►
Description
This week Jonny and Andrew are joined by the Kasten team to discuss data protection and recovery, including with VMs deployed using OpenShift Virtualization.
If you have persistent applications that need protection, this is a great stream join and learn more about one of our great partners!
Want to learn more?
Try Red Hat OpenShift: https://red.ht/3uVxkya
Follow us:
Andrew Sullivan https://twitter.com/practicalandrew
Jonathan Rickard https://twitter.com/jrocktx1
OpenShift https://twitter.com/openshift
Subscribe to Red Hat's YouTube channel: https://www.youtube.com/redhat/?sub_c...
About Ask an OpenShift Admin:
Do you have questions? We want to answer them! Every Wednesday at 11am ET, hosts Andrew Sullivan and Jonathan Rickard sit down to answer your questions about Red Hat OpenShift.
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 on redhat.com https://red.ht/3rYz5ZS
OpenShift on GitHub https://github.com/openshift/
#RedHat #OpenShift #Kubernetes #ContainerPlatform
C
Hello
and
welcome
to
this
week's
ask
an
openshift
admin
office
hours
live
stream.
I
am
Andrew,
Sullivan
joined,
as
always
by
my
lovely
and
amazing
co-host,
Mr
Johnny
Rickard.
How
are
you
today,
sir
dude
I'm
I'm
hanging
in
there
man?
How
are
you
I'm
good,
it's
good
to
be
back
in
the
saddle.
I
feel
like
we've
said
that
a
lot
you.
C
C
Yeah
I
missed
interacting
with
everybody,
I
mean
you
know.
Normally,
when
we
do
those
things
we
have
chats
and
you
know
try
and
answer
questions
and
all
that,
but
C'est
La
Vie
right
indeed
so.
So
today's
topic
is
protecting
apps
with
castin
K10
by
veeam,
and
this
is
something
that
I've
been
looking
forward
to
for
a
little
while
you
know
Mark
and
I.
C
We
chatted
gosh
like
six
or
eight
weeks
ago,
and
it's
just
it's
been
a
topic
that
has
come
up
regularly,
both
here
with
our
audience
as
well
as
kind
of
a
cross,
Red
Hats
and
all
the
various
teams
that
we've
talked
to
and,
in
particular
I
think
what
really
connected
us
together,
Mark,
despite
the
fact
that
you
and
I
have
done
other
things
together
was
the
openshift
virtualization
team.
They're,
like
oh
it'd,
be
great.
C
D
D
Of
course,
I
help
with
K10
our
commercial
product,
but
I'm
responsible
for
helping
marshalling
our
upstream
and
downstream
contributions
to
cube
vert,
Red,
Hat
engineering,
our
own
open
source
projects,
which
we
may
talk
about
a
little
bit
later
on
and
and
certainly
with
K10
product
management
as
well
I'm
joined
by
my
two
colleagues
and
I'll.
Let
Alex
and
Michael
introduce
themselves
now.
E
I
will
go
first,
hi.
Everyone
I'm
really
really
happy
to
be
there.
So
my
name
is
Alex
I'm,
I've,
drunk
cast
and
roughly
one
year
ago,
a
bit
more
than
one
year
ago,
and
today,
I
go
after
the
pre-sales
and
the
solution
architect
team
for
the
casting
business
unit.
So
I
had
the
pleasure
to
work
with
Michael
as
a
solution
architect
so
Michael.
If
you
want
to
introduce
yourself.
B
Sure,
thank
you
Alex.
My
name
is
Michael
Michael,
Kors,
CM
solution,
architect
at
Gaston,
and
to
tell
you
my
story
before
in
the
past,
something
like
three
years
ago.
I
was
an
open
shift
admin.
Can
you
believe
that
and
I
was
working
in
a
company
called
soccer
Australia
and
we
were
walking
since
openshift
free
free
seats,
I
believe
then
we
moved
to
free.
B
We
had
to
jump
to
openshift
four,
and
the
question
is:
how
are
we
going
to
do
that?
We
were
only
something
like
a
lot
of
cluster
with
a
lot
of
namespaces
and
there
was
no
other
solution
than
migration,
so
we
looked
for
a
tool
for
migration
and
that's
how
I
met
Gaston
and
after
that,
I
I
moved
to
some
other
stuff
and
one
day,
castan
called
me
and
asked
me
to
join.
B
C
C
So
and
Steve,
thank
you
for
your.
Your
good
wishes
there
I'm
much
I'm
very
happy
that
I'm
feeling
better
too
and
a
cadellis
would
be
great
if
coover
could
have
some
kind
of
BMC
like
BMC,
Associated
I,
don't
know
if
you're
a
red,
Hatter,
there's
some
home
lab
folks
who
do
crazy
things.
I,
don't
know
if
that's
one
of
them,
but
it
might
be
worth
asking
over
in
that
in
that
Community.
Well,
the
programs
might
also
be
paying
attention,
I'll
be
sure
to
break
it
up.
C
So
you
really
appreciate
you
know
introducing
yourselves.
It's
always
fascinating
to
me
all
the
different
backgrounds
and
that's
right
as
how
we
get
our
technical
expertise,
how
we
end
up
at
the
companies
we're
at
and
all
that
other
stuff,
and
you
know
Mark,
like
I
said
we
we
worked
together
when
you
were
somewhere
else
and
now
it's
it's
good
to
see
you
here
and
thank
you.
Yeah
yeah.
D
B
C
Yeah,
it's
we're
working
for
global
companies
is
a
blast
in
my
opinion,
because
I
I
have
folks
on
my
team
that
are
in
France
Germany.
Let's
see
New
York
North
Carolina
India,
like
just
all
over
the
place,
I
used
to
make
a
joke
about
how,
like
the
Sun,
never
sets
on
my
Empire,
because
you
know
we're
we're
literally
worldwide
and
you
know,
there's
always
somebody
awake
and
doing
something.
Yeah.
C
C
No
I
remember
casting
before
veeam
I
had
I
had
a
number
of
conversations
with
you
all
when
I
worked
for
a
storage
vendor,
which
was
really
interesting,
so
I'm
I
would
say
I'm
familiar
with
the
technology,
but
really
that
was
like
five
years
ago.
So
I
can't
can't
actually
say
that
it's
I'm
sure
it's
dramatically
different.
Now.
D
C
C
This
week,
I
know
that
we
really
want
to
focus
on
Caston
and
all
of
the
capabilities
there,
because
when
we
were
talking
earlier
today
getting
ready
for
this,
you
all
just
had
an
amazing
amount
of
content
and
it's
stuff
that
I'm
really
excited
to
to
talk
about
and
to
show
the
audience
so
I
think
from
a
top
of
Mind
perspective.
Kind
of
the
biggest
thing
is
Johnny
and
I
are
going
to
start
running
through
a
series
of
streams
around.
C
What's
new
for
administrators
with
openshift
4.13,
so
we've
got
a
whole
list
of
stuff
custom.
Metrics,
Auto,
scaler,
Network
policy,
generator
openshift
virtualization,
with
pipelines,
hosted
control
planes
with
openshift
virtualization.
Let's
see
what
else
is
here,
Progressive
sync:
with
get
Ops,
what's
get
ups
with
multiple
sources,
Johnny
I,
don't
know
that.
A
One
yeah
so
get
up
to
multiple
sources
is
when
you
have.
You
know
you
have
like
a
Helm
chart
and
then
you
have
a
values
file
in
another
repo,
so
you
can
actually
pull
values
from
another
repository.
So
that
way
you
can
actually
you
can
have
your
defaults
that
go
in
and
then
you
can
have
another
source
to
actually
add
to
that
value
or
to
add
values
from,
and
then
it
allows
you
to
have
some
flexibility
on
how
you're
deploying
your
applications
using
get
Ops.
So
it's
it's
a
pretty
cool
feature.
D
A
C
Very
cool,
so
so
I
I
think
with
that.
You
know
for
audience.
Just
keep
an
eye
out
for
the
next
series
of
streams.
Remember
we
will
be
not
streaming
during
the
week
of
red
hat,
Summit
I
know
I
will
be
at
Summit.
So
if
anybody
is
there,
please
don't
hesitate
to
come
and
find
me
I'll
be
in
the
booth.
I'll
be
doing
Labs
I
think
I've
got
two
Labs
two
times
just
all
kinds
of
stuff
going
on.
C
D
Sure
and
and
again
I'll
take
a
stab
at
this
I
will
defer
to
my
colleagues.
