►
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
Intro: KubeVirt BoF - Daniel Gonzalez Nothnagel, SAP & Scott Collier, Red Hat
KubeVirt extends Kubernetes with the ability to run virtual machine workloads, as Kubernetes itself is not able to launch those. To do so KubeVirt introduces dedicated custom resources around virtualization to have enough flexibility to build and tune VMs towards existing and new use-cases. While Kubernetes is a great platform for cloud-native workloads there are also many workloads that require more isolation than containers can provide, or are just in general better served with VMs than with containers. KubeVirt aims at a seamless integration with Kubernetes in order to create one consistent cloud-native API for container and virtual workloads, sharing the same storage, network, and compute resources. In this BoF we will be discussing the basics and different use-cases of KubeVirt and adjacent projects. As an example we will explore how KubeVirt VMs can be used as nodes for a Kubernetes cluster, effectively allowing to run Kubernetes on Kubernetes.
To Learn More: https://sched.co/Ixrt
A
So
welcome
to
the
Qbert
intro.
This
is
a
bird
of
a
feather
session.
My
name
is
Scott
Collier
and
I
work
for
Red,
Hat
and
I
focus
on
the
upstream
Qbert
community.
However,
I
do
have
some
history
with
Seattle.
I
did
live
here
about
for
about
five
to
six
years.
Twenty
years
ago
met
my
wife
here.
My
daughter
was
born
here
and
I
was
actually
stationed,
30
minutes
south
of
here
at
Fort
Lewis
in
Washington.
B
A
A
Well,
okay.
So
this
is
an
intro,
hopefully
by
the
end
of
this,
which
will
walk
away
with
is
a
better
sense
of
how
to
get
started
and
take
it
back
and
kick
the
tires
on.
It
do
a
proof
of
a
proof
of
concept
on
it
now
another
hand-raising,
how
many
people
have
existing
virtualization
like
infrastructures,
on-premise?
A
That's
what
I
thought
this
is
the
perfect
session
for
you,
okay,
so
what
we've
got
here
is
destroy
so
far.
We're
really
looking
at
like
the
intersection
of
managing
virtual
machines,
as
well
as
application
containers
in
the
same
umbrella
right.
So
we
want
to
make
sure
that
we
have
a
platform
that
provides
a
consistent
experience,
no
matter
where
the
workload
resides
with
application
containers.
We
can
leverage.
You
know
the
past
five
or
so
years.
A
You
know
of
knowledge
that
we've
gained
and
with
virtualization
we've
got
like
a
history
that
spans
back
even
further
to
the
past.
You
know
10
to
15
years,
perhaps
for
some
of
us
with
libvirt
and
what
we've
got
with
what
kubernetes
is
we
basically
need
a
couple
of
you
know:
infrastructure
providers
to
provide
us
virtual
machines.
A
We've
got
OpenStack
AWS
over
we've
got
our
public
cloud
providers
like
AWS
DCP,
even
as
your
we
provision
kubernetes
on
those,
and
it
gives
us
this
kind
of
consistent
platform
and
and
kind
of
satisfies
that
abstraction
for
the
open,
hybrid
cloud
that
we
need.
We
can
grab
rural
machines
and
provision
cube
on
top
of
that.
So
what
we
have
now
is
we
have
our
application
containers
that
we
need
to
manage
with
with
kubernetes,
and
then
we
have
some
use
cases
where
we
just
need
like
stricter
isolation
for
some
of
these
virtual
machines.
A
So,
for
example,
on
the
slide,
you
see
some
examples
like
cata
containers,
orgy
visor
from
Google
or
you
know,
isolates
from
CloudFlare
and
then
we've
got
firecracker.
That
was
just
announced
by
AWS
a
couple
of
weeks
ago
out
reinvent
now,
these
provide
like
stricter
isolation,
kind
of
runtimes
for
for
your
applications
there
it's
a
little
bit
different
of
a
problem
space
than
what
we're
looking
at.
A
What
we're
looking
at
is
Qbert
for
virtual
life,
basically
managing
and
virtualizing
existing
workloads,
so
for
the
container
rising
infest
that
we
have,
you
know,
we've
got
application,
containers
that
are
coupled
with
kubernetes
orchestration,
that's
basically
becoming
the
new
standard
for
our
applications.
Right
and
we've
got
some
work
loads
that
you
know
we
can
rip
tin
ship
straight
into
containers,
but
it's
isn't
always
the
case
and
there's
a
few
reasons
for
that.
