►
Description
OpenShift Commons Gathering @ Kubecon/NA San Diego November 18 2019
Portworx, StorageOS, Dynatrace, MongoDB
A
Great
thanks,
Rob,
alright
folks,
so
the
next
panel
is
going
to
be
our
operators
in
excellence
panel
and
so
we're
gonna
bring
them
on
board.
These
are
five
is
V
partners
that
have
developed
operators
and
not
just
any
operator,
but
operators
that
are
really
advanced
capability
operators,
and
not
only
that
they're
also
operators
certified.
So
what
we
wanted
to
do
is
I'm
going
to
call
each
of
our
panelists.
We
actually
are
awarding
for
the
first
time
ever
our
operator
excellence
awards.
A
A
A
C
So
I
just
wanted
to
underscore
the
importance
of
that
capability
model
and
building
the
maturity
of
the
operator
ecosystem
is
really
important
for
both
the
customers
of
these
companies
to
be
successful,
building
these
operators
and
using
them
on
their
clusters,
as
well
as
like
building
the
next
wave
of
kubernetes
growth.
You
know
we
kind
of
are
in
the
stateless
applications.
C
Everything
is
getting
a
little
bit
more
complicated
and
it
means
that
you
need
operational
expertise
and
running
that
or
you
defer
to
one
of
these
really
awesome
operators
to
do
that
for
you,
and
so
it's
a
game
changer
for
consuming
applications
on
kubernetes
and
consuming
a
new
product
on
kubernetes,
and
so
that's
super
super
important.
So
we're
really
happy
to
work
with
all
of
these
folks
and
their
teams
to
make
that
happen.
So,
let's
just
go
down
and
introduce
yourself,
give
us
an
idea
of
what
you
do
and
then
we'll
continue.
E
My
hey
hi,
my
name
is
Jason
mimic,
I'm,
a
product
manager
at
MongoDB
and
I
guess:
I've
been
working
with
the
Red
Hat
and
the
kubernetes
for
a
couple
years.
Now
we
didn't
have
an
operator
up
until
about
a
year
ago.
We
released
it,
but
before
that
it
would
just
had
like
a
openshift
template
and-
and
things
like
that,
so
it's
been
great
working
with
with
Rob
and
the
team
and
seeing
it
really
mature.
F
B
C
And
so
we
like
to
hear
all
right
first
question
huge:
this
is
to
you:
I
saw
it
when
I
was
doing
some
research
for
this.
That
port
works
has
this
config
generator
for
open
ship
for
to
help.
You
come
up
with
your
spec,
for
what
you
pass
into
the
operator.
Tell
us
about
that.
How
did
that
come
about,
and
how
do
your
customers
use
that
yeah.
F
Sure
so,
when
we
first
started
deploying
in
kubernetes,
which
was
more
than
two
years
ago,
we
had
we
used
to
deploy
it
as
a
daemon
set
like
what
work
is
to
get
install
as
demon
said,
and
then
there
were
other
slowly.
We
started
adding
components
like
config
Maps
different
permissions.
So
it's
a
huge
spec
and
people
don't
want
to
look
at
that
spec.
F
So
we
came
up
with
a
spec
generator
where
people
could
just
choose
some
parameters
like
select
their
storage
network,
and
this
just
by
clicking
on
some
checkboxes
drop
box,
fill
in
some
fields
and
they'll
get
a
huge
kubernetes
spec
and
they
just
have
to
apply
that
spec
and
install
port
works,
but
with
that
with
operators,
it's
just
a
20
line,
spec
that
they
have,
they
can
even
handwrite
it.
They
don't
have
to
use
the
spec
generator,
but
we
that's.
That
was
the
reason
why
we
added
the
spec
generator
awesome.
C
F
C
Perfect
all
right,
this
one
is
for
Jason
for
MongoDB.
Can
you
talk
about
how
your
operator
has
helped
customers
embrace
like
this
kind
of
container
native
Cloud
Nate
world?
You
mentioned
that
you
know
the
operators
been
in
existence
for
a
short
while
now
how
our
customers,
embracing
that,
how
is
it
helping
them?
Well,.