Who've
got
a
lot
more
experience
and
a
lot
more
time
in
the
field
pitching
casting,
but
but
I'll
I'll
do
my
best
job
and
and
they
will
feel
them
for
me.
So
as
we
mentioned,
veeam
acquired
us
about
just
a
little
over
two
years
ago
and
we
had
and
the
the
company
had
very
much
started.
D
A
D
D
And
the
focus
from
the
very
onset
was
you
know,
and
and
it's
still
our
tagline
is
number
one
kubernetes
backup,
and
what
does
that
really
mean?
And
and
how
do
we
slot
into
veeam
I'll?
Give
you
the
big
picture?
The
long
story
short
is
that
veeam
is
the
pretty
much
the
global
leader
in
backup
and
Recovery.
You
know,
by
by
millions
of
measurements
by
millions
of
analysts.
D
Different
Cloud
providers
different
distributions,
but
we
found
the
most
success
with
openshift
clearly
in
in
trying
to
help
people
manage
their
workloads,
protect
their
workloads,
secure
their
workloads
and
migrate,
their
workloads,
so
that
that's
what
we
do
we
do
that
with
Simplicity.
We
do
that
with
Incredible
scale,
and
we
do
that
with
Incredible
customer
customization
and
automation
built
in
so
being
able
to
move
a
portion
of
a
workload,
persistent
volumes,
only
per
system
volumes
and
secrets.
You
know
looking
at
a
whole
kubernetes
namespace
or
looking
at
kubernetes
itself,
the
cluster
any
aspect
of
kubernetes.
D
We
can
basically
take
a
look
at
all
those
objects,
serialize
them
filter
them,
transform
them
back
them
up,
export
them
not
just
on
cluster
but
off
cluster
and
That's
essential
for
disaster
recovery
and
therefore,
by
the
same
token,
we
can
automate
with
policies
the
import
and
the
restoral
so
that
we
can
have
active,
active
recovery.
Basically
continuous
Disaster
Recovery
is
one
kind
of
mode
of
operation.
Although
most
customers
aren't
comfortable
with
that
so
they're
they
basically
have
disconnected
and
periodic
restores
for
Disaster
Recovery,
but
then
we
so
we
handle
all
of
these.
D
We
handle
all
of
these
as
automated
policies,
and
basically
we
get
to
the
point
where
you
don't
need
to
do
backup
recovery
as
a
post
process,
or
something
that
you
add
in
with
simply
annotating
a
namespace.
We
can.
We
can
take
care
of
an
entire
application,
all
of
the
objects
underneath
it
back
it
up
on
costs
or
export
it
off
cluster
import
it
on
clust
onto
the
next
cluster
and
whether
that's
openshift,
three
and
open
shift.
Four
on
other
ends
that
doesn't
matter
we
can
accommodate
Pearl
that's
transforms.
D
D
All
of
this
seamless,
simple
scalable,
we
give
you
the
GUI.
Everything
can
be
addressed
as
yaml
and
declaratively
automated,
and,
and
so
we
basically
make
it
such
that
it
can
be
seamless
automation,
policy
based
on
workloads
just
by
adding
an
annotation.
That's
how
simple
it
is
to
kind
of
yeah.
It's.
C
Funny
right
because
I
remember
talking
to
you,
know
Caston
way
back
when
and
they
were
some
of
the
first
that
recognized.
You
know
what
what
encompasses
an
application
in
kubernetes
you
know,
and
it's
more
than
just
a
PV,
it's
more
than
just
a
pod
definition
right.
There's
all
of
these
other
metadata
objects
that
surround
it
and
all
this
other
stuff.
You
know
Secrets
config,
Maps
a
lot
of
things
that
are
that
are
generated
dynamically,
that
all
have
to
go
into
protecting
that
application.
C
So
it's
really
and
it's
funny-
I
distinctly
remember
sitting
at
the
very
first
cncf
meeting
for
kubernetes.
You
know
in
in
New
York
City,
and
there
was
a
debate
around
whether
or
not
at
the
at
the
time
they
were
called
pet
sets
right.
Whether
or
not
persistence
was
even
needed
inside
of
kubernetes
right.
Everything
should
be
ephemeral.
Everything
should
be,
you
know,
just
deployed
and
into
and
and
introduced
to
the
cluster.
So
why
do
you
need
things
like
backup?
C
And
you
know
of
course
here
we
are
27
28
versions
later,
and
you
know
something
like
that.
That
was
pre
1.0
so,
and
you
know
clearly,
persistence
is,
is
something
that's
significant
and
important,
so
it
just
makes
sense
that
a
company
like
veeam
would
be
you
know,
integrated
and
and
a
major
player
in
that
that
Arena
exactly.
D
E
Also,
the
adoption
of
the
public
Cloud
leveraging
as
well
different
favor
of
openshifting
to
the
public
cloud,
and
as
soon
as
we
can
demonstrate
to
them
that,
through
a
quite
easy
process,
they
can
quite
rapidly
move
an
application
from
one
place
to
another
one.
All
of
the
sudden,
you
are
in
the
position
where
you
can
solve
a
lot
of
challenges
for
customers
in
the
market
right
now,
so
definitely
adoption
of
stateful
workload,
adoption
of
operators
and
that
whole
discussion
around
application.
D
Let
me
just
jump
in
and
answer
a
few
questions
and
and
then
augment
on
that.
So
so
Steve
asked
you
know
if
it's
correct
that
Beam
still
very
much
VMware
focused
and
the
long
story
short.
Is
that
traditionally
absolutely
right?
You
know
you
could
also
say
that
a
lot
of
those
VMware
workloads
are
windows
based
and,
and
that
would
be
a
majority
of
our
traditional
audience,
but
not
so
when
we're
talking
about
Cloud
native
workloads.
D
Although
this
is
where
openshift
virtualization
starts
to
come
into
the
story,
we
have
a
lot
of
customers
that
are
on
VMware,
that
have
more
traditional,
you
know
VMS
and
they
need
to
do
application,
modernization
or
infrastructure
modernization
or
get
me
to
the
public
cloud
and,
by
the
same
token,
we
we
so
that
lift
and
shift
strategy
we've
heard
everywhere
and
that's
nothing
new
Under,
the
Sun.
What
is
new
is
that
that
lift
and
shift
now
is
to
a
multi-cloud
infrastructure
provider
called
kubernetes
yeah
and
in
many
cases
that's
open
shift.
D
In
fact,
the
majority,
when
we're
talking
about
virtualized
and
more
traditional
workloads,
we're
almost
exclusively
talking
about
open
s,
virtualization
and
and
the
keyboard
project,
and
so
yes,
while
veeam,
has
a
lot
of
more
traditional
customers.
We
definitely
are
driving
very
much
into
things
like
kubernetes,
backup,
disaster
recovery
and
application
Mobility,
but
also
things
like
Salesforce,
Office,
365
or
Microsoft
365
backup,
because
even
these
cloud
and
SAS
providers
don't
recommend
that
you
rely
on
them
exclusively
for
backup
and
Disaster
Recovery.
C
Yeah
we've
talked
about
that
before
of
AWS
actually
has
a
really
great
document
that
you
know
shows
the
demarcation
between
like
what
they're
responsible
for
and
what
you're
responsible
for,
and
you
know,
they're
responsible
for
the
infrastructure,
and
you
know:
can
you
provision
a
new
ec2
instance?
Can
you
create
a
new
volume?
You
know
EBS
volume
not
like.
We
will
guarantee
that
you
can
recover
precisely
the
ec2
instance
that
was
there
before.
You
know
that
type
of
stuff.
D
C
D
C
I'll
I'll
point
out
that
you
know
with
openshift,
virtualization
and
VMware
and
any
hypervisor
for
that
matter.
There's
always
guest
based
agents,
the
that
you
can
use
so
that
from
within
that
operating
system
within
that
virtual
machine,
you
can
backup
data
from
inside
of
there.
I
think
where
veeam
has
really
been
successful.
C
Historically,
with
VMware
is
doing
hypervisor
level
backups,
you
know
you
can
do
them
more
efficiently,
more
quickly
in
many
ways
right,
it's
sort
of
abstracted
from
all
these
things,
I
think
if
I
remember
correctly,
I
have
the
ability
to
like
look
inside
of
the
guest
operating
system
to
extract
files
out
of
there
and
stuff
like
that.
Absolutely
so
you
know
bringing
that
type
of
capability,
a
hypervisor
level,
backups
and
I've
I,
don't
know
about
like
opening
up
the
disks
or
anything
but
bringing
hypervisor
level
backups
into
openshift.