A
We
have
technical
reasons
and
we
have
business
reasons
right
so
for
the
business
reason
we
have
like
the
cost
of
basically
D
decomposing,
these
in
virtual
machines
into
micro
services
and
managing
those
with
kubernetes,
and
we
also
have
time
to
market
like
that,
takes
a
long
time
to
do
that.
Some
of
these
applications
weren't
designed
to
run
this
way
and
some
of
the
technical
reasons
may
be.
A
Some
customers
have
the
need
to
you,
know:
access
certain
types
of
hardware
like
GPUs
or
SR
Iove
devices.
That
may
be
another
reason
to
to
go
at
a
bare
metal
and
then,
as
the
mix
changes-
and
we
have
you
know,
customers
who
have
a
need
to
deploy
their
new
applications
on
top
of
as
kind
of
containerized
micro
services
they've
got
their.
You
know
legacy
existing
virtual
machines
that
are
running
in
their
environment.
We
need
to
basically
provide
a
platform
for
them
to
have
that
consistent
management
experience.
A
A
We
can
leverage
the
knowledge
that
we
built
out
with
KBM
over
the
years
and
take
that
user
space.
That
represents
a
virtual
machine
and
basically
wrap
that
into
a
container
and
manage
it
with
kubernetes
right.
So
one
of
the
use
cases
is
that
a
developer
may
may
have
is
they've
got
a
unified.
You
know
they
need
a
unified
platform
to
build
and
deploy
applications
on,
regardless
of
whether
or
not
they
reside.
A
In
applications
or
or
containers
or
virtual
machines-
and
then
you
know,
Qbert
also
represents
basically
the
Kuban
eddie's
community's
commitment
to
to
the
future
of
application,
orchestration
and
the
in
the
common
fabric
across
the
open
hybrid
cloud.
So
it
enables
us
to
combine
those
existing
virtualization
workloads
onto
the
same
platform
if
they're
running
in
virtual
machines
once
you
get
them
on
the
same
platform,
you're
at
least
buying
yourself
some
time
so
once
you're
managing
you
know
both
application
containers
and
virtual
machines
under
that
same
kind
of
umbrella.
A
You
can
at
that
point
start
looking
at
how
best
to
decompose
your
application
out
of
that
virtual
machine,
manage
it
with
kubernetes
and
have
that
same
existing
kind
of
experience
from
an
Operations
perspective
for
sure.
So,
for
a
few
more
details,
notice
on
the
top
we've
got
Qbert
dot
IO
listed.
That
website
has
documentation
for
the
project.
It
has
examples
to
get
you
started
and
it
has
lots
of
blog
post
that
the
engineers
and
contributors
in
this
room
have
been
have
been
adding
over
the
past
year.
So
please
do
go
check
that
out.
A
So
Qbert
is
basically
it's
a
drop
in.
On
top
of
your
existing
kubernetes
cluster
right,
it's
it's
an
extension
to
the
to
your
cube
cluster
by
CRD,
the
custom
resource
definition
and
CR
DS.
Basically,
in
kubernetes,
allow
you
to
extend
the
cuban
eddie's
API
and
a
pluggable
way
to
add
new
functionality,
and
that's
what
Cuba
brings
to
your
cluster?
Okay,
it
leverages
a
native
approach,
meaning
that
once
the
API
is
extended,
you
can
use
the
same
tools
that
you're
familiar
with
to
manage
your
other
cube
resources
and
objects
the
same
way.
A
A
A
And
OpenShift
UI
I'll
talk
about
that
a
bit
more
once
once
Daniel
gets
past
his
use
case,
but
we
basically
have
a
UI
with
some
extensions
that
allow
you
to
manage
and
monitor
your
virtual
machines.
When
they've
been
deployed
on
an
open
ship
cluster
we've
got
verts
ETL.
This
is
a
tool,
that's
basically
an
add-on
to
the
infrastructure
that
allows
you
to
do
things
like
start
VMs,
stop
beams,
connect
to
the
VMs
by
serial
console
or
VNC,
perhaps
so,
just
kind
of
in
a
nutshell.
A
B
So
understand
our
use
case.
I
have
to
set
the
context
here
this
so
I
guess
most
of
you
have
heard
of
a
city.
We
are
quiet,
large
software
company
focused
on
the
enterprise
market
and
actually
a
lot
of
stuff,
the
company
in
Europe
and,
of
course,
and
basically
all
the
other
software
companies.