E
There's
certainly
a
ton
of
an
interest
I
mean
I'd,
say,
like
I
said
in
the
past
two
years
that
we've
really
been
active
in
this
space,
I
mean
there
hasn't
been
a
week
on
by
that
I
haven't,
got
calls
for
new
accounts
or
new
customers
that
are
interested
in
in
kubernetes
and
running
MongoDB
in
these
environments
and
frankly,
most
of
them
are
pretty
confused.
There's
a
relatively
small
subset
that
are
any
experience
with
kubernetes.
Most
of
our
customers
are
brand
new,
and
so
they
really
look
to
us
to
help.
E
You
know,
show
them
the
right
way
and
the
and
the
sort
of
certified
way
to
run
MongoDB
in
these
environments.
What
else
is
really
confusing
is
like
there's
tons
of
MongoDB
in
docker
in
containers
and
kubernetes
out
there
and
open
source,
but
like
none
of
that
came
from
MongoDB
until
recently
with
with
our
operator,
and
so
that's
typically,
you
know
more
of
our
enterprise
customers
that
really
need
that
support
and-
and
you
know
they
can
rely
on
on
a
company
to
back
it,
but
the
operators
just
made
it
a
lot
easier.
E
So
if
you
look
at,
if
you
go
to
like
find
just
your
random
MongoDB
deployment
out
there
kubernetes,
it's
gonna
be
a
lot
of
the
ammo
like
we
were
talking
about
now
with
the
MongoDB
operator.
It
reduces
down
to
as
little
as
nine
or
so
for
a
simple
model.
Db
cluster.
We
have
one
C
Rd,
so
that
makes
it
a
lot
simpler
and
a
lot
of
people
want
to
run
MongoDB
as
a
service.
This
is
the
tool
for
these
enterprises
to
actually
offer
their
own
database
services
inside
their
own
private
data.
Centers
yeah.
C
That's
something
that
I've
seen
is
really
popular.
Is
you
know,
especially
when
you
get
to
these
big
banks
and
insurance
companies
and
folks,
like
that
they
all
have
a
central
database
team?
That's
running,
you
know
pick
whatever
you
want
of
a
database
and
they've
got
a
bunch
of
DBAs
that
run
it
for
you,
but
they
can
offload
their
work
to
your
expertise
in
the
operator.
Hood.
E
Yeah
yeah
in
a
way,
but
like
also
that
one
other
point
I
wanted
to
make
was
like
our
operator,
is
kind
of
you.
Well,
maybe
not
unique,
but
has
basically
it
has
our
enterprise
management
monitoring
solution
called
mama-to-be,
ops
manager.
That
is
more
or
less
bundled
with
our
operator,
because
we
already
had
a
bunch
of
a
bunch
of
features
in
that
around
automation
and
we
could
basically
do
what
you
know
what
kubernetes
does
in
terms
of
managing
the
pods
just
with
VMs
in
our
own
agents.
E
So
we
took
a
lot
of
that
feature
and
functionality
and
we
kind
of
use
it
with
our
operator.
But
the
neat
thing
is
those
DBA
teams.
They
actually
do
get
this
like
completely
independent
control,
plane
where
they
could
just
look
at
the
data
drill
into
collections
and
things
like
that.
They
don't
even
know
that
it's
necessarily
running
an
open
shift
so
that
you
can
have
a
nice
separation
of
those
of
those
responsibilities.
Awesome.
C
Yeah,
that's
what
we
need
to
get
everybody.
That
sounds
perfect.
All
right.
Next
question
is
for
Simon
with
storage,
OS,
I'm
curious.
We
all
know.
Storage
is
really
critical
for
the
cluster
and
your
operator.
You
know
it's
a
big
part
of
running
a
healthy
storage
tier.
You
talk
about
some
of
the
features
that
the
operator
has
to.
You
know,
keep
your
storage
running
correctly
and
performant
on
an
open
ship
cluster.
B
Yeah
sure
and
I
mean
the
the
operator
SDK
itself,
apart
from
being
a
really
good
starting
point
with
the
scaffolding
when
you
create
your
first
operator
in.
It
includes
a
lot
of
tooling
around
testing
as
well,
so
not
just
unit
testing
but
being
able
to
describe
all
of
the
permutations
of
configurations
you're
able
to
through
the
operator
test
framework
test
all
those
combinations
as
well
as
that
it
helps
you
run
some
end-to-end
tests
so,
for
example,
deploying
on
various
versions
of
kubernetes
and
things
like
that.