C
Virtualization
is
fascinating
to
me
and
I
say
that,
because,
like
PVS
are
they're
different
than
a
traditional
virtual
machine
disk-
and
you
know
openshift
virtualization
stores,
vmdest
inside
of
PVCs,
so
I'm
curious
to
know
you
know
and
as
we
go
on
and
as
we
look
at
things,
I
I'd
love
to
learn
more
about
all
that,
but
we'll
get
there
we'll
get
there.
I'm,
not
I'm,
not
overly
worried
about
it.
D
D
We
look
at
that,
especially
when
we
compare
that
to
K10.
K10
is
our
commercial
product
for
doing
all
of
this
as
a
really
good
fundamental
basic
Tool,
a
good
place
to
start.
But
when
we
talk
about
what
organizations
need,
especially
on
the
scale
of
what
openshift
can
do
on
all
of
its
capabilities,
especially
when
we
get
all
the
way
to
openshift
virtualization
any
any
complex
application,
any
staple
application
just
using
oadp
would
be
a
good
start.
D
D
C
A
C
So
I
I,
don't
want
to
or
or
I
think
it
would
be
interesting.
I
think
it
would
be
good
if
we,
you
know
I,
don't
know
if
it's
possible
to
show
if
we
have
anything
else
that
we
want
to.
You
know,
chat
about
or
or
discuss
in
that
respect,
but
I'm
all
about
you
know,
seeing
things
and-
and
you
know,
asking
questions
and
I'd
love
to
see
what
it
looks
like.
If
that's
a
possibility.
E
C
E
One
of
the
important
point
when
we
start
discussing
especially
with
Enterprise
type
of
customers,
is
the
way
they
can
operate.
The
overall
solution
and,
and
one
of
the
one
of
the
thing
that
he
that
really
catch
the
attention
of
a
new
administrator,
especially
operating
an
openshift
platform,
is
the
fact
that
we
skating
you
have
a
UI
where
you
can
basically
perform
any
type
of
action
when
it
comes
to
protect
an
application
in
any
form
of
fashion
and
just
getting
a
simple
UI
where
you
can
easily
create
a
policy.
E
Protecting
one
application
or
set
of
application
with
a
very
low
level
of
effort
is
quite
quickly
and
a
great
asset,
especially
when
you
try
to
link
that
with
some
security
requirements.
The
fact
that
you
can
connect
the
K10
platform
with
almost
any
type
of
authentication
system
we
could,
we
could
name
the
openshift
authentication
mechanism.
We
could
also
mention
the
support
of
active
directory
or
the
oidc
connector
that
we
have
allowing
our
customers
to
implement
a
multi-factor
authentication
mechanism
to
get
access
and
secure
the
access
to
K10.
E
Those
type
of
things
becomes
more
and
more
interesting
for
the
type
of
population
we
used
to
we
used
to
deal
with,
and
on
top
of
that,
once
you
have
implemented
an
authentication
mechanism,
we
can
also
quite
easily
show
the
American.
You
know
what
I
can
do
it
right
now.
We
can
also
show
to
them
the
capability
of
K10
to
implement
all
the
exact
role-based
access
control
so
as
soon
as
you
will
need
to
delegate
totally
or
partially
some
actions
on
that
platform.
E
E
So
once
you
have
implemented
that
authentication
mechanism,
then
you
can
set
up
a
bunch
of
different
type
of
roles,
redefine
the
set
of
action
with
the
level
of
permissions.
You
would
like
to
give
to
your
users
or
project
owners
and
so
on,
and
then
you
can
very
rapidly
Implement
a
kind
of
self-service
action
on
the
on
the
cast
and
UI.
So
this
is
something
that
is
really
powerful
and
and
one
of
the
important
points,
especially
when
we
discuss
with
financial
services
customer
or
ways
customer
that
would
be
on
the
public
sector
vertical.
E
D
D
Do
through
the
GUI
you
pretty
much
can
do
completely
automated
and
it
sell
it
itself.
We
are
in
the
red
hat
Marketplace,
so
as
soon
as
you
create,
basically
a
cast
and
names
space
or
whatever
appropriate
to
install
in
you
can
use
the
casting
K10
operator
to
install
it
as
well
as
we
have
Helm
charts.
So
so,
basically,
we
are
a
cloud
native
application
on
a
cluster
that
then
allows
you
to
do
this
very
much
a
GUI
or
or
or
multi-cluster
multi-tenant,
our
back
self-service
model
or
completely
automated.
A
Awesome:
hey
real,
quick
Alex,
so
I'm
glad
you
mentioned
that
Mark
about
the
you
know:
yeah,
you
have
the
GUI,
but
you
also
have
the
the
Manifest
capability
as
well,
which
is
awesome.
Have
you
all?
Have
you
all
done
like
any
integration
for
like
plugins
on
like
backstage
or
anything
like
that?
So
you
know
like
backstage
and
idps,
and
so
that's
like
the
new
hotness
right
now
platform
engineering.
It
goes
into
all
the
self
self-service
conversation
like
do
you
all
have
plugins
forecasting
for
platforms
like
that.
D
At
this
point
in
time,
I
don't
I'm
not
aware
of
any,
but
the
truth
is
it.
This
is
100
doable
right.
It's
a
matter
of
you
know
getting
the
right
customer
probably
to
to
get
Michael
to
be
interested
in
doing
that.
Michael
talk
about
some
of
the
other
customer
asks
that
we've
had
say
with
tecton
and
pipelines,
but
but
for
platform
engineering
certainly
possible.
Although.
B
A
Yeah
just
look
at
the
rbac
and
all
the
stuff
that
you
can
do
from.
From
that
perspective,
it
seems
like
that
would
be
an
awesome
thing
to
have
kind
of
built
into
this
thing
where
it's
like.
Oh
hey,
look,
you're
an
app
developer.
You
blew
yourself
up,
go
click
this
button
to
restore
yourself.
You
know
that
yeah.
B
C
That
never
happens,
but
that
always
reminds
me
of
like
you
know,
as
when
I
was
a
VMware
administrator.
You
know
you
have
regularly
scheduled
snapshots
for
like
when
Patch
Tuesday
happens,
because
you
never
know
if
something's
gonna
happen
and
being
able
to
integrate
that
in
the
pipeline
again
super
powerful.
Yes,.
B
C
B
D
B
We
can
do
also
what
we
call
logical
backup
or
we
can
even
Leverage
The
API
of
the
database.
So
on
the
first
case,
the
thing
we
do
is
we
The
Logical
backup,
so
so
the
things
I
explained
we
do
what
we
call
a
crash,
consistent
backup.
Why
I'm
calling
that
crash
consistent,
because
it's
consistent
with
the
file
system
just
before
the
crash,
which
means
that
we
take
a
snapshot
snapshot
is
all
the
files.
B
In
the
same
time,
if
you
don't
take
a
snapshot,
if
you
do
something
like
very
Rook,
for
example,
you
capture
the
file
system
by
searching
the
monpoint
on
the
different
nodes
and
you
do
a
backup
on
the
live
PVC,
and
the
consequence
of
that
is
that
you
capture
a
dirty
backup
when
you
restore
you
never
know,
what's
going
to
happen,
yeah
the
thing
a
proper
backup
is
a
snapshot.
Then
a
snapshot
is
really
tight
to
the
storage.
So
you
need
to
put
this
snapshot
outside
of
the
of
your
infrastructure.
B
Besides,
you
want
to
have
a
portable
snapshot.
You
can
imagine
that
you
have
your
application
on
odf,
for
example,
with
a
rook
safe,
and
you
want
to
restore
on
EBS
on
AWS,
so
your
snapshot
has
to
be
portable.
So
what
we
do
we
take
a
snapshot
from
the
snapshot.
We
rebuild
the
volume
and,
on
top
of
this
volume
we
attach
the
data
converter
to
send
it
out.
So
that's
the
first
kind
of
backup.
The
second
kind
of
backup
that
we
can
do
is
what
we
call
a
logical
backup.
B
So,
instead
of
capturing
the
PVC,
we
capture
a
dump,
we
capture
MySQL
down
the
postgres,
Dom
and
elastic
dump,
all
kind
of
dump,
and
when
we
restore,
we
recreate
an
empty
PVC
and
we
execute
the
restore
with
the
dub.
So
that's
the
second
one,
but
something
we
can
do
as
well
is
to
leverage
the
eventual
AP
backup
API.