We
are
currently
heavily
looking
into
kubernetes
and
building
it
into
multiple,
any
product
and
so
on,
and
because
of
that,
we
have
a
very
large
developer.
B
Community
is
currently
looking
to
get
kubernetes
cluster,
and
because
of
that,
we
have
to
need
to
actually
manage
this
large
fleet
of
cluster,
and
but
we
are
not
only
having
development
cases.
Here
is
also
more
and
more
production
use
cases
coming
on
yeah
and
so
to
do
that
if
50
starter,
then
our
own
open
source
projects
to
do
it
and
called
product
Gardner
which
we
use
to
manage
to
create.
You
need
upgrades,
gags,
laughter
and
I
won't
talk
about
protocol.
B
B
But
what
we
see
is
that
OpenStack
is
actually
a
quite
big
and
complex
technology
stack,
but
we
show
you
on
the
next
one.
What
I
mean
by
that?
Exactly
not!
The
thing
is
that
it
was
pretty
hard
to
maintain,
as
I
mentioned
earlier.
We
need
some
stable
infrastructure
and
how
to
maintain
and
save
it
infrastructure.
B
And
the
next
thing
is
that
if
you're
talking
about
education
formants,
so
we
have
customers
who
want
to
do
like
small
deployments
and
if
you
have
this
big
and
complex
OpenStack
control
plane.
That
does
not
really
work.
In
that
context,
it
yeah
just
a
few
servers.
Difficulty
put
up
and
taken
down
yeah.
So
let's
have
a
look
at
what
I
mean
by
the
complex
bank.
So
you
don't
need
to
understand
that.
B
But
if
you
move
across
like
architecture,
that
pictures
would
be
like
first
thing
that
you
find
on
Google
I
even
cut
out
some
of
the
stuff
like
horizon
the
API
and
so
on.
But
what
you
see
here
are
basically
two
core
component
of
kubernetes
with
all
their
API
services
agents,
plugins
databases
aim
to
teach
use
and
so
on,
and
basically,
if
any
single
one
of
these
components
here
doesn't
function
correctly,
you
get
stability
issues
on
top
of
it.
Well,
it's
just
an
example:
the
the
message
queue
for
Neutron
completed
without
tenure.
B
You
get
problems
in
your
network
funding
and
then
you
have
Gardiner
trying
to
provision
clusters
or
object
classes,
I,
don't
know
basically
creating
nodes
and
the
creation
of
the
node
itself
works
on
OpenStack,
but
the
network
funding
doesn't
work,
and
so
your
clusters-
or
you
know
it's-
never
really
come
online
and
because
everything
is
automated
Garnica,
when
is
we
try
and
try
again
to
create
new
node
and
some
but
nothing's
coming
online?
And
there
forget
yeah,
high
stability
issues
here,
so
we
we
are
thinking
about.
B
They
need
to
dynamic
API
to
create
cluster.
To
operate
them
and
so
on
and
for
static
deployment
or
sort
of
question
here
and
the
next
idea
that
we
had
was
well.
We
like
the
idea
of
having
kubernetes
at
the
infrastructure
level,
why
not
provision
binaries
class
up
on
top
of
it,
and
but
first
we
are
having
some
problems
there,
because
on
the
one
hand,
the
container
isolation
not
export
that
you
could
use
it
to
do
real
multi-channel
deployment
from
there
and
the
other
thing
is.
B
It
is
actually
quite
hard
to
run
the
queue
blood
inside
containers,
but
yeah.
Then
we
came
across
prepared
and
we
had
the
idea.
Why
not
use
cubed
VMs
to
provision
our
nodes
and
then
use
gardener
to
to
use
this
cubed
VMs
to
create
clusters
on
top
and
so
what
this
does
for
us?
It
simplifies
this
back.
You
don't
need
all
just
OpenStack
control
plane
here.
You
just
have
to
kubernetes
control
plan,
which
is
much
simpler
and
OpenStack.
B
Also
the
amount
of
operational
knowledge
that
you
need
the
super
reduced,
because
you
don't
need
any
knowledge
about
the
OpenStack
components.
You
basically
have
kubernetes
all
the
way
down
and,
lastly,
doing
like
edge
computing
cases
with
this
data
is
much
easier
because
your
control
planes,
just
so
lightweight
so
in
our
lab.
We
actually
have
single
no
deployments
with
clusters
on
them.
I
mean,
of
course,
that's
just
web
deployment.
You
won't
get
any
high
availability
out
of
them,
but
it's
possible.
If
you
have
edge
case
deployments
with
like
three
to
five
servers.
