►
From YouTube: Intro + Deep Dive: IBMCloud SIG - Sahdev P. Zala, Khalid Ahmed, Brad Topol & Nimesh Bhatia, IBM
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 + Deep Dive: IBMCloud SIG - Sahdev P. Zala, Khalid Ahmed, Brad Topol & Nimesh Bhatia, IBM
In this session, the IBM Cloud SIG leads will first provide an overview of the SIG and highlight key SIG activities. The introductory presentation, in the beginning, will be followed by a deep dive into SIG upstream activities, updates to the cloud provider code, and the ongoing collaborative work with the SIG Cloud Provider. The presentations will be followed by discussions between SIG leads and attendees on current and future work items. Everyone --active contributors, new contributors, and conference attendees -- with interest in the SIG IBM Cloud are welcome to attend. During the session, SIG leads will be available all the time for discussions.
https://sched.co/MPks
B
A
All
right,
so
this
is
a
marathon
session.
We
have
80
minutes
to
talk,
it's
a
combination
of
both
introduction
and
deep
dive.
So
there's
no
rush.
You
know
we're
gonna
call
a
lot
of
things
around
this
see.
If
you
have
any
questions,
feel
free
to
ask
any
time
we
can
have
one
of
them
after
presentation
as
well
for
discussion.
So
we
can
use
that
time
as
well,
so
agent
down
pretty
simple,
we're
gonna
provide
a
seek,
introduction,
I
see
faces
here,
who
probably
are
not
so
much
familiar
with
the
sake
so
I'm
gonna
go
over.
A
The
introduction
would
basically
cover
the
overview
sake
structure
activities.
We
do
just
high
levels
and
this
that
will
be
followed
by
the
overview
of
IBM
cloud,
because
this
is
IP
IBM
cloud
specialist
group,
so
college
will
be
giving
glory
of
IBM
public
and
private
cloud
and
the
hybrid
strategy,
and
that
was
followed
by
a
little
deep
dive
with
the
C
sub
projects.
A
So
we're
gonna
talk
about
the
cluster
API
and
we
gonna
talk
about
a
game
park
forward,
an
interface
somewhere
to
talk
and,
like
I,
said
we're.
Gonna
have
a
lot
of
time
for
the
discussion
afterwards,
just
quick
questions
or
everyone
familiar
with
the
concept
of
see
ya.
Everyone
know
sig
right
here
and
within
six.
So
there
are,
you
know:
36.
There
are
few
six.
Are
cloud
providers
things
right,
so
all
the
major
clock
for
that
they
have
their
own
six
and
IBM
cloud,
for
that
is
one
of
those
things.
A
A
Some
the
like
I,
said
I
did
call
more
detail,
but
we
have
a
I
care,
so
IBM
cloud
cabinet
service,
the
to
public
service,
and
then
we
know
ICP,
which
is
private.
The
both
participate
in
the
CNC
have
certified,
Cuban
is
conformance
program
and
the
both
are
certified,
which
is
we
stay
on
top
of
those
certifications?
A
A
We
discussed
so
many
things
about.
You
know
what
we're
going
to
be
focused
on
it's
all
those
things
follow
under
the
sink
and
and
then
you
know
you
sort
of
communicate
directly
with
the
IBM
cloud,
tackle
aids
and
focusing
more
in
the
open
source
side,
all
right,
so
the
structure
arm
we
have
three
co-chairs.
This
is
this
is
funny,
but
this
is
more
like
you
know,
cluster
right
cluster.
You
know
four,
so
three
leads
are
good.
You
know
in
case
one
or
two
are
not
there.
A
A
Then
what
you
get
there
is
the
you
know
the
scope
of
the
see
like
it's
not
for
commercial
discussion
and
in
other
things,
and
then
we
follow
the
sync
governance,
typically,
its
its
same
governance.
All
the
sig
follows
we
have
made
in
list.
So,
if
you're
interested,
please
join
the
mailing
list,
you
to
stay
up
to
date
on
what's
going
on,
and
you
can
learn
more
on.
The
read
me
page
on
the
github
community
site.
I
should
use
this
one
activities.
C
A
A
Meeting
time
you
know
we
are
always
on
slack
it's
sort
of
24
hours
somebody's.
You
know
some
discussions,
some
questions,
you
know
what
not
right
sick
mailing
list
is
another
place
to
to
keep
connected
in
a
share
the
views
right.
All
the
meetings
are
recorded,
I,
make
sure
that
you
know
every
meeting
definitely
is
recorded
and
then
in
I
I
try
to
put
it
on
the
YouTube
channel.
The
cubed
is
YouTube
channel
just
the
same
day
or
the
next
day.
So
all
the
videos
are
there.
If
we
have
an
attorney
previously.
