I
promise
I'll
keep
it
short,
but
I
do
want
to
take
just
a
moment
before
I
answer
that
question
to
emphasize
something
that
chris
said,
we
do
have
an
event.
Weedworks
is
putting
on
get
up
stays
it's
a
two-day
event
june,
9th
and
10th.
My
colleague,
tamo
nakahara,
is
a
magician.
She
runs
our
dx
program.
She
is
a
magician
when
it
comes
to
putting
on
events
like
this.
We
will
have
a
live
dj
between
sessions.
It's
going
to
be
super
fun,
so
please
do
come.
B
B
B
This
really
impactful
statement
that
has
had
an
impact
on
your
life
and
other
people's
lives
in
the
years
since
here
it
is
three
years
later,
and
it's
still
such
a
re,
a
moment
of
remembrance
for
you
and
you
recognize
the
rippling
effects
that
that
moment
has
had
that's
what
gets
me
up,
because
I
believe
that
every
day
has
a
possibility
of
having
one
of
those
boulders
in
in
the
pond.
That
causes
these
huge
rippling
effects
and
that's
what
I
love
about
life
wow.
A
What
an
amazing
I'm
I'm
so
glad
that
you
were
able
to
say
that
an
amazing
amazing
finish
to
this
episode,
so
thank
you
so
much
cmx.
Thank
you!
So
much
cornelia.
I
appreciate
your
time.
I
appreciate
all
the
efforts
you
putting
into
the
community.
You
know
at
at
large
for
everyone
out
there
trying
to
make
their
lives
better
in
general.
Right,
like
devops,
is
essentially
making
lives
better,
in
my
opinion,
so
and
using
technology
to
do
it.
A
A
No
problem
at
all
coming
up
next
folks
is
the
one
and
only
sebby,
and
he
will
be
talking
about.
He
has
a
special
guest
talking
about
porcus
today,
so
tune
in
they're
gonna
do
some
stuff
with
doom
the
video
game,
so
that
should
be
fun
so
again.
Thank
you,
cornelia
and
cmac.
Thank
you
audience,
and
we
will
see
you
again
soon.