B
B
A
Thanks,
ok,
so
just
quickly
on
the
Red
Hat
use
case,
we've
got
Red
Hat
does
have
a
product
called
CNB,
which
is
container
native
virtualization,
which
runs
on
top
of
openshift,
which
is
our
basically
our
enterprise
release
of
kubernetes,
and
it's
got
kind
of
a
developer
friendly
spin
to
it
with
the
web
console
that
you
can
use
to
manage
your
virtual
machines,
as
well
as
the
storage
integration
being
backed
by
bluster
and
and
some
of
the
other
capabilities
there.
So
I
think
this
is
kind
of
what
a
lot
of
you
are
waiting
for.
A
A
Okay,
you
can
see
it.
I
can
see
it
that
way.
I
don't
have
to
turn
my
head
too
much
all
right.
So
this
is
a
demo
that
all
of
you
can
do
it's
a
very
low
barrier
to
entry.
Just
need
your
laptop
I'm,
just
using
mini
cube
to
do
this
and
we'll
talk
about
some
other
ways
that
you
can
get
your
hands
on
it
in
a
bit.
So
the
first
thing
that
I
want
to
do
I'm,
gonna
kind
of
like
copy
and
paste
over
here,
so
don't
mess
up
anything
typing
it.
A
Just
gonna
set
my
version
to
eleven
0.11,
which
is
the
latest
one
and
then
just
kind
of
take
a
look
at
my
cube
environment.
So
keep-keep
cube,
cuddle,
get
nodes
shows
me
that
I've
got
a
working
kind
of
Cuban
Eddie's
cluster
I
can
get
my
CR
DS,
which
don't
exist
yet,
and
I
can
also
get
all
of
my
pods.
Oh.
A
Okay,
so
we
can
see
here
that
this
is
a
basic
cube,
install
and
now
what
I'm
going
to
do
is
just
provision
cube
Bert
on
top
of
this.
So
this
everybody
in
here
you
take
your
existing
kubernetes
cluster.
You
can
do
the
same
things
and
all
I'm
gonna
do
is
apply.
This
cube,
Bert
yamo
file
and
it's
gonna
install
it
just
in
a
couple
of
seconds
right.
So
that's
done.
It's
installed
now
I'm,
just
gonna
check
again
for
the
CR
DS
that
got
added.
You
can
see
here
that
we've
got
our
virtual
machine
instance.
A
A
And
you
can
see
here
that
they
are
in
the
process
of
being
provisioned.
Now
I
am
on
the
event
Wi-Fi
network,
so
I'm
not
exactly
sure
how
long
it
will
take,
but
it
shouldn't
take
too
long.
Let's
just
watch
that
for
a
second
and
in
the
meantime,
all
the
containers
are
creating.
Let's
take
a
look
here.
Can
everybody
see
that?
Okay,
okay,
let's
just
take
a
look
at
that
Qbert
yeah
Mel
file
that
got
created,
so
we
can
take
a
look
at
different
custom
resource
definitions
in
here
that
got
created.
A
A
The
machine
wrecked
instance
replicas
sets
and
then
preset
and
then
finally,
the
virtual
machine
instances
and
the
migrations
so
that
got
created
and
let's
take
a
look
over
here,
looks
like
our
containers
are
starting.
Now,
that's
good.
Let's
just
get
our
namespaces
our
pods
one
more
time,
everything's
up
and
running
so
now
Qbert
is
provisioned.
We've
got
our
CR
DS,
we've
got
our
pods
I'm.
Just
gonna
make
this
a
little
bit
smaller.
A
Bring
this
to
the
top
of
the
screen,
we're
just
gonna
grab
this
version
of
the
based
off
of
the
off
of
the
github
repo,
and
you
can
see
that
the
test
VM
was
created.
So
now,
we've
got
this
new
object,
that's
being
presented
you
in
kubernetes
right!
So
let's
do
a
cube.
Cto
get
VMs!
This
wasn't
there
before
now,
we've
got
a
test
VM
that
was
created.
The
VM
hasn't
started
yet,
but
it's
there.
So
let's
go
ahead
and
now
take
a
look
at
the
VM
mo
file
that
we
actually
imported.
A
So
you
can
see
here
that
this
is
based
off
of
the
Qbert
API.
That
was
extended,
it's
a
kind
virtual
machine
and
it
was
named
test
VM
so
again
like
when
I
was
talking
earlier.
These
are
like
tools
that
you're
all
familiar
with
the
same
kind
of
gamo
that
you're
used
to
configuring
and
creating
to
deploy
virtual
machines.