B
We're
able
to
do
performance
regression
testing
we're
able
to
do
more
kay
we're
able
to
do
more
chaos
either
on
bit.
We
were
able
to
do
the
more
sort,
a
chaos,
monkey
style
testing
to
ensure
that
data
is
is
consistent.
So,
for
example,
John
Willis
earlier
was
talking
about
Merkle
trees.
We're
able
to
through
block
checksums
verify
the
integrity
of
replicas
before
we
failover
to
them.
Yeah
awesome.
C
Yeah
I
think
that's
the
exact
type
of
expertise
that
you
know.
We
want
to
depend
on
these
operators
for
instead
of
having
to
understand
that
about
a
storage
layer.
Even
so
that's
super
powerful
and
you
know
you
get
all
that
baked
in
and
release
after
release,
all
right
and
last
but
not
least,
for
our
first
question
Peter
how
you
doing
so.
Your
customers
appraise
your
operator
for
its
ease
of
use.
Can
you
just
talk
about
that
experience
and
how
you've
designed
it
and
what
you
were
looking
to
do?
Maybe
in
the
future
absolutely.
D
So
dynaTrace
comes
out
of
the
APM
space
and
the
monitoring
enterprise
monitoring
and
what
we've
developed
this
operator
together
with
Red
Hat
when
they
were
initially
bringing
the
operator
to
market
that
framework,
we
used
that
we
created
a
go
operator
and
the
focus
around
dynaTrace
is
being
able
to
do
it
like
make
it
as
easy
as
possible
for
our
customers
right.
We
don't
want
our
customers
to
have
to
really.
You
know,
have
super
expertise,
and
you
know
their
application
code
to
try
and
instrument
things
and
such
so.
D
We,
our
goal,
was
always
to
make
it
as
easy
to
deploy
the
agent
and
get
the
and
get
the
democratize
the
data
that
comes
out
of
it.
So
our
agent
based
solution
in
the
you
know
non-criminal
space.
It
would
be
an
install,
but
because
of
the
operator
instead
of
having
to
use
a
daemon
set
where
you
would
have
to
you
know,
take
it
down
to
you
know,
to
put
up
any
updates
or
put
up
a
new
one,
the
with
the
OLM
and
the
operator.
D
Basically,
through
just
yamo
files,
you
could
have
the
entire
operator
deployed
running
in
your
entire
cluster
and
the
one
agent
our
agent
technology
would
automatically
be
deployed
to
all
the
nodes
in
the
cluster,
providing
visibility
into
all
the
application
stacks.
So
what
this
really
drives
for
our
customers
is
the
providing
monitoring
is
part
of
the
platform,
but
deep
monitoring
and
and
full
stack
view
of
the
entire
cluster
yeah.
C
Obviously,
a
critical
capability
for
any
cluster
yeah,
so
we
can
turn
this
more
into
a
little
bit
of
free-for-all.
Now,
I
want
to
touch
on
two
topics
that
we
visited
earlier
in
the
day:
around
get
ops
and
then
Multi
cluster,
and
so,
let's
start
with
multi
cluster
I'm,
just
curious
any
of
you
with
your
customers.
Are
you
seeing
them
run
multiple
clusters
and
how
is
the
operator
helping
them
get
that
consistency
between
the
clusters,
like.
D
D
D
It's
there's
customers
that
have
these
large
legacy
data
centers
spread
around
their
organization,
but
then
they've,
you
know,
started
to
build
out
into
a
cloud
whether
it
be
AWS
or
Google
or
measure,
and
having
the
ability
through
the
operator
to
be
able
to
kind
of
see
all
of
the
clusters
and
all
the
nodes
for
those
clusters
really
gives
us
an
ability
to
help
monitor
all
of
that
across
the
board.
So
that's
that's
something
that
said
that
was
critical
for
our
customers,
as
well
as
for
for
having
the
operator
and
in
the
first
place
anyone.
E
I'll
just
say
that
honestly,
I
haven't
heard
much
about
Mustard
multi
cluster.
In
a
few
months.