So
if
I
take
an
example,
a
classical
example
is
elasticsearch.
B
If
you
read
the
documentation
of
elasticsearch,
you
should
not
take
any
snapshots.
Well,
the
reality
is
is
that
if
you
take
snapshot
on
elasticsearch
most
of
the
time
when
you
restore
it's
working,
but
if
you
read
the
official
documentation,
it's
clearly
stated
that
you
should
not
take
a
backup
of
elastic
surgery,
snapshot
and
elasticsearch.
The
company
is
not
going
to
support
you.
B
If
you
do
that,
you
have
to
use
the
official
snapshot
API
that
you
can
leverage
on
the
elasticsearch
cluster,
and
we
have
case
where
custom
is
doing
that,
instead
of
taking
the
logical
backup
instead
of
taking
snapshot,
we
just
leverage
this
snapshot
API.
So
that's
the
Third
Kind
and
we
also
do
things
like
working
with
external
database.
Sometimes
your
application
is
on
a
namespace,
but
is
communicating
with
an
RDS
instance
which
include
SQL
instance,
and
we
can
take
back
up
over
there
as
well.
B
C
I
I
didn't
know
that
that's
really
cool
and
I
I
think
you
touched
on
something
there
that
I've
been
saying
for
years
and
usually
people
give
me
strange,
looks
which
is
snapshots
alone
are
not
backups,
no
they're
a
point
in
time,
recovery
and
most
storage
systems.
It's
you
know,
they're,
not
making
a
clone
of
that
data
and
calling
it
a
snapshot.
It's
part
of
the
logical
volume.
So
if
the
underlying
discs
fail,
your
snapshots
aren't
going
to
protect
you
from
anything
they're
they're,
just
a
point
in
time:
recovery
for
that
y
volume.
C
D
Vmware
admin
who's
just
really
used
to
the
GUI
great.
Let's,
let's
use
the
GUI,
but
once
they
start
to
understand
how
to
apply
policies
for
automated
compliance
and
backup
and
recovery
and
imports
and
Disaster
Recovery.
Well,
then,
then
they
understand
our
back
and
self-service
and
multi-cluster
and
so
and
then
they
can
delegate
back
to
people,
but
even.
C
D
You
know
we
try
to
help
keep
mature
customers
to
the
third
stage,
which
is
completely
work.
Loser,
annotated
applications
are
back
up
and
recovered
automatically
all
based
on
policy,
and
we
can
show
you
when
you
drift
out
of
policy
compliance
and
that's
really
where
we
try
to
drive
customers
from
Hands-On
I.
Do
it
all
through
the
GUI
to
all
right,
I've,
automated
my
policies
and
I've
delegated
self-service
across
multiple
clusters
to
completely
lights
out
automated
governance
and
Recovery
everywhere,
so
that
that's
the
journey
we
take
people
on,
but
but
what
Michael
also
hasn't
hasn't?
D
D
How
do
we
Trigger
elasticsearch
or
mongodb
dump
so
casting
K10
for
the
most
part
works
with
CSI
drivers
back
to
the
storage
providers,
whatever
they
are
right
and
we
drew
in
volume
snapshots
and
that's
only
one
small
operation,
it's
probably
the
most
important
operation
amongst
all
the
serialization
of
all
the
kubernetes
objects,
the
config
Maps
the
deploy
sets
everything
the
PVS
as
well
as
the
PVCs.
All
that
information
consists
of
everything
that
it
requires
to
do
an
application
restore
for
disaster
recovery.
So
it's
not
just
the
snapshot.
D
If
it
was
just
the
snapshot,
then
you
could
probably
just
use
beam
right
or
or
all
of
the
other
traditional
Solutions,
but
that's
not
what
it
takes
to
restore
and
migrate
applications
and
do
Disaster
Recovery
in
the
kubernetes
space.
So
we've
even
had
to
deal
with.
You
know
all
right.
What
if
there
isn't
what,
if
we're,
not
dealing
with
CSI
storage?
What
if
we're
dealing
with
more
traditional
storage
or
off
off
cluster
storage
or
NFS
mounts,
or
even
we
even
drive
back
to
vbr,
right
so
vbr
repositories?
D
So
there's
not
only
one
way
to
deal
with
things.
We
have
to
even
deal
directly
back
to
the
VMware
file
system
itself
in
some
cases,
because
that's
where
customers
are
okay,
so
so
yeah
Michael's,
given
you
like
the
second
you
know
level
down
the
next
level
down-
is
that
well,
there's
lots
of
permutations
of
complexity
to
deal
with
our
customers,
traditional
environments,
traditional
constraints.
D
We
try
to
do
everything
kubernetes
native
and
everything
that
we
do
is
kubernetes
native,
but
it
can
reach
back
to
more
traditional
storage
providers
and
more
traditional
operations
that
are
required,
we'll
execute
it
completely
from
K10
and
from
a
kubernetes
cluster.
But
if
it
requires
API
cli's,
you
know
mongodb
Dom
elasticsearch
dump.
Well,
we
actually
extend
K10
with
blueprints,
and
this
is
an
open
source
project
called
canister
and
canister
really
allows
us
to
not
have
to
customize
K10
for
every
possible
situation
right.
D
We
build
in
the
native
drivers
to
different
file
systems,
put
profiles
for
compliance
around
all
these
different
things,
but
then
blueprints
are
kind
of
our
expansion
release
mechanism,
where
we
can
customize
to
say
all
right
bring
up
a
container
in
this
pod.
With
these
certain
tools,
like
mongodb
dump,
do
this
quiescene
operation
before
allowing
K10
to
go
on
and
then
snapshot
the
the
persistent
volumes
and
so
on,
and
then
afterwards
restore
those
operations
to
unquiesce,
restore
back
to
a
load.
Balancer
bring
up
mongodb,
you
know
as
complex
as
needed.
D
Orchestrational
operations
in
the
traditional
world
we
can
shell
out
to.
We
can
provide
those
tools
all
from
the
kubernetes
control
plane,
but
they
can
do
the
most
traditional
operations
as
needed,
and
that's
how
we
really
meld
the
world
together
and
deal
with
all
the
Enterprise
complexity
of
what
it
requires
to
kind
of
bring
applications
on
an
open
shift,
no
matter
what
form
they
come
in
to
a
safe
and
auditable
and
restorable
world
of
data
protection.
Yeah.
C
Yeah,
audible,
being
important,
Steve
I
see
your
questions
there
I'm
going
to
skip
over
them
briefly
because
I
know
I,
know
Mark,
like
that's.
That's
right
up
your
alley.
I,
don't
think
you
might
have
paid
Steve
I'm,
not
sure
to
ask
those
questions.
I
I
want
to
address
Mira's
question
there
about.
So
why
do
we
not
back
up
from
the
live
volume?
What's
the
difference
doing
it
with
the
snapshot?
Doesn't
this
use
more
storage
I
feel
like
I'm
missing
something
so
Michael
I,
don't
know
if
that's
something
you
want.
B
So
the
question
the
question
basically
is:
is
it
correct
to
do
a
backup
on
a
live
volume?
Why
should
we
use
Snapchat
and
that's
that's
a
fair
question,
so
I'm
going
to
explain
that
so
imagine
that
you
do
a
backup
on
a
live
volume,
something
which
is
constantly
changing.
You
see
my
screen
by
the
way.
Yes,.
B
And
clear
yeah,
it's
clear,
okay,
so,
let's
take
a
very
classical
example
of
a
parent
relationship
in
a
database.
So
you
have
an
invoice
text
file
which
is
storing
invoice
and
you
have
a
line
inverts
text
file,
which
is
storing
the
learning
voice,
referencing
the
first
file
invoice.
So
this
is
a
very
classical
parent
relationship.
B
So,
let's
imagine
that
you
create
this
data
T1
at
time,
one
and
then
at
time
two
you
create
a
certain
invoice
with
the
corresponding
children
line,
inverse
two
one,
two,
two
two
three
and
so
on.
So
I
think
everybody
understand
at
this
point
but
question
what
happened
if
your
volume
is
urged
and
the
time
to
backup
the
first
file
and
the
time
to
backup
the
certain
times
are
not
the
same.
Let's
imagine,
for
example,
that
you
backup
invoice
at
T1
and
your
backup
line
invoice
at
T2.
B
B
C
B
C
Yeah
so
I
think
to
answer
the
other
part
of
your
question
Mira.
So
it
doesn't
just
use
more
storage.