A
One
thing
here
is
that
the
spec
is
set
to
running
false,
which
means
it
doesn't
start
when
the
object
gets
created,
and
you
can
also
see
that
we've
got
like
the
ability
to
configure
different
attributes
for
the
virtual
machine.
So
this
is
kind
of
a
trivial
example
that
we're
using
on
a
local
laptop.
But
again,
if
you
go
to
Q,
Bert
dot,
IO
take
a
look
at
the
documentation.
You
can
see
some
of
the
other
attributes
that
you
can
add,
but
here
we've
got
a
container
disc.
We've
got
network
interfaces.
A
We've
got
the
amount
of
memory
that
this
VM
is
going
to
consume.
We
can
take
a
look
at
the
networks
that
it's
going
to
connect
to
and
by
default
it's
the
pod
network
and
we'll
be
able
to
demonstrate
that
it
actually
got
an
IP
address
off
of
that
once
we
boot
up-
and
this
is
the
image
that's
actually
getting
booted
here-.
A
And
it
says
that
the
test
VM
was
created
now.
I
can
also
go
back
and
take
a
look
at
I
can
describe
that
VM
and
make
sure
that
the
state
changed
from
running
false
to
running
true
get
the
m's
I
can
then
describe
VM
test
VM,
and
if
we
scroll
up
here
just
a
little
bit,
we
can
actually
see
that
that
state
did
indeed
pivot
to
true.
A
A
A
And
I
can
see
that
it's
pulling
the
image
right
now,
so
it's
been
scheduled
to
anode,
in
this
case
the
only
node
in
the
cluster
and
it's
pulling
down
the
image.
So
what
I'm
waiting
for
is,
if
I
well
I
can
try
to
connect
to
it
now,
but
I
yeah
I'll
go
ahead
and
try
so
like
vert,
CTO
console
test
VM
and
you
can
see
here.
A
A
It
shouldn't
take
too
long.
It's
just
basically
a
matter
of
it,
pulling
it
down
over
the
event
Wi-Fi.
Sometimes
it
takes
a
minute.
Sometimes
it
takes
a
little
longer,
but
while
we're
while
we're
waiting
here,
do
we
have
any
questions,
while
this
is
kind
of
booting
up
from
anyone
in
the
audience
lots
of
hands?
You,
sir.
A
C
Yeah,
so
we
basically
have
a
couple
of
different
ways
that
you
can
get
an
image
from
where
you
have
it
to
your
virtual
machine
for
this
demo
we're
using
a
container
disk
which
is
an
ephemeral
image.
That's
pulled
from
a
registry
and
it
starts
as
part
of
the
vert
launch
or
pod
and
then
becomes
available
to
the
virtual
machine
to
use.
So
that
would
be
an
ephemeral
just
like
a
normal
container
image.
C
But
if
you
want
something,
that's
persistent,
you
can
actually
have
a
persistent
volume
claim,
that's
associated
with
your
virtual
machine
and
then
Scott
mentioned
CDI,
which
is
a
way
that
actually
will
pull
a
queue,
how
to
image
or
a
raw
image
from
wherever
you
have
it
on
a
web,
server
or
other
location
into
the
PVC,
so
that
you
can
start,
and
in
that
particular
mode.
It
would
start
over
and
over
persistently,
like
you
would
expect
thanks.
A
C
So
we
started
off
using
just
the
funny
oh
yeah,
so
the
question
is:
is
our
using
block
mode
or
file
mode
TVs?
So
when
we
first
started
the
design,
the
only
thing
that
was
really
available
was
file
mode.
So
that's
how
it
works.
Today
we
actually
just
store
the
disk
image
as
a
single
file
on
the
persistent
volume
on
the
file
system,
but
we're
also
working
to
add
raw
block
volume.
Support
as
well
so
I
think
it's
actually
working
in
Qbert
today.
A
Just
just
really
quickly,
the
demo
has
about
2
more
minutes,
so
let
me
wrap
that
up
and
then
we'll
transition
over
into
QA,
so
the
virtual
machine
has
been
launched
now
so
now
we
can
actually
connect
to
the
console-
and
this
is
the
test
VM
login,
so
we've
connected
to
it
login
with
Cirrus
and
go
Cubs
go
and
now
we're
we're
logged
in
here.
So
now,
I
can
take
a
look
at
the
IP
address.
