►
From YouTube: OpenShift Common Briefing #125: OpenShift on OpenStack with Ramon Acedo Rodriguez (Red Hat)
Description
In this briefing, Red Hat's Ramon Acedo Rodriguez presents the strategy, vision and current state of affairs for containers within Red Hat OpenStack Platform. In this session you will learn what role containers play in OpenStack, what we as Red Hat are focusing on, what we already have brought to our customers and what you can be expecting in the future in infrastructure as well as tenant layers.
Containers provide a compelling alternative virtualization for modern application development. OpenShift running on OpenStack allows combining the exposition of resources in the infrastructure layer such as network, compute and storage, provided by OpenStack, with in the consumption of these resources from the platform side by OpenShift.
A
You
hello,
everybody
and
welcome
to
yet
another
OpenShift
Commons
briefing
this
week
is
coming
on
the
heels
of
red
hat
summit
and
the
lead
up
to
OpenStack
summit.
So
we
are
really
pleased
to
have
Ramon
with
us,
who
is
the
OpenStack
product
manager
here
at
Red
Hat,
to
give
us
a
an
update
on
all
things:
open
shift
on
OpenStack
some
insights
into
some
new
things
that
are
on
the
roadmap
and
I'm
going
to
let
not
gonna
steal
any
thunder,
but
I'm
gonna.
Let
Ramon
introduce
himself
and
he's
last
30
minutes
afterwards.
We'll
have
live
Q&A.
A
B
Thanks
I
am
well
welcome
everybody.
My
name
is
Ramon
Acedo
and
I'm
the
product
manager
for
a
few
things
in
the
OpenStack
team
at
Red
Hat.
One
of
them
is
an
integration
with
OpenShift,
which
is
an
open
stack
but
I'm
going
to
cover
today,
let's
start
by
well
the
exposition
and
consumption
of
resources.
In
this
context,
so
on
the
one
hand
we
have
red
cart,
OpenStack
platform
open
and
stuck
in.
B
On
the
other
hand,
we
have
open
shipped
so
OpenStack,
as
probably
many
of
you
know,
is
infrastructure
as
a
service
platform,
where
you
have
resources
exposed
to
you,
which
are
consumed
too,
usually
by
tenants
or
by
other
applications.
Right
is
the
case
with
open
shift
right.
Those
resources
are
commonly
compute
storage
network
right
and
you
know
they
get
quite
sophisticated
as
well.
We
will
see
some
of
them
as
well,
a
load
balancing
a
service,
for
example,
and
then
on
the
consumption
side
of
things
we
have
open
shape.
B
Open
shift
is
well
what
am
I
gonna
say
to
your
open
tip
that
you
don't
know
what
it's
a
platform
as
a
service
that
can
consume.
Many
of
these
resources
royal
by
a
OpenStack
in
this
case
on
top
of
OpenStack,
and
we
are
doing
it
in
the
same
way
I
saw
when
she
does
it
in
other
platforms,
such
as
AWS,
for
example,
or
a
zero,
etc.
Okay,
so.
B
Yeah
sorry
I
skip
one
slight
exposition
of
resources,
so
here
we
have
open
stack,
which
is
platform
that
allows
us
to
automate
the
provisioning
of
bare
metal
nodes,
virtual
machines,
storage,
starting
as
block
devices,
for
example,
but
by
a
similar
or
even
as
file
systems
with
Manila.
We
have
also
Network
so
noble
stack,
tenants.
You
can
create
your
own
networks,
you
can
create
network
topology
with
Reuters
and
you
can
connect
it
to
the
outside
wall
or
you
can
keep
it
inside.
B
Only
for
your
virtual
instances,
for
example,
and
all
of
these
mas
I
said
earlier-
is
automated
it's
API
driven.
This
is
important
as
well,
and
thanks
to
that,
we
can
integrate
access
to
this
platform
or
with
any
other
platforms
that
I
understand
you
know
using
api's.
At
the
same
time,
OpenStack
itself
has
been
designed
to
scale
pretty
well,
so
the
OpenStack
infrastructure
is
composed
of
a
number
of
services
and
the
services
scale
pretty
well.
B
So
when
you
have
a
platform
when
you
have
in
this
case,
our
warming
workloads
will
be
containers
all
right,
I
hope
in
shift
on
OpenStack
and
if
you
need
to
scale
the
OpenShift
cluster
well
with
open
start
that
that's
an
easy
task,
because
OpenStack
has
been
designed
to
easily
and
seamless
X
scale,
for
compute,
for
storage
and
for
the
services
managing
these
services
and
on
the
consumption
of
resources.