Maybe
you
can
yeah,
maybe
I
think
it
depends
on
your
on
your
particular
storage
vendor
and
how
they
do.
You
know:
storage
protection,
oops
clicking
the
wrong
button
here.
Excuse
me
not
storage
protection,
but
how
they
Implement
snapshots.
C
So
some
storage
vendors
they
use
like
a
log
append
file
system
so
that
the
only
capacity
the
only
additional
capacity
that's
used
is
effectively
the
data
that
changes.
So
if
I
have
one
gigabyte
of
data,
I
take
a
snapshot
and
three
megabytes
of
that
data
gets
changed.
The
snapshot
only
consumes
three
megabytes
of
of
capacity
yeah,
some
storage,
vendors
Implement
snapshots
as
a
duplicate
of
the
volume,
and
you
know
so.
C
D
Vendor
exactly
and
and
this
is
why
your
mileage
will
vary
to
to
answer
these
questions
right.
Everything
is
nuanced
and
it
comes
down
pretty
much
to
the
particular
permutation
of
all
of
the
vendors
and
all
of
the
configuration
in
each
environment,
and
they
can
vary
for
the
same
vendors
with
just
slightly
different
configurations.
D
But
we
for,
for
the
most
part,
we're
dealing
with
longer
backup
Windows,
because
we
are
doing
monolithic
volume
snapshots
of
these
PVS
of
these
persistent
volumes.
And
so
that
is
why
one
of
the
one
of
the
projects
that
casting
engineering
is
involved
in
is
the
kubernetes
enhancement
proposal.
3314
for
change,
block
tracking
to
be
added
to
CSI
I
was
going
to
ask,
and
so
this
is
something
that
we're
actually
working
and
coordinating
a
little
bit
with
red
hat
engineering
and
even
the
Qbert
project.
D
Although
we're
in
the
very
early
stages
of
these
discussions,
in
fact
we're
still
kind
of
establishing
these
discussions,
but
but
long
story
short,
is
that
yeah
we're
we're
interested
in
making
sure
that
these
backup,
Windows
and
now
you
know
we
do
more
than
just
the
persistent
volume
snapshots
right.
It's
it's
everything
around
the
application
as
needed.
D
That's
that's!
Where
we're
trying
to
make
these
backup
Windows
as
small
as
possible
and
as
efficient
as
possible
such
that
we're
not
dealing
with
huge
monolithic,
Blobs
of
volume
snapshots
at
each
and
every
time
you
have
a
backup
window
in
the
worst
case
scenario,
and
that
is
certainly
a
huge
concern
of
our
customers,
especially
our
veeam
customers.
C
If,
if
I
can
rephrase
that
a
little
bit
Mark,
so
what
you're
talking
about
is
the
difference
more
or
less
between
a
full
backup,
each
time
versus
incremental
backups,
exactly
whereas
so
today
there
is
no
change
block
tracking.
There
is
no
ability
to
determine
what
has
changed
or,
or
only
what
has
changed
since
the
last
backup.
So
every
time
you
take
a
backup,
you
know
my
my
pvc
is
one
gigabyte.
It's
one
gigabyte
of
data,
that's
stored
on
the
backup
system,
each
time
with
change,
block
tracking.
D
Yes,
that's
exactly
right,
and-
and
so
this
is
this-
is
you
know,
an
expectation
that
everybody
has
coming
from.
You
know
the
traditional
world
of
VMS
and
bare
metal
and
veeam.
When
we
get
to
kubernetes
again,
it
depends
on
the
storage
provider,
exactly
what
is
capable,
and
so
we
have
Cloud
providers
and
storage
providers
and
you
know
open
shift.
You
know,
data
protection
has
some
of
these
capabilities,
but
to
expose
it
to
kubernetes
consistently
to
make
the
operations
consistent,
no
matter
what
the
vendor
or
no
matter
what
the
situation.
That's.
D
E
E
That
will
allow
you
to
deploy
K10
on
the
cluster,
having
up
to
50
work
or
not
so
five
zero
and
after
that,
first
month
of
trial,
automatically
that
embedded
license
will
be
converted
into
a
free
license
that
will
allow
you,
then,
to
cover
or
protect
a
cluster
having
up
to
five
worker
nodes.
The
objective
of
that
free
license
is
to
allow
you
to
evaluate
the
product,
so
you
won't
get
any
limitation
or
restrictions
on
the
functionalities
or
the
features
of
the
product
itself.
It
will
allow
you
to
test
the
whole
product
without
any
limitation.
E
The
only
restriction
will
be
about
the
support,
the
access
to
the
veeam
and
the
cast
and
support
itself.
You
won't
be
allowed
to
open
a
case
because
you
are
not
supposed
to
protect
production
environment
with
that
free
license.
So
from
that
perspective
there
is
a
free
license
emitted
into
the
project.
Everybody
can
download
and
install
K10
on
a
kubernetes
cluster
and
you
will
get
access
to
all
the
the
features
of
the
products.
That's
the
current
situation
right
now,.
C
I
gotta
say
that
aligns
pretty
well
with
the
the
openshift
developer
license
as
well.
Right
of
you
get.
What
is
it
16
entitlements?
Johnny,
you
know
no
charge,
you
can
use
them
for
single
user
non-production
use
cases,
and
you
know
we
were
talking
a
minute
ago
about
you,
know
integrating
with
pipelines
and
stuff
like
that
and
how
you
can
do
all
of
that
in
your
development
cluster.
D
Yeah
so
so
casting
K10
is
a
commercial
license.
It's
not
open
source.
However,
we've
talked
already
about
all
of
the
kind
of
needs
to
do
these
complex
data
operations
for
quiescene
and
application
consistent
backups,
and
that
is
an
open
source
project
canister
and
that
can
be
used
by
anybody.
You
don't
have
to
use
it
with
K10
and
we
have
a
few
folks
actually
using
it,
so
that
that's
on
GitHub
and
all
that
stuff,
and
that
is
one
of
our
three
open
source
projects.
D
B
Yeah,
maybe
we
can
speak
about
kopia
as
well,
is
a
very
important
project
for
us,
because
it's
a
world
data
mover
and
copia
is
an
open
source
project
and
we
have
something
like
three
or
four
developers
that
are
completely
dedicated
to
work
with.
So
we
really
contribute
actively
to
kavya
and
kopia
is
losing
many.
Many
other
use
cases,
but
we
are
the
main
contributor
of
Korea
today.
C
And
so
prakashas
does
cast
and
handle
application
aware
and
crash
consistent
backup
by
integrating
with
pre
and
post
activities
for
various
applications
and
Mark
I
know
you
were
talking
earlier
about
that.
Australia
was
Michael,
talking
about
logical
backups,
I,
don't
know
if
those
two
are
the
same
or
maybe.
B
Not
exactly
the
same,
we
have
some
new
skills
where
you
are
not
taking
logical
backup,
but
you
need
to
do
few
things
like
flushing
and
freezing
and
and
unlock
an
unfreezing
when
the
backup
is
finished.
We
support
that
as
well.
I
didn't
speak
of
this
use
case,
but
this
is
one
of
the
use
case.
Yes,
yeah.
D
C
D
D
Is
you
know
for
those
traditional
workloads
that
show
up
in
kubernetes-
or
you
know
up
here
under
openshift
virtualization
as
well
right,
and
how
do
we
again
if
we
can
coordinate
this
all
through
one
namespace
again
casting
looks
at
this
world
as
applications
and
we
try
to
make
the
Clusters
a
lot
more
ephemeral
right?
That's
that's
really
our
stance
and
that's
really
how
we
help
our
customers
get
to
these.
E
E
E
Okay,
all
right,
so
just
in
the
meantime,
what
we
are
capable
to
demonstrate
today
is
basically
we
have
two
type
of
environments:
The
Source
environment
is
based
on
openshift
deployed
in
Asia,
using
the
AR
or
service,
and
on
top
of
that
openshift
Crystal.
Of
course
we
have
deployed
K10.
This
is
the
interface
that
you
can
see
on
my
screen
right
now
and
if
I
go
on
the
application
view,
basically
I
will
be
capable
to
list
all
the
namespaces
all
the
applications
running
at
that
time
on
that
openshift
cluster.
E
So
if
I
use
the
filter
function
in
the
UI,
I
can
notice
that
there
is
one
specific
namespace
called
the
errorive
dash
Pacman,
and
this
application
is
basically
a
stateful
application,
and
what
we
can
try
to
show
right
now
is
the
way
we
can
create
a
policy
and
protect
this
application
running
on
openshift
in
azure.