A
A
Then
we
can
also
use
cube
cuddle
to
delete
the
VM,
and
if
we
want
to
clean
up
our
environment
completely,
we
can
just
do
a
cube,
cuddle,
delete,
f,
Qbert
and
that
wipes
our
environment
clean
so
that
we
can
start
over.
So
that
concludes
the
demo
that
worked
now.
Let's,
let
me
just
transition
really
quickly
back
over
into
a
present
mode.
I've
got
like
two
more
slides:
they
don't
want
to
cover
with
everybody,
and
then
we
will
OH
does
that
on
keep
a
revert,
keep
okay!
Thank
you.
Okay,.
A
Just
a
quick
update
on
the
events
that
we
have
in
two
hours.
We
have
a
deep
dive
on
this,
where
Adam
and
David
are
going
to
talk
a
little
bit
more
about
architecture.
So
if
you're
interested
in
learning
a
little
bit
more
about
it,
please
do
show
up
for
that.
We've
got
a
demo
at
the
Red
Hat
booth
tomorrow
at
8:30
a.m.
if
you
want
to
take
a
look
at
that
I
believe
it's
a
virtual
machine
migration
from
one
node
to
another,
and
then
we've
got
a
Qbert
gathering
this
evening
and
that
location
change.
A
Let
me
just
pull
that
up
really
quickly:
a
spin
room,
okay,
very
good,
so
that's
also
gonna
be
posted
on
the
blog
post,
a
spin
room,
second
floor,
the
Sheraton
and
then
the
blog
post
is
listed
here.
If
anyone
wants
to
take
a
look
at
that
and
get
a
get
an
idea
of
all
the
events
so
now
getting
started
back
on
the
website
on
the
homepage
of
the
website,
you've
got
like
four
buttons.
You
can
get
started
on
AWS
or
GCP.
You
provide
your
credentials.
We
actually
roll
Cubert,
bm's
and
publish
them
on
every
update.
A
So
you'll
be
using
the
latest
code
launch
those
into
your
Amazon
or
DCP
account,
and
you
can
run
through
the
same
demo.
You
can
also
deploy
it
on
mini
cube,
which
is
listed
there,
as
well
as
on
bare
metal
clusters.
Fabien
Deutsch
who's.
Not
here,
has
also
created
a
cat,
a
code,
a
demo
which
is
zero
barrier
to
entry.
A
The
link
is
on
the
slides
and
you
can
basically
do
the
same
demo
that
we
just
went
through
here
on
your
browser
and
then
finally,
the
Qbert
demo
that
gets
updated
on
on
every
release
is
also
on
the
Qbert
github
page
and
then
for
collaborating.
We
do
have
the
website
the
github
page.
You
can
open
issues
against
Cubert
there.
If
you
start
testing
it
out
and
have
any
problems.
We
have
a
mailing
list:
Qbert
dev,
it's
a
google
group,
mailing
list
IRC
as
well
as
slack.
A
A
The
question
was:
if
you're,
using
a
cue
count
to
image,
can
you
pass
in
cloud
in
it
data
and
if
you
are
doing
that,
do
you
configure
it
in
your
yellow
file?
The
answer
is
yes
to
both.
You
can
plus
passing
cloud
in
it
data
and
you
do
configure
that
in
the
yellow
file
that
was
actually
shown
in
the
example
config
that
I
had
we
passed
in
a
password
that
way.
Yeah
next
question
go
ahead.
A
A
So
Steve
said
basically
like
we're
pivoting
we're
pivoting
the
operators
which
basically
provide
a
way
to
extend
kubernetes
and
provide
lifecycle
management.
So
it
will
be
once
once
that
pivot
is
made
you,
sir,
the
question
was:
does
it
support
multiple
networking
interfaces?
The
answer
is
yes,
it
does
support
the
Malta's
plug-in
and
you
can
take
a
look
at
Cubert
dot,
IO
in
the
documentation
and
there's
an
example
on
how
to
set
that
up
as
well.
Yes,
sir.
C
A
C
So
what
we
actually
use
for
that
mode
is
it's
a
separate
container.
The
question
is:
is
how
does
the
container
disc
mode
that
we
we've
seen
here
work?
Is
it
just
dumping
the
filesystem
from
the
from
the
container
image
into
the
VM?
The
way
that
this
works
is
we
actually
have
a
separate
container
that
runs
in
the
pod
that
specifies
that
container
image
and
then
we're
able
to
point
the
the
pod
that's
actually
or
the
container?
That's
actually
running
the
virtual
machine
over
to
that
filesystem
as
its
expanded.