A
What
we
do
in
the
meeting
so
presentations
on
various
topics?
What
some
of
the
overviews
you
know,
the
things
like
you
know,
upgrade
strategy,
Cuban
ease
of
the
strategy
in
high
chaos,
ICP
anything
which
comes
out
a
new
feature.
From
that
perspective,
we
have
somebody
from
ICP,
I
guess
talk
about
it
in
the
first
10
50
meets,
and
it's
not
just
IPM
a
lot.
Folks
like
cystic
guys,
you
know
they
provided
the
talk
just
three
weeks
back,
so
it's
saying
or
any
member
interested
in
games
a
movie.
They
provide
it
as
more
educational.
A
A
A
A
But
close
to
a
VI
project,
it's
I
think
it's
a
part
of
the
project
under
sig
question
lifecycle.
So,
as
you
know,
we
know,
Q
breezes
is
getting.
You
know,
high
adoption
right,
it's
just
growing
and
with
that
the
requirements
from
folks
are
also
growing.
Oh
I
want
these.
Oh,
how
about
these
white?
What
now.
A
Requirements
was
about
to
managing
the
cluster
right,
so
Cuban
Aires
has
you
know
a
set
of
api's,
but
you
do
so
many
things
like
a
diplomat
resources,
even
things
like
load,
balancer
right.
Yes,
you
can.
How
does
he
peers
out
there?
One
thing
which
is
not
there,
it's
out
of
stock
unit
school
is
managing
the
cluster
itself
to
the
AP
is
right.
How
do
you
manage
your
cluster,
creating
a
node
removal
deleting
it?
You
know
dealing
with
the
lifecycle
of
the
node
right
and
the
cluster,
so.
A
A
And
we
have
a
link
coming
up,
so
the
following
slide
has
the
link
for
the
github
repo
as
well
yep.
We-
and
this
is
you
know
this-
is
when
the
project
we
would
like
to
have
more
contributors
contribute
there.
We
already
have
six
seven
contributors
up
there,
so
many
things
going
on
right
now,
like
I
said
we
announced
it
three
weeks
back,
so
it's
still
in
its
funny
face.
We
like
to
help
people
contribute.
You
know
this
is
a
good
place
to
contribute
drug
talk.
A
A
So
when
what
figures
be
a
couple
is
back
some
of
the
early
adapters
on
the
provider
side.
They
had
their
own.
You
know,
provider
interface
on
the
Cuban
is
poor
on
the
master
side,
so
those
codes
are
considered
like
entry,
it's
part
of
the
cube
controller
controller
manager,
so
it's
sitting
there
they've
been
using
it
eventually.
The
community
figure
out
that
hey.
This
is
not
the
right
way
to
do
this
kind
of
things.
You
know
this
is
adding.
B
A
Need
to
do,
then
they
still
need
to
create
some
versions.
Whatnot,
we
haven't
have
our
cloud
portal
out
there
on
the
it
up
externally
we
have
it
internally.
We
have
what
going
on
to
make
it.
You
know:
how
are
they
both
externally?
It's
a
the
open
suicide
has
been
already
approved
internally,
so
right
now
I'm
just
working
progress.
We
should.
A
With
this
it's
going
to
be
out
of
creams
out
of
trees.
You
know
it's
basically
loosely
coupled
now.
So
all
the
entry
providers
there's
this
cloud
for
their
sake.
I
will
be
talking
about
they
are
trying
to.
You
know,
encourage
everybody
and
everybody
agrees
to
move
their
code
from
entry
to
our
country,
so
that
you
know
this
binary.
They
can
live
on
their
own
quite
loosely,
coupled
from
the
Cuban
spent
cube
controller
manager,
KCM
architecture,
so
is
a
new
architect
for
CCM
I'll,
be
talking
about
that,
but
see
also
participates
in
those
activities.
A
So
there's
a
seek
cloud
provider.
Anybody
familiar
with
the
cloud
corridor,
sick
yeah,
so
popular
seek
was
created
last
year
and
and
the
main
focus
right
now
is
two
things:
the
concentration
of
individual
clapper,
six,
including
second
cloud
right,
all
the
clever
things
we
want
to
be
all
of
them
to
be
under
one
umbrella,
better
sing
and
what
they're
doing
so,
I
think
right
now
by
July,
before
around
the
time
of
CFP,
for
the
next
coupon
in
North
America,
to
send
you
one
I
think
that's
the
line
to
have
this
happen.
A
A
E
A
few
years
back,
when
you
know
IBM,
we
decided
to
make
kubernetes
kind
of
a
strategic
element
of
our
cloud
strategy
and
we
started
with
the
public
cloud
side
being
basically
being
able
to
you
know:
let
the
cloud
provider
us
IBM
manage
your
kubernetes
structure
right,
so
you
don't
have
to
worry
about
the
kubernetes
part.