The
policy
that
we
have
created
will
create
a
backup,
a
local
backup
that
will
remain
into
the
source
environment
and
once
the
local
backup
will
be
done,
then
automatically
K10
will
create
what
we
call
an
export.
E
In
that
case,
we
are
going
to
leverage
an
AWS
S3
bucket
to
get
a
copy
outside
of
the
production
environment,
and
something
that
could
be
interesting,
for
instance,
would
be
to
take
that
backup
that
take
that
K10
export
based
located
in
AWS
and
try
to
restart
the
same
application
on
a
different
cluster
and
then
on
the
secondary
site
on
the
disaster
recovery
site.
Let's
call
it
that
way.
E
We
have
a
bar
metal,
openshift
cluster,
running
into
a
demo
Center
or
data
center,
based
in
in
Switzerland,
in
the
demon
Center
of
HP,
where
we
have
a
bomb
it
or
openshift
cluster,
and
probably
we
can
try
to
restart
this
application
over
there
and
doing
the
restore
process.
There
are
a
couple
of
challenges
that
we
can
highlight
during
that
demo.
E
The
first
challenge
will
be
about
reconfiguring
the
PVC
with
a
different
storage
class
because,
of
course,
between
an
application
running
in
Azure
and
the
same
application
running
on
the
barometer
cluster
in
a
private
Cloud.
Necessarily,
the
storage
service
would
be
different.
So
the
challenge
is:
how
can
we
easily
reconfigure
the
PVC
during
the
restore
process
to
make
sure
that,
during
the
restart
process,
the
PVC
will
claim
a
persistent
volume
from
the
Right
Storage
service?
E
So
this
is
one
of
the
challenge
that
you
will
have
to
deal
with
when
it
comes
to
moving
application
from
one
place
to
another
one,
and
something
specific
in
that
case,
because
we
will
move
between
operand
shift
into
the
public
cloud
and
openshift
on-premise
will
be
about
how
can
I
reconfigure
the
route
itself?
In
this
example,
we
have
two
different
environment
and
two
different
DNS
domain
and
when
it
comes
to
restart
that
application
on
the
destination
site,
the
challenge,
the
other
challenge-
could
be.
E
How
can
I
easily
reconfigure
my
route
to
make
sure
that
I
will
use
the
DNS
domain
that
I
have
implemented
into
my
private
Cloud?
So
that's
the
type
of
thing
we
can
try
to
show
you
and
it
will
be
for
us
the
opportunity
to
demonstrate
the
the
power
of
that
application.
Mobility
use
case
that
we
have
in
that
case.
E
So
if
we
move
on,
let's
go
back
on
the
main
UI
and
try
to
have
a
look
to
the
policy
view
into
that
into
that
view
we
have
a
specific
policy
that
will
allow
me
to
protect
the
the
errorive
Pacman
namespace
that
I've
mentioned
just
before.
So,
if
I
click
on
edit
I
can
quickly
show
you
the
way
you
can
Define
the
policy,
because
everything
will
start
indicating
everything
will
start
from
the
definition
of
a
policy.
E
So
things
are
pretty
pretty
simple,
pretty
straightforward,
you
will
Define
the
type
of
action
that
you
want
to
perform
and,
of
course,
at
the
beginning,
you
will
need
to
create
a
backup
and
an
export.
So
that's
the
reason
why
you
need
to
select
the
snapshot
type
of
action,
then,
basically,
you
will
need
to
select
the
backup
frequency
if
I
try
to
change
the
setup
of
the
of
the
policy.
E
So,
in
order
to
manage
the
rotation
of
the
local
snapshot,
we
have
a
snapshot,
retention
policy
or
strategy
that
you
can
rely
to
manage
that
part
of
the
the
process.
Once
it's
done,
you
can
also
ask
K10
to
automatically
create
that
famous
export
that
will
allow
you
to
float
your
backup
outside
of
the
production,
environment
and.
D
E
Logic
is
pretty
much
the
same.
You
have
a
frequency
to
Define.
In
that
specific
case,
you
will
need
to
Define
and
provide
the
different
parameters
to
get
access
to
this
external
storage
repository.
So
we
used
to
call
that
a
location
profile,
so
we
have
defined
one
in
that
environment
and
same
thing
on
that
external
storage
repository
which
you
could
decide
to
apply
it
exactly
the
same
retention
strategy
between
local
backup
and
exports,
or
you
could
decide
to
have
a
bigger
retention
on
that
external
strategory
position.
E
E
C
E
Now,
if
you
need
to
be
a
bit
more
granular
into
that
process,
during
the
backup
and
also
during
the
restore
process,
you
could
leverage
the
filter
function
and,
for
instance,
you
could
ask
K10
to
only
backup-
and
this
is
just
an
example-
only
backup-
the
price
system
volume
for
all
the
namespaces,
all
the
applications
that
will
be
covered
or
protected
by
this
policy.
So
you
could
exactly
Define
what
you
want
to
include
what
you
want
to
exclude
in
terms
of
kubernetes
object
and
that
logic
you
can
apply
it
on
at
the
namespace
level.
E
Basically,
but
you
could
do
also
exactly
the
same
thing
for
all
the
resources
that
will
be
deployed
at
the
cluster
scope
level.
So
especially
if
you
have
applications
deploying
or
consuming
specific
crds
or
cluster
roles
or
Crystal
or
bindings,
you
could
ask
K10,
during
the
same
backup
process,
to
also
capture
those
informations
at
the
cluster
level,
and
you
will
get
exactly
the
same
logic.
E
Just
for
the
purpose
of
that
demonstration,
we
are
going
to
keep
things
simple
in
the
way
we
Define
the
policy
and
basically
I
will
trigger
that
policy
manually,
but
we
will
keep
the
export
function
enabled
and
we
will
offload
a
copy
of
that
local
backup
on
that
AWS
F3
bucket.
So
if
we
run
manually
the
policy,
let's
do
something
like
that
and
click
on.
Yes.
D
E
D
Want
to
add
one
thing
here:
Alex
along
along
the
way
that
we've
had
yaml
buttons
at
the
bottom
of
the
dialogues.
So
again,
this
is
all
being
done
through
the
GUI,
but
all
of
these
actions
are
declarative.
You
can
export
the
yaml
and
therefore
download
it.
Cube
control
apply
it
later
on.
You
don't
have
to
use
the
GUI
we're
CLI,
friendly
and.
E
So
if
you
have
a
quick
look
on
the
execution
of
the
policy,
you
can
see
on
the
screen
that
at
the
beginning,
K10
we'll
execute
the
backup
action
and
once
it
will
be
done,
then
automatically
we
will
start
working
on
the
creation
of
the
export.
So
during
the
export
phase,
once
again,
it
will
be
about
creating
a
copy
of
that
local
backup
getting
all
the
Pro
System
data,
all
the
kubernetes
specification
getting
all
of
that
into
a
specific
object
that
we
used
to
called
an
export
and
during
the
export
process
by
default.
E
E
So,
just
in
the
meantime,
the
export
will
probably
take
one
or
two
minutes.
There
is
something
else
that
we
can
show
you
on
the
way
you
can
implement
the
products
in
your
environment.
We
used
to
talk
since
a
couple
of
minutes
about
leveraging
an
external
storage
repository.
This
is
something
that
we
called
location
profile
for
instance,
and
if
I
edit
this
one.
This
is
exactly
the
way
you
can
configure
an
external
NFS
export
or
any
type
of
object,
storage
service
that
you
could
leverage
into
the
public
cloud
or
into
your
private
Cloud.
E
A
A
E
D
B
B
So
in
this
case,
we
leverage,
if
you,
if
you
most
of
the
time
we
are
working
with
access,
key
and
secret
key
for
this
free
bracket,
but
we
also
support
assuming
roles.
Okay,
was
it
your
question
and
we
also
support
workload
identity
when
we're
working
with
the
Google
bucket
and
if
we
are
working
with
things
like
NFS?
B
B
E
All
right
so,
as
you
can
see
on
the
screen
now,
the
backup
and
the
export
is
done
successfully,
which
is,
of
course,
good
news.
And
now
the
thing
that
we
will
need
to
do
is
to
go
on
that
destination
environment.
Try
to
import
the
export
that
we
have
offloaded
into
that
every
bucket
and
see
how
we
can
restore
and
especially,
transform
that
application
to
make
sure
that
the
PVC
will
work
with
the
Right
Storage
class
and
see
how
we
can
potentially
reconfigure
the
root
in
order
to
restore
the
networking
access
to
this
application.