Well,
we
have
OpenShift.
B
Openshift
itself
is
also
distributed
in
in
a
cluster
when
shifters
in
the
infrastructure
running
open
shift
and
as
you
know,
OpenShift
is
a
self-service
portal.
It
helps
developers
to
define
applications
to
share
applications
to
do
the
life
cycle
of
these
applications
in
a
way
that
we
all
know
about
right,
/,
natives,
but
on
asteroids.
B
Ok,
why
are
we
doing
this
integration
in
particular,
which
opens
you
know
among
all
the
products
that
we
can
run
OpenShift
on
which
you
know
it's?
It's
a
number
of
them,
so
open
stock
in
them.
But
let
me
start
first
by
open
shipped,
so
open
shift
is
workload
driven.
That
means
that
if
I'm
not
a
developer
and
I'm
using
open
shift,
I
don't
really
care
about
the
underlying
platform
right.
It's
open
style!
B
We
have
the
underlying
platform
as
open
stunts
in
this
case,
and
we
want
to
keep
that
experience
that
the
developer
expects
right,
meaning
that,
if
I
do
this
in
AWS
or
if
I
do
this
in
OpenStack
I
want
my
application
to
work
equally,
alright.
So
these
on
the
1r,
then
why,
on
openness
tank?
Well,
it's
deeply
integrated
what
we
are
doing
between
our
chieftain
OpenStack
but
I.
What
does
that
mean?
Well
I'm,
open
shift.
B
Moreover,
optimizations,
which
are
part
of
the
integration
as
well
network
wise
we
have
Korean
Korea,
is
on
the
opposite
side
and
Sdn
right
connected
to
the
OpenShift
CNI
and,
on
the
other
hand,
in
OpenStack,
it's
a
plug-in
for
Neutron
right,
so
Korea
will
be
able
to
make
connectivity
without
performance
penalty
and
right
between
openshift
containers
and
VMs
on
top
of
OpenStack.
If
you
think
about
this,
openshift
itself
uses
v
excellent
tunnels
to
communicate
the
containers
running
on
top
of
them,
and
it
happens
that
OpenStack,
that's
the
same
as
well.
B
B
What
else
OpenStack
is,
as
I
said,
API
driven
100%
with
you
know
all
the
actions
that
you
can
do
on
top
understand.
For
storage
for
networking
for
compute
right,
meaning
that
you
can
program
these
tasks
perfectly
from
your
applications
from
the
OpenStack
provided
for
open
shift
as
well
or
our
combination
of
the
two
all
right,
because
okay
I
keep
saying
that
the
developer
doesn't
care
too
much
about
the
underlying
platform,
but
what?
If
he
or
she
is
right?
B
Well,
in
that
case,
you
could
combine
the
two
platforms
and
your
application
could
perfectly
be
making
use
of
OpenStack,
because
you
know
that
you
are
running
containers
on
top
rope
instead
and
openshift
right,
and
this
is
common
as
well,
when
you
are,
for
example,
migrating
traditional
application
running
on
the
ends
running
on
OpenStack
and
your
container
izing
it.
While
you
in
that
process,
you
may
want
to
have
access
to
the
yeah
applications,
no
all
the
same
application
running
on
the
two
platforms
right,
so
you
have
access
through
api's
from
your
application.
B
If
you
want
OpenStack
as
well,
what
else
well
fully
managed?
What
does
that
mean?
A
dread
cut?
We
have
a
number
of
applications
to
provide
management
for
this
platform
12
in
shape
OpenStack.
One
of
them
is
cloud
forms
which
is
mana
tied.
You
up
stream
another
open
source.
You
can
download
check
it
out
and
it's
you
know:
it's
like
a
single
pane
of
glass
for
multiple
providers
being
providers,
OpenShift
OpenStack,
Amazon,
Web
Services
right.
So
in
one
view
you
have
access
and
management
for
all
these
platforms.
B
Right
again,
this
comes
very
handy
when
we
are
working
with
this
integration.
When
you
are
the
operator,
you
really
appreciate
to
have
this
visibility
of,
in
this
case,
these
two
platforms
and
you
can
even
add
more
platforms
and
supported
by
managed
IT
work
or
cloud
forms.
As
you
know,
we
work
with
ansible
at
record
until
is
part
of
our
portfolio,
and
it's
also
integrated.