A
while
ago,
I
had
a
there
was
a
lot
of
customers
interested
in
it,
and
so
we've
kind
of
been
following
the
cig,
multi
cost
cluster
and
seeing
where
that's
going,
but
I'm,
not
really
sure
I
think
the
the
verdict
is
still
out
on.
If
people
a
lot
of
people
are
going
to
be
using
those,
you
think.
E
F
Think
what
he
said,
that's
the
problem
we
are
trying
to
solve
with
storage,
so
we
already
support
a
lot
of
clouds
like
Google,
Amazon,
IBM
Azure,
and
with
this
with
kubernetes
and
the
operators.
It
is
a
seamless
experience
for
customers.
They
can
just
run
port
works
without
knowing
they
just
tell
us
what
capacity
they
need
internally.
We
go
and
provision
the
capacity
and
they
can
just
migrate.
Their
apps
using
port
works
from
one
cloud
to
another.
We
have
few
customers
who
are
running
on
multiple
clouds.
F
C
F
B
C
C
E
That,
like
we
also
so
we
have
MongoDB
as
a
service.
I
want
to
be
Atlas,
which
is
fully
hosted
by
us,
but
that
is
actually
available.
We
have
an
open
service
broker
for
MongoDB
Atlas
that
just
came
out
this
past
summer.
So
in
a
way
because
Atlas
supports
Google,
GCP
and
annasher,
you
could
have
multiple
kubernetes
clusters
that
just
consume
services
running
in
those
same
clouds
so
functionally
there's
a
lot
of
ways.
I
think
you
can
get
around
some
of
those
requirements
that
might
not
need.
C
Point
moving
on
to
get
up
so
we
saw
a
example
of
using
both
MCM
and
agro
based
workflows
for
doing
this.
One
of
the
things
I
think
is
really
powerful
about
operators
is
when
you
do
have
these
complex,
distributed
applications
or
storage
layer
that
are
made
up
of
a
ton
of
different
gamal
objects.
When
you're
doing
like
a
PR
review
on
a
get
ops
change,
you
can
start
looking
if
it's
an
operator
really
you're
looking
at
one
yellow
file
for
this
CRD.
C
That
might
represent
a
ton
of
changes
under
the
hood,
but
you
don't
need
to
see
that
complexity
and
so
I
think
that's
a
really
powerful
way
to
do.
Git,
ops
and
I'm
curious.
If
y'all
are
starting
to
see
this
with
your
customers,
are
they
doing
these
types
of
flows,
or
is
that
still
more,
it's
the
audience
at
coop
con,
but
it
hasn't
made
its
way
down
into
the
fortune
500.
That
kind
of
thing.
F
E
I,
don't
think
it's
quite
come
down
to
the
the
common
common
folks
right
now,
but
just
internally
at
MongoDB
we
have
on
our
own
in
Turku
Nettie's
cluster,
that
has
a
CI
CD.
Very
similar
I
mean
it's
drone,
but
it's
the
same
kind
of
thing
and
I
can
really
see
that
model
is,
is
really
solid,
I
think
I
mean
because
we
write
code.
We
like
source
control.
Why
not
put
all
that
stuff
in
source
control?
It's
great.
D
How
do
they
validate
and
quality
test
things
like
that,
and
so
I
mean
I'm
wearing
this
shirt
capped,
and
this
is
an
open
source
project
that
we
started
at
iron,
trace
to
try
and
address
this
specific
kind
of
issue
around
that
get
up
story
about
the
git
flow
and
the
model
and
and
ideally
I.
You
know
I'm
working
with
the
team
here
to
try
and
build
operators
to
do
the
deployment
of
captain
as
well.
So
that
should
be.
C
It's
not
going
anywhere,
it
started
with
terraform
and
the
like
on
the
infrastructure
side
and
it's
moving
its
way
up
all
right,
I
think
that's
all
we
have
for
this
panel
today.
I
want
to
give
a
warm
congratulations
to
all
of
these
companies
check
out
their
operators
they're
on
operator
hub
they're
inside.
If
we're
open
ship
cluster,
if
you
need
any
of
these
services,
you
know:
storage
databases,
monitoring
security
tools,
queues
web
servers,
it's
all
out
there.
So
thank
you
all
so
much
and
congratulations.