E
So
just
on
that
Source
environment
you
can,
you
can
have
a
look
and
note
that
the
the
DNS
name
is
based
on
azure.cast10.io.
Okay.
This
is
the
source
environment.
Now,
if
I
try
to
go
on
the
Dior
site
and
I'm
going
to
change
the
the
UI
in
that
case,
you
can
see
that,
of
course,
the
color
of
the
UI
is
different
and
you
can
also
notice
and
pay
attention
to
the
to
the
DNS
domain
name,
which
is
Reddit
dot,
hpecic.net,
okay.
E
On
that
destination
site
once
again,
openshift
barometer
cluster
K10
is
installed.
If
we
have
a
look
to
the
UI,
that's
of
course
pretty
much
the
same,
and
the
first
thing
that
we
need
to
do
on
that
destination
site
would
be
to
figure
out
the
way
to
import
and
get
that
export
right
now
located
into
that
F3
bucket
available
on
that
destination
site.
So
the
first
thing
to
do
basically
is
to
create
what
we
called
an
import
policy.
E
But
basically
that
information
comes
from
the
source
environment
and
at
this
stage,
basically,
we
just
need
to
run
that
in
Block
policy.
The
execution
of
the
import
policy
is
pretty
quick,
because
at
this
stage
we
don't
move
the
export
out
from
the
location
profile
itself.
We
just
open
it
thanks
to
the
encryption
key
and
extract
a
couple
of
metadata
to
add
to
the
inventory
on
the
the
K10
application,
the
the
additional
restore
point
that
we
have
exported
at
the
previous
stage.
E
D
E
Yeah
absolutely,
and
the
last
step
is
how
can
I
restart
now,
this
application
on
different
environments,
different
type
of
storage
infrastructure
and
different
type
of
cloud?
Let's
put
it
that
way.
So,
if
I
go
back
on
the
application
View
and
try
to
see
if
the
air
rif
Pacman
is
there,
we
can
see
that
the
application
is
not
yet
there
because
we
have
not
yet
restored
this
application.
So
if
we
go
on
this
removed
view,
I
can
now
see
the
restore
point
that
we
have
just
imported.
E
So
if
I
click
on
restart
there
and,
for
instance,
select
the
last
export
or
restart
points
available,
I
will
now
be
capable
to
work
on
the
restock
Point
panel
and
Define.
The
way
I
would
like
to
restore
my
application
so
just
to
show
you
what
we
can
see
from
the
UI
if
I
scroll
down
from
the
interface
I
can
now
see
all
the
content
available
into
that
restart
point.
E
So
I
can
see
there
is
a
persistent
volume
claim
available
into
my
export
that
persistent
claim
comes
from
a
storage
class
called
manage
premium,
and
this
is
something
that
is,
that
belongs
to
the
Azure
storage
infrastructure.
Of
course,
when
it
comes
to
restore
this
application
on
my
new
environment,
this
is
typically
the
type
of
things
I
would
like
to
reconfigure
change,
transform
and
if
I
continue
to
scroll
down,
I
will
see
all
the
kubernetes
specifications
an
objects
that
K10
by
default
will
restore
during
that
process.
E
Of
course,
we've
presented
to
you
the
different
filters
that
are
available
during
the
backup
process
and
exactly
the
same
way
you
can
select
or
unselect
the
objects
that
you
would
like
to
restore.
So
by
default
we
will
restore
everything,
but
you
could
decide
to
restore
only
a
specific
config
map
or
specific
secret
or
specific
PVC.
If
this
is
something
that
would
be
required
for
you
now,
let's
talk
about
the
Transformations.
This
is
really
something
that
is
powerful
in
the
indicating
solution.
E
So
in
my
case,
I'm
going
to
show
you
how
to
do
it
manually
and
then
I
will
show
you
what
we
call
the
transform
set,
which
is
basically
a
kind
of
library
or
transformation
that
you
can
create
over
time,
and
you
will
be
capable
to
reuse
a
multiple
time.
Existing,
transform
actions
that
you
could
have
defined
previously.
So.
C
B
B
C
E
So,
basically,
when
it
comes
to
Transformer
PVC,
this
is
the
way
it
works
on
the
cast
hand
side.
So,
of
course,
you
you
will
have
to
select
the
right
resource
which
makes
sense,
and
then
you
will
have
to
define
the
type
of
operation
that
you
would
like
to
apply
on
this
type
of
resource.
So,
let's
add
a
new
operation,
as
you
can
see
in
the
UI,
there
are
a
bunch
of
different
type
of
operations
that
you
could
apply
on
kubernetes
resources.
E
In
the
specific
case
about
reconfiguring
a
persistent
volume
claim,
we
are
going
to
select
the
replace
type
of
action.
So
if
we
do
that,
we
need
to
provide
the
Json
pointer
path
to
the
definition
of
the
starch
class,
and
then
we
will
have
to
provide
the
value
so
basically
the
name
of
the
storage
class
that
we
would
like
to
use
in
order
to
replace,
in
my
case
here
the
manage
premium
storage
class
that
is
defined
in
my
export.
So
if
I
want
to
do
that,
basically
I
will
need
to
provide
the
path
to
the
resource.
E
I
would
like
to
modify
and
in
order
to
identified
the
storage
class
I.
Would
like
to
use
this
is
not
something
that
cast
and
Kate
and
can
Define
on
your
behalf.
This
is
the
type
of
parameter
that
you
will
have
to
provide
to
K10
and
once
you
have
provided
the
right
parameter
or
the
right
fitting,
then
K10
will
operate
that
transformation
for
you.
So
if
we
try
to
very
quickly
get
a
c
c
which
storage
class
we
would
like
to
use,
of
course,
this
is
not
the
right
one.
This
is
here.
E
E
So,
basically,
by
doing
this
I'm
telling
2K10
to
during
the
restart
process,
reconfigure
the
PVC
by
replacing
manage
premium
by
that
storage
class
available
on
my
destination
environment
and
just
to
make
sure
that
this
transform
action
will
properly
will
be
properly
executed.
I
can
even
test
that
type
of
operation
from
the
K10
UI.
So
if
I
click
on
the
test
button,
I
can
select,
the
persistent
volume
claim
that
I
can
see
here.
Click
on
this
arrow
and
try
to
compare
what
is
the
definition
of
my
object
before
and
after
the
transform
action.
E
So
let's
try
to
scroll
down
a
bit
and
I
can
see
the
definition
of
the
storage
class
here
and
if
I
try
to
scroll
down
once
the
transformation
will
be
applied,
I
can
see
that
the
name
of
the
storage
class
will
be
different,
and
this
is
exactly
the
setting
that
we
would
like
to
position
during
the
restart
process.
So
this
is
a
way
to
validate
and
verify
that
the
transformation
that
we
are
defining
will
work
properly.
D
Alex,
if
I
can
just
join
in
here's
another
example,
we've
had
to
do
this
for
openshift
virtualization.
You
know
we
need
to
delete
the
IP
address,
delete
the
MAC
address,
because
those
will
be
reprovisioned
either
on
the
fly
or
or
against
a
different
set
of
policies.
So
we
and
and
in
some
cases,
we've
actually
built
in
these
Transformations
or
made
them
part
of
a
library
or
even
built
them
completely
to
make
them
transparent,
because
we
understand
specific
VMI,
you
know
objects,
so
we
can
do
these
transforms
automatically
but
prior
to
us.
E
Yes,
absolutely
and
if
I
can
give
you
another
example
on
how
to
reconfigure,
for
instance,
a
route
because
we
are
changing
the
openshift
cluster.
In
that
case,
the
domain
name
is
completely
different.
You
could
do
something
like
that
and
we
are
going
to
use
as
well
the
test
operation
to
show
you
how
we
can
verify
that
the
change
of
the
route
will
work
properly.
C
Effectively,
you
know
making
it
super
simple
to
like
you,
don't
even
have
to
have
application.
Level
awareness
going
in,
like
you
can
use
castin
to
truly
do
the
entire
recovery
process.
Not
hey,
dump
the
yaml.
You
know
somewhere.
Let
me
make
some
modifications
because
it's
a
different
cluster,
like
all
of
it
is,
is
handled.
That's
amazing.
B
A
C
E
But
basically,
we
have
already
combined
and
I
will
show
you
that
during
the
restore
process,
basically,
we
have
combined
those
two
transform
actions
into
one
transform
set.