B
You
know
it's
integrated
with
OpenStack
it's
in
treated
with
OpenShift,
as
we're
gonna
see
in
a
minute,
and
all
of
this
is
part
of
a
bundle
that
we
call
the
Red
Cloud
suite
right.
That
includes
things
us
such
as
satellite.
If
you
know
satellite
well,
it's
a
very
cool
tool
that
allows
you
like
cycling
or
packaging
visibility
of
what's
installed
where
and
to
what
version
and
etc.
And
it's
why
we
use
my
brush.
B
You
know
any
open,
sorry,
open
status
world,
but
read
cut
in
general
users
and
at
least
stock
that
this
slide
is
having
at
the
bottom.
You
see:
Red,
Hat,
Enterprise,
Linux
and
Wow.
This
needs
no
presentation
right,
really
solid
foundation.
It's
an
operating
system
that
has
been
trusted
and
respected
for
many
years
in
the
industry
and
all
the
stock
is
very
stone
on
rail
as
well.
Okay,
let's
see
a
Libra
life,
not
very
deep,
but
you
know
slightly
deeper
how
all
of
these
is
working
to
the
era.
B
So
here
you
can
see
at
the
bottom
the
hard
work
right,
so
you
know
cloud
our
servers
running
somewhere.
We
all
know
that
and
those
servers
are
representing
here
what
we
have
in
the
data
center
and
on
top
of
these
servers
we
are
installing
OpenStack
right
we
are
installing
safe,
safe.
Here
is
what
we
suggest,
and
you
know
it's.
The
most
used
towards
Vacanti
knob,
understand,
I
would
say.
B
B
An
OpenStack
cloud
forms
is
very
flexible
in
the
sense
that
if
you
come
up
with
a
workflow,
because
it's
something
that
you
are
doing
over
and
over
during
the
day,
you
can
very
easily
call
it
in
cloud
forms,
create
a
button
and
create
a
display
to
see
the
output,
and
you
can
have
it
in
in
there
integrated
in
your.
In
your
view,
right.
B
This
is
just
an
example.
I
mean
it
really
is
a
very
sophisticated
tool,
simple
to
install
by
the
way
people
working
with
cloud
forms
know
that,
like
you,
know,
OpenStack,
which
traditionally
has
been
more
of
a
complex
twist,
not
so
much
now,
we
have
really
simply
fight
over
the
stock.
A
lot
in
the
last
I
would
say:
10
yeah
especially
started
in
OST
10,
which
was
Newton,
and
we
are
now
about
to
release
our
speed
13,
which
is
based
on
Queens,
and
it's
way
way
simpler
to
install.
B
Actually,
it's
a
couple
of
comments
now
what
else?
Okay
I
mentioned
satellite
earlier
as
part
of
the
management
suite,
but
then
we
have
OSP
director
as
part
of
well.
This
management
OSP
director,
you
probably
have
heard
of
triple
o
as
well.
It's
the
OpenStack
installer
right
triple
o,
because
it's
OpenStack
on
OpenStack
right
director
itself
is
an
only
one
installation
of
OpenStack.
That
happens
to
include
also
ironic
right
in
this
one.
One
installation
and
ironic
I'm
gonna
talk
about
ironic
later
today
is
the
bare
metal
provisioning
service,
but
it's
with
open
stuff
right.
B
So
thanks
to
that,
we
have
a
two
capable
of
deploying
and
configuring
software
on
bare
metal
right,
and
this
tool
is
mainly
plain
OpenStack
by
the
way
when
I
say
Oh
SP,
as
the
OpenStack
distribution
from
record
address
that
we
have
a
policy
of
upstream
first.
So
it's
no
different
from
the
upstream
distribution
of
openness
that
you
have
the
services.
Everything
that's
downstream,
but
I
said
comes
from
upstream
with
no
changes
right
should
know
that,
then
we
also
have
an
civil
uncivil
is
used
integrated
with
director.
B
Uncivil
is
also
used
us
or
with
the
installer,
for
open
shift
in
general
and
for
OpenStack
as
well
sorry
I'm
for
yeah
yeah
for
OpenStack
as
well.
Okay,
let's
cover
a
little
bit
more
of
our
career,
so
I
said
before
fury
er
is
doing.
That
is
avoiding
the
double
encapsulation
that
was
introducing.
You
know
a
level
of
performance
penalty
up
until
now,
curious
ships
with
our
speed
13
as
I
said,
is
going
to
be
released.
Hopefully
next
month
during
you
know,
June,
and
it
connects
both
platforms.
B
It
connects
to
the
sea
and
I
part
of
hope
and
shift
and
to
Neutron
in
the
OpenStack
side
of
things
right
and
that
helps
with
what
I
said
earlier,
with
performance
in
particular
and
well.
Open
chief
can
really
add
the
containers
you
know,
but
you
can
really
make
use
of
it
for
an
efficient
communication
between
VMs
and
containers
and
containers
between
in
between
containers
as
well
running
on
top
of
VMs
ironic
and
they're
metal
as
a
service.
B
Some
will
refer
to
it
many
times,
ironic
again,
it's
installed
as
part
of
OpenStack
and
in
the
same
way
that
you
have
compute
nodes,
running
KVM
as
a
hypervisor.
You
will
have
ironic
running
plainly
and
directly
on
the
bare
metal.
Note.
Sorry,
you
won't
have
ironic
running
their
ironic
runs
on
the
controller's
ironic,
we'll
manage
those
bare
mental
notes,
I'm
going
to
explain
a
second
okay,
let's
review
a
little
bit
an
overview
of
the
topology
of
how
this
works
so
in
the
boxes
in
the
middle.
What
do
you
see
what
they
represent?
B
Our
build
well
instances
on
open
stuff,
alright,
so,
for
example,
you
have
one
built,
for
instance,
on
the
Left
called
the
bastion
house.
Well,
this
will
be
a
host
which
you
SSH
and
you
start
everything
from
there.
You
store
the
installation,
etc.
This
is
running
on
an
existing
OpenStack
platform
right,
so
you
have
created
this
VM,
you
SSH
into
it,
and
you
start
working
on
the
rest
of
the
installation.
If
you
are
working
on
installation
from
it,
you
have
load
balancers
as
well.
B
A
load
balancer
will
be
a
VM
right,
sophisticated
VM,
but
a
VM,
and
it
will
run
and
it
will
connect
to
openshift.
Whenever
you
need
load
balancing
services
for
your
applications
right,
then
you
have
the
infrastructure
of
openshift
itself
running
on
VMs.
By
the
way
we
are
working
as
we're
gonna,
see
later
in
the
roadmap
on
distributing
everything
with
this
integration
between
BMS
and
bare
metal
nodes
at
will.
B
So
us
an
operator
will
be
able
to
choose:
what's
your
best
topology
and
instructing
solar
to
do
it
as
you
as
you
defined
right
today,
you
can
do
it,
but
it's
a
more
of
a
manual
process
that
we
are
automating
with
inclusion
of
bare
metal
right
then,
at
the
bottom
you
see
open
stats
in
the
volumes.
The
OpenStack
has
a
service,
that's
called
cinder,
which
offers
you
block
devices
that
are
presented
to
the
virtual
machines.
B
Basically,
when
you
attach
a
new
block
device
to
agree
to
a
machine,
all
you
see
is
like
a
new
hard
drive
right,
/,
def
/
as
the
C,
sdd,
etc,
and
well
with
this
integration.
What
we
are
doing-
and
this
is
used
usually
for
persistent
data
right-
you
are
familiar
with
the
concept
of
the
persistent
volumes.
So
this
exactly
right.
Actually
it's
time
it's
on
a
one-for-one
like
when
you
present,
as
in
the
volume
to
work
on
training
as
a
PD.
Okay.
In
this
case,
we
were
also
using
it
for
the
infrastructure
itself.
B
So
the
registry
is
worked
by
a
single
volume,
for
example.
Okay,
then,
the
networks
here
you
see
two
of
them,
one
in
blue,
the
public
network
and
another
one
displayed
in
green,
the
private
network.
Both
our
networks
are
created
as
part
of
the
installation
on
OpenStack
right,
so
they
are
networks
managed
by
this
case
the
Newton
service,
which
is
the
networking
service
in
OpenStack.
Okay,
then
you
see
things
like
dnas,
LDAP
and
TP.
B
Well,
of
course,
you
may
want
to
have
DNS
for
the
external
resolution
I'm
supporting
internal
resolution
as
well,
and
then
what
you
can
have
is
an
extra
VM
with
bind
and
it
will
be
part
of
your
infrastructure
as
well
right,
and
the
same
applies
to
all
that
if
you
are
using
a
lab
and
if
you
want
to
add
an
NTP
server
instead
of
using
an
external
one,
this
can
be
part
of
your
infrastructure
as
well.
Oh
with
open
stun.
Okay,
all
right,
let
me
cover
real,
quick,
B
and
C.
B
It's
based
on
uncivil
2.4
and
well,
you
will.