So
that's
the
reason
why
I
decided
to
click
on
add
a
reference
to
an
existing
transform
set,
so
This
ocp
Cloud
migration
will
combine
the
change
of
the
reconfiguration
of
the
PVC
with
the
storage
class
and
same
thing
about
the
root.
E
So
let's
try
to
use
that
I'm
going
to
click
on
the
restore
I'm
going
just
to
do
an
additional
modification
before
restoring
the
application,
and
hopefully
we
won't
get
the
demo
effect
and
everything
will
work
properly.
But
just
before
restoring
and
just
in
order
to
illustrate
the
filter
function
that
you
can
leverage
during
the
restore
process.
If
I
try
to
scroll
down
up
to
the
bottom,
you
will
see,
of
course,
because
it's
coming
from
the
source
environment,
the
definition
of
the
initial
storage
class.
E
So
this
is
typically
the
type
of
resource
that
we
don't
want
to
restore
in
that
example,
because
it
doesn't
make
sense
to
restart
the
setting
for
the
manage
premium
storage
class
on
this
environment
based
on
a
private
Cloud
infrastructure.
So
just
an
example
of
the
way
to
illustrate
the
the
the
way
you
can
leverage
that
filter
function.
Okay,
so
at
this
stage
we
are
ready
to
restart
the
application.
Let's
try
to
do
it.
Let's
click
on
restore
and
let's
see
if
everything
will
be
restored
properly.
E
So
on
the
main
dashboard,
you
can
see
that
the
wrist
direction
will
start
really
soon
and
in
the
meantime,
probably
we
can
go
back
on
the
openshift
Management
console
and
try
to
see
what's
going
on
from
an
openshift
Management
console
perspective.
So
let's
go
on
the.
Maybe
let's
go
on
the
pods
View
and
try
to
go
on
the
air
ride,
Pacman
namespace
that
is
already
restored
by
the
way
and
now
from
the
UI.
You
can
see
that
during
the
restart
process
we
are
going
to
trigger
a
couple
of
temporary
pod.
E
So
that's
the
case
for
the
Affinity
PVC
pod,
where
we
will
try
to
identify
the
most
relevant
worker
node
into
that
cluster
in
order
to
restore
the
the
PVC,
and
during
that
phase
we
will
also
operate
and
reconfigure
the
the
change
of
the
storage
class.
So
this
is
typically
the
moment
where
we
will
apply
the
transform
that
we
have
defined
during
the
restart
process.
So
it's
going
to
take
couple
of
seconds.
B
B
E
E
E
So
it
was
a
way
for
us
to
show
you
how
you
can
move
an
application
from
one
site
to
another.
One
that
case
it
was
a
kind
of
migration
between
openshift
running
into
the
public
Cloud,
to
an
operative
platform
running
on
premise.
It
could
be
the
other
way
around.
Actually
it
doesn't
mix
it
doesn't
really
matter
for
us.
E
We
can
move
an
application
from
one
side
to
another
one,
and
this
was
something
we
wanted
to
show
you,
especially
when
we
talk
and
when
we
discuss
with
openshift
administrator,
they
used
to
be
really
interested
by
the
the
way.
It
can
be
easy
to
move
a
such
type
of
application
from
one
place
to
another.
One.
C
No,
this
is
this
is
really
awesome.
I
I
mean
even
as
a
developer.
You
know
we
talked
earlier
about
different
capabilities
for
integrating,
for
you
know,
development
type
purposes
of
copying,
data
out
or
replicating
an
application
into
a
development
cluster,
or
you
know
a
CI,
CD,
Pipeline
and
being
able
to
duplicate.
All
of
that-
and
it's
it's
really
awesome
to
me
so
I
I,
think
I
hope
we've
covered
quite
a
bit
of
stuff
here,
I
feel
like
we've
covered
quite
a
bit.
C
C
That
being
said,
if
you're
not
watching
us
live,
if
anything
occurs
to
you
afterwards,
please
don't
hesitate
to
reach
out.
So
you
can,
of
course,
reach
out
to
me
at
andrew.sullivan
redhat.com
or
on
social
media,
at
practical
Andrew,
so
I'm
on
Twitter
and
Reddit
and
all
of
the
places
and
all
of
the
things
you
can
reach
out
to
Johnny,
Johnny,
redhat.com
or
Jrock
tx1
I
will
also
put
in
a
plug
for
the
kubernetes
slack
Channel
or
sorry
slack
Team.
What
are
they
called?
C
Slack
Channel
yeah
Channel
yeah
anyways,
the
kubernetes
slack,
where
there's
a
openshift
users
channel
on
there.
It's
a
great
place
to
go
and
connect
with
the
community.
Johnny
and
I
are
there,
along
with
a
whole
bunch
of
other
red
Hatters.
So
it's
a
great
way
to
interact
with
us.
So
prakash
does
kastin.
Take
SCD
backups.
B
Yes,
of
course,
we
have
a
blueprint
for
that
because,
as
you
know,
the
etcd
bird
is
is
not
a
managed
pod.
It's
I
can't
remember
exactly
the
name,
but
wait
we
deal
with
that
and
we
are
able
to
take
etcd
backup,
but
don't
get
me
wrong,
you're
not
going
to
restore
ATC
it's
interesting
to
take
a
backup
of
the
tcd
for
forensic
reason
or
for
eventually
legal
reason,
but
you're
not
going
to
restore
Etc.
B
B
C
B
D
Absolutely-
and
this
is
one
of
the
first
questions
that
we
usually
get
asked
after
you
know
why,
why
staple
workloads
on
kubernetes,
then
the
sex
thing
is
well,
isn't
SCD,
backup,
good
enough,
and
the
answer
is
no
not
ever,
and
so
that's
really
just
helping
our
customers
progress
along
to
understand
persistence,
kubernetes
objects,
disaster
recovery
and
and
all
the
nuances
that
we've
started
to
cover
here.
Yeah.
C
A
C
E
All
right,
no,
it
was
it
was
I-
was
really
happy
to
be
there
with
you
guys.
I
hope
that
the
demo
was
useful
for
people
to
really
understand
the
capabilities
of
the
solution.
Of
course,
there
are
plenty
of
other
topics
that
we
could
cover
as
well.
Michael
had
a
great
presentation
about
how
to
integrate
K10
into
detect
on
Pipeline.
E
B
D
Temptation
and
and
just
to
add
and
bring
it
all
together-
I
mean.
That
is
why
we're
we're
excited
to
be
a
good,
Red,
Hat
partner.
That's
why
we'll
be
at
Red
Hat
Summit,
but
but
most
importantly,
it's
because
we
solve
these
problems
for
our
joint
customers
together
on
the
largest
scale
and
we
meet
them
where
they're
ready,
right,
you're,
the
VMware
admin
with
Windows
workloads.
You
can
use
the
GUI
if
you're
ready
to
kind
of
do
everything
declaratively.
Well,
let's,
let's
export
the
ammos.
D
If
you
need
to
do
it
in
the
GUI
or
not,
and
then
let's
start
Cube
control,
applying
those
and
then
correspondingly,
when
you're
ready
to
do
this
all
through
operator,
driven
aspects
and
installs,
and
such
we're
already
ready
to
meet
you
there
and
in
fact,
as
you,
we
didn't
even
talk
about
this,
but
canister
even
drives
operators
right.
So
we're
really
very
much
ready
to
meet
you
where
you
are
and
very
much
take
you
through
all
the
stages
of
maturity
to
finally
Drive.
D
C
Yeah,
so
I
will
I'll
throw
you
under
the
bus
mark,
because
your
your
contact
information
is
there
on
the
screen,
so
anybody
has
any
questions
feel
free
to
reach
out
to
Johnny
or
I.
You
know
we'll
make
sure
to
include
the
team
from
cast
in
here,
or
you
can
of
course
reach
out
to
Mark
directly
since
since
he
volunteered
his
his
contact
information
I,
don't
ever
ask
folks
to
do
that
because
you
never
know
what's
going
to
happen
all
right
so
to
our
audience.
Thank
you
so
much
for
joining
us
today.
C
Again,
if
you
are
not
watching
us,
live
if
anything
occurs
to
you
later,
don't
hesitate
to
reach
out
to
the
team
from
Caston.
Thank
you
so
much.
This
has
been
really
enjoyable
and
really
informative.
So
greatly
appreciate
your
time
and
Johnny
I
will
see
you
at
the
same
open
shift
time
on
the
same
openshift
channel
next
week.