If
you
go
to
the
link
that
I
showed
to
you
earlier
and
click
on
the
documentation
file,
you
will
see
the
details
of
how
to
do
this.
How
to,
for
example,
configure
OpenStack?
What
do
you
want
out
of
opens
that?
How
many
of
the
ends
you
want?
How
many
networks
and
what
are
they
names,
etc?
The
same
applies
to
the
OpenShift
configuration
right.
Do
you
want
one
infrastructure?
No,
do
you
want
3
infrastructure
nodes,
DNS
configuration
as
well,
so
what's
my
DNS
server?
B
How
do
I
access
my
DNS
server
so
that
I
can
make
the
updates
within
you?
Can
change,
set
cetera,
right,
Curie
or
networking
configuration
Curia
is
one
option?
It's
the
best
option
right
in
this
context,
but
you
may
want
to
tell
the
opportunity
so
that
you
want
a
difference
at
the
end.
Well,
this
is
where
you
would
do
it
in
the
opportunity,
installer,
etc.
Right
here
you
can
see
that
I
also
mentioned
the
credentials
configuration.
B
Good,
so
let
me
cover
real,
quick,
the
bare
metal
service,
us,
an
open
stock
block
manager.
I
also
look
after
ironic
itself,
but
said,
however,
provisioning
as
we
call
it,
which
includes
ironic
ironic,
is
probably
the
most
famous
product
within
our
provision
in
attractions,
an
ironic.
It's
an
enterprise
ready
bare
metal
as
a
service
running
on
top
of
OpenStack.
It's
an
open
stock
core
service.
B
It's
a
trusted
platform.
It's
been
there
for
many
years
now
it's
multi
tenant.
It
integrates
with
the
rest
of
the
OpenStack
services
right
with
Neutron
for
the
network,
with
Nova
for
computers
you're
going
to
see
in
a
second
with
scenar.
The
iteration
with
cinder
is
mainly
for
allowing
you
to
boot
bare
metal
nodes
on
cinder
volumes.
Right.
Imagine
you
could
have
these
class
servers
and
you
still
want
to
use
them
right,
so
you
could
do
that.
You
could
have
seen
the
volumes
created
and
then
have
your
bare
metal
out
booting
from
them
right.
B
The
goal
here
and
our
strategies
in
the
beginning
has
been
to
have
a
platform
or
bare
metal
provisioning
of
our
future
wise
with
that
of
the
virtual
instances
on
opens
not
right
and
I'm
gonna
cover
everything
that
I've
written
in
this
slide,
but
this
is
probably
a
good
summary
highlighting
all
the
benefits
and
features
that
you
have
with
ironic
right,
starting
by
its
simplicity,
how
it
is
used
in
different
use
cases,
for
example,
as
an
installer
for
open
star,
give
me
under
cloud
use
case
right.
It
covers
inspection
of
the
hardware
itself.
B
It's
able
to
talk
lldpe,
to
extract
information
from
switchboards
for
danique
cards,
that
the
metal,
note,
half,
etc
right.
So
it's
a
really
really
comprehensive
service
future
reach,
and
let
me
say
that
today
very
easy
to
use,
so
the
workflow
is
basically
I.
Have
a
new
server
I
registered
e
server
by
pixi
booting
it
to
the
ironic
service
and
after
that,
that
server
is
ready
to
be
used.
Ok,
it
integrates
directly
in
the
servers
that
are
there
ready
for
the
tenants
or
OpenShift
in
this
case
to
use
ok.
B
Our
tenant
on
the
left
is
OpenShift
and
civil
right
to
start
with,
which
is
the
installer,
the
installer
of
openshift,
and
how
that's
open
she
returns.
He
will
install
them
and
shift
on
OpenStack,
using
both
ironic
and
beautiful
machines.
Virtual
instances.
No,
but
it's
not
well
via
the
API
fine.
So
basically
you
instruct
the
API
hey.
Give
me
a
number
of
networks.
Give
me
a
number
of
virtual
instances
give
me
a
number
of
physical
modes
right
when
all
of
this
is
ready.
Let
me
know
and
I'll
go
ahead
and
deploy
OpenShift
on
them
right.
B
So
this
is
what
you
have
until
we
hear
the
API.
There
are
multiple
API
in
OpenStack.
This
conversation
between
our
point
IFFT
and
the
API
is
mainly
through
Nova
right.
Obviously,
you
end
up
always
talking
to
the
credential
service,
which
is
Keystone
and
other
services
such
as
similar
right,
but
in
this
case
the
most
important
communication,
I
would
say
to
start
with,
for
creating
the
resources.
B
Is
that
it's
via
the
integration
of
ironic
I,
made
a
note
here
in
open,
chieftaincy
long
is
currently
a
work
in
progress
and
a
lot
of
work
that
we
are
doing
as
we
speak,
we
plan,
and
you
know
this
is
a
plan.
Don't
don't
take
my
word
for
for
each
because
you
know
we
have
delays
for
finish
earlier
as
well
the
plans
by
the
end
of
the
year.
We
should
be
ready
to
do
this
end-to-end.
B
B
B
Definitely
this
summer,
hopefully
next
month
in
Queens,
OpenShift
ansible
will
be
able
to
create
all
the
resources
on
OpenStack,
as
I
said,
without
any
interaction
from
the
Installer
from
the
person
using
OpenShift
antilock
right
up
until
now,
operative
1004
open
star
expected
that
you
had
created
the
resources
right,
the
VMs
accept
and
when
you
have
them
do
yes,
tell
me
chief
91,
hey
here:
are
my
resources
go
ahead
and
deploy
everything,
but
since
OpenStack
is
API
driven?
Why
not
using
the
same
installer
to
do
this,
and
this
is
what
we've
done.
B
We
finish
this
work
and
hopefully
you'll
be
able
to
to
test
it
pretty
soon.
The
other
big
improvement,
as
I
mentioned
earlier
in
our
speaker
team
opens
winds.
Is
the
introduction
of
career
right
career
to
solve
or
to
improve
rather,
the
performance
of
the
network.
When
you
are
communicating
containers
and
VMs
containers
and
containers
in
different
VMs
right
and
octavia
load,
balancing
integration-
and
this
is
also
a
big
one-
right-
Octavia-
is
supported
in
in
Queens,
and
the
plan
is
that
it
will
be
fully
integrated
as
well
with
open
shift.
B
So
the
default
load
balancer
for
when
chieftain
opens
up
will
be
okay.
What
else
are
we
doing
we're
doing
many
things
this
year
and
next
year?
Hopefully,
everything
that
you
see
on
the
middle
column
on
the
right
one
will
we'll
be
able
to
ship,
starting
by
the
integration
with
my
metal
that
I
explained
earlier,
so
something
if
you're
familiar
with
triple
o
or
maybe,
if
you
in
our
developer,
you
may
have
seen
that
triple
o
is
now
able
to
deploy
open
shift
right
so
via
an
zero.
B
So
basically
using
ansible
OpenShift
on
civil
right
now
triple
o
knows
about
it.
It
has
templates
that
basically
call
play
books.
You
know
which
is
tangible
and
that's
everything
in
a
very
similar
way
to
what
do
you
do
for
installing
OpenStack
right?
So
this
pretty
cool?
This
is
very
exciting
and
it's
really
a
very
nice
way
and
quick
to
deploy
OpenShift
on
bare
metal
right,
plain
and
simple
we're
also
doing
the
same.
This
is
more
sophisticated
in
the
over
cloud,
so
the
temps
under
cloud
and
over
cloud.
B
The
under
cloud
is
the
node
that
has
triple
o
in
it
by
record.
We
talked
about
downstream,
and
the
over
cloud
is
the
OpenStack
platform
that
your
users,
your
tenants,
are
using
right
with
all
the
services,
not
just
those
services
for
installing
the
over
cloud
as
the
under
cloud,
but
everything
else,
including
ironic,
of
course,
and
we
want
openshift
ansible
to
be
able
to
point
at
any
other
cloud
and
an
empty
over
cloud.
B
You
know
an
empty
OpenStack
platform
and
say
please
install
Apache
for
me
me
being
a
tenant,
not
an
operator,
in
this
case,
just
a
tenant
who
has
access
on
OpenStack
platform
and
after
the
installation
I
have
my
own
OpenShift
infrastructure
ready
for
me
to
use
right.
So
your
only
limit
will
be
the
quotas
that
your
administrator
or
patron
has
set
for
you
right.
So
this
pretty
cool.
This
could
be
called
something
like
open
shift.
B
As
a
service,
if
you
like
right
because
it
is
pretty
much
what
it
is,
it's
a
way
of
self
servicing
openshift
in
OpenStack
platforms.
Well,
we
are
expanding
the
integration
with
storage
right
I
said
earlier
that
other
a
history.
For
example,
if
you
use
the
open
substantial
it's
configured,
we
seem
that
and
that
works
very
well,
but
we
see
a
lot
of
use
of
Ceph
and
in
particular
safe
with
you
know,
the
Swift
compatible
API
has
three
right:
Rallos
gateway,
so
why
not
using
safe?
B
Now
as
well
with
all
those
gateway,
just
like
seem
like-
and
you
say,
I'll
be
back
to
my
cell
pull
the
registry
right,
there's
demand
for
that.
We
see.
That
is
what
many
customers
are
doing.
So
we
are
adding
this
feature
to
work
on
shift
on
CEO
right,
something
important
as
well.
You
probably
have
similar
in
the
kubernetes
repositories
and
a
standalone
Cena.
There
is.
What
does
that
mean?
Well,
similar,
as
I
said,
is
what
gives
you
block
devices
from
OpenStack
when
you
are
a
bare
metal
node,
you
can
vote.
B
You
can
boot
that
note
from
a
similar
volume,
but
you
can
note
once
this
bare
metal
note
is
have
been
running
with
an
operating
system.
You
cannot
attach
volumes
to
it
right,
that's
not
something
that
seem
that
is
able
to
do,
but
if
you
have
containers,
if
you
have
open
ship
containers
in
done
not
with
pods,
you
still
want
to
be
able
to
present
seen
the
volumes
from
the
PDS
alright.
B
So
for
this,
what
you
do
is
using
the
standalone
scene
that
you
will
install
the
it's
done
or
not
you,
the
openshift
ansible,
will
do
that,
for
you
install
it
in
a
very
mental
notes
and
all
of
a
sudden,
you
have
again
the
same
experience
that
you
have
with
virtual
machines
were
transparently
you
get
cinder
block
devices
connected
to
the
virtual
machines
and,
in
turn
becoming
tvs
for
your
containers.
The
same
exact
experience
with
bare
metal
knobs
right
thanks
to
a
standalone,
sinner
check
it
out.
It's
there
in
the
kubernetes
repositories.
B
B
Also
again,
the
ability
to
scale
out
scale
back
in
from
on
civil
right.
You
can
scale
out
and
scale
in
nope,
Romero,
you're,
probably
doing
right
now,
but
what
we
want
is
an
automation
or
peace,
a
manual
automation,
automation.
What
I'm
saying
right
now
is
we
found
zero
me
as
an
operator
having
control
of
this
right.
Saying:
okay,
I
need
to
scale
my
workers
with
three
more
bare
metal
notes.
Alright,
so
I
go
there
and
I
say,
plus
three
install
them
and
by
the
time
the
playbooks
finished
boom.
You
have
them
okay.
B
So
this
what
we
are
doing
you
know
with
high
priority
as
we
speak
and
beyond
that.
Well,
we
have
big
plans
and
one
of
them
is
situation
with.
Manila
manila
brings
to
the
table
something
really
important,
which
is
multiple
rights
to
the
containers
right
rwx,
and
when
we
have
this
integration
through
vanilla,
the
PVC
can
be
presented
also
from
Manila.
When
you
have
this
use
case,
where
you
want
to
have
multiple
containers,
reading
and
writing
from
from
the
same
persistent
volumes.
Okay,
what
else?
B
Well,
we
are
also
working
on
auto
scaling
right,
I
skipped
one
on
purpose,
because
some
of
the
plans
from
that
are
changing,
which
is
the
deployment
of
OpenStack
itself.
You
know
which
shift
right
so
don't
we
that
one
I'll
I,
belated
and
operative
total
scaling
is
also
something
pretty
cool.
But
you
know
it
has
some
requirements
that
we
are
working
on
as
well,
and
we
plan
to
introduce
these
having
openshift,
auto
scaling
based
on
metrics
when
running
on
top
opens
and
then
integration
with
a
DNS.
B
If
you
know
I
mentioned
earlier
that
the
most
common
thing
is
running
mines
by
nine
use,
your
external
and
maybe
internal
as
well
neighbor
solution
through
it,
but
OpenStack
has
designate
the
NSS
a
service.
So
with
that,
what
we
want
is
to
integrate
it
with
our
chief,
so
making
OpenShift
ansible
aware
of
the
service
and
and
and
use.
C
B
A
B
B
So
I'm
not
familiar
with
DNS.
Thank
you,
but
designate
will
be
the
an
option
right
that,
instead
of
using
bind
or
other
DNS
resolvers
that
you
may
have,
for
example,
one
not
complete
chorus.
You
can
have
DNS
as
a
service
via
designate
connected
to
your
OpenShift.
It
will
be
an
option.
You
will
need
to
tell
the
installer,
but
you
want
to
use
the
NS
SS
eries
yeah
and
maybe
let
me
add,
when
we
are
doing
this.
Many
of
the
things
that
you
will
see.
B
For
example,
I,
mentioned
courier
few
times
its
options
that
you
have
as
an
operator
to
deploy
and
design
the
architecture
and
the
topology
of
what
you
want.
I
mean
it's
not
like
this
integration
forces
you
to
use
all
the
services
that
are
integrated,
but
only
those
that
you
think
are
right
for
your
environment.
C
A
C
Okay,
so
just
one
one
suggestion
I
might
add
a
slide
that
shows
that
just
kind
of
visualizes,
the
over
cloud
in
the
under
cloud,
because
you
do
mention
times
and
I-
think
it's
important
that
people
understand
that
relationship.
So
when
you
talk
about
you
know
ironic
and
the
under
cloud
versus
the
over
cloud,
they
they.
C
B
Is
it
yeah,
that's
the
most
common
use
case
this?
What
we're
planning
for
so-
and
this
is
the
first
thing
that
we
are
testing,
but
you
can
also
install
your
infrastructure
service
if
you
want
on
bare
metal
right,
but
the
most
common
thing
will
be
that
having
your
infrastructure
on
virtual
instances
and
a
mix
of
VMs
and
bare
metal
nodes
were
the
workers.
B
Director
will
deploy
only
environmental,
as
you
know,
so.
Director
will
have
to
deploy
everything
on
bare
metal.
The
infrastructure
and
everything
else.
Work
on
yeah
yeah
in
the
over
cloud,
though,
is
where
us
a
tenant
when
run
over
shift
on
until
against
it
will
be
able
to
do
some
mixed
environment.
So.
C
B
C
B
That's
not
something
directly
related
to
what
we
are
doing
right,
so
that
would
be
two
questions.
One
is
vgpu
supported
in
the
OpenStack
Nutkin
I
would
like
to
have
the
answer.
I
believe
it's
yes
right.
There
plans
around
GPU,
B,
GPU
I,
don't
know
the
details
of
them
and
then
is
it
supported
on
the
opposite
side
of
things.
C
C
So
there's
a
PCI
path,
pass-through
support
if
you're
familiar
with
overt
or
Reve,
that's
kind
of
the
H
dev
model
to
expose
it.
So
what
that
means
is
that
if
you
spin
up
a
virtual
machine,
you
can
attach
a
GPU
to
it
and
the
GPU
will
be
available
to
that
virtual
machine
exclusively
for
its
lifetime.
Until
you
delete
it
the
so
you
can't
kind
of
take
it
off
a
running
vm
and
give
it
to
somebody
else,
and
you
can't
share
a
single
GPU
across
multiple
virtual
machines
that
support
is
coming.
C
We
have
it
in
rev
for
two
right
now
so
once
that
version
of
k
vm,
the
support
is
supported
by
OpenStack
you'll,
be
able
to
share
a
GPU
across
multiple
virtual
machines
and
disconnect
a
GPU
from
a
virtual
machine
without
destroying
the
instance.
We
also
support
GPUs
directly
in
containers
and
for
people
using
openshift
on
top
of
OpenStack
I'd
recommend
they
expose
the
GPU
through
OpenShift,
because
you
can
disconnect
and
reconnect
a
GPU
to
a
container
and
OpenShift.
Today.
A
B
A
A
Well,
we
will
all
be
at
Red
Hat
summit,
Trevor,
Jacob,
I,
think
Ramon
and
the
whole
gang
of
other
people
from
the
open
ship
team,
Brian,
Jarvan
and
Josh
would
are
giving
a
training
on
the
Sunday
before,
which
is
the
Sunday
coming
up,
which
I
think
they
are
about
five
more
seats.
Oh
and
it's
free
its
containers
ribbon.
It
is
an
open
shift.
You
can.
You
have
to
be
registered
for
OpenStack
summit
to
attend,
but
if
you're
coming,
please
do
got
a
full
room.
So
far
we
can
hold
98
about
almost
90
people
registered.
B
A
A
And
if
you're
thinking
about
it,
we'd
love
to
talk
to
you
too,
and
get
your
feedback
in
requirements
and
make
sure
they're
incorporated
into
this
roadmap
and
future
ones
as
well.
So
there
you
go
I,
think
that's
it
I'm!
Looking
to
see
it
once
more
time
has
any
more
questions
and
I
will
stop
talking
and
we
will
meet
you
all
next
week.
Take
care
guys
thanks.
Mom
thank.