You
worry
about
your
application,
then
we
also
because
we
have
a
lot
of
enterprise
customers.
They
have
security
requirements
and
data
privacy
requirements
and
so
on.
We
needed
an
on-premise
solution.
E
So
that's
where
we
took
kubernetes
and
pockets
not
to
be
able
to
be
deployed
in
your
environment
that
you
manage.
So
you
manage
the
lifecycle
of
kubernetes.
That
gives
you
that
flexibility
to
configure
kubernetes
according
to
your
security
requirements
and
your
application
needs
and,
most
importantly,
probably,
access
data
that
is
in
your
environment.
That
is
not
you
don't
want
to
put
on
to
the
public
side.
E
Thank
you
and
then
you
know.
Obviously
now
you
have
public
and
private,
and
not
only
do
you
have
you
know
just
one
public
cloud.
We
see
a
lot
of
clients
have
more
than
one
cloud.
There's
reasons
that
that
community
thought
will
get
into
that.
So
then,
how
do
you
bridge
all
these
environments
right?
How
do
you
manage
all
these
different
kubernetes
environments
distributed
across
multiple
clouds
and
clusters
and
make
them
you
know
consistent
and
even
format?
And
so
on
so
I'll
give
you
a
quick
overview
of
the
kubernetes
services.
E
Iks
is
the
IBM
public
kubernetes
service
servus,
and
it's
very
mature
right.
It's
been
around,
for
you
know,
three
years
almost
and
what's
more
importantly
or
what's
kind
of
differentiated
about
it
is
the
breath
of
workloads
that
it
provides
right.
It
supports
a
lot
of
the
you
know,
obviously,
open
source
tools
as
well
as
IBM
middleware.
We
leverage
it
for
IOT
applications.
We
leverage
it
for
Watson
blockchain
and
a
and
all
those
kinds
of
workloads
run
on
kubernetes
for
IBM
and
all
the
middleware
that
IBM
has
especially
enterprise.
E
Middleware
is
all
in
the
process
or
has
being
or
is
in
the
process
of
being
containerized
and
being
able
to
be
delivered
on
the
cloud,
and
this
is
important
element
because
it
gives
you
that
portability
to
move
between
public
and
private
and
public
back
again
move
back
and
forth
between
your
environments.
You
might
develop
your
application
on
the
public,
but
because
your
data
is
inside
your
firewall,
then
you
need
to
be
able
to
deploy
it
that
way.
So
the
IPS
service
provides
a
variety
of
kind
of
configurations.
E
You
obviously
have
you
know,
signal
kubernetes,
cluster
and
one
availability
zone.
You
can
also
take
a
kubernetes
cluster
and
we
support
multi
zones
right,
so
you
can
have
workers
front
across
different
zones,
so
one
availability
zone
go
down
and
the
other
availability
zones
also
are
are
up
and
running
in
order
to
you
know,
keep
your
applications
available.
Okay,
we
also
support
gpu-based,
kubernetes
environments
for
AI,
workflows
or
high-performance
workloads
and
so
on,
and
we
support
this
across
the
multiple
data
centers
right
in
six
different
regions
over
25
plus
data
centers.
E
We
have
regional
data
centers
to
support
the
IKS
so
giving
you
that
workload,
flexibility
across
if
you're
an
enterprise
company
you
can
have,
and
you
have
data
local.
You
know
data
privacy,
kind
of
gdpr
or
those
kinds
of
requirements.
Then
you
can
locate
your
cluster
where
your
applications
or
where
your
users
are
in
terms
of
the
capabilities
that
we
provide
it.
We
kind
of
emphasize
the
simplicity
aspect
of
it
right
because
we
are
kicking
over
the
process
of
installing
and
upgrading
and
updating
your
kubernetes.
E
And
if
you
don't
have
to
worry
about
that
right,
you
just
click
a
button
when
you
decide
to
upgrade
and
the
system
will
go
off
and
and
do
that
so
the
simplified
cluster
management
is
a
key
aspect
of
it
and
we
give
you
the
flexibility
to
how
you
want
your
cluster
to
design
design
which
domes
in
which
that
how
many
knows
whether
you
need
GPUs
or
not
and
all
those
kinds
of
considerations
are
there.
And
obviously
you
know
when
you're
configuring.
All
these
security
is
important.
E
F
E
Think
yeah
we
do
have,
we
do
have
fast
internet
I,
don't
know
the
exact
number
I
think
at
least
ten
kegs
of
25
gigs,
which
we
can
within
the
data
centers
I
can
find
out
more
about
that.
If
that
that's
important,
I,
think
you're,
probably
more
concerned
about
between
the
on-premise
and
the
cloud
right
and
I,
think
yeah
I'm
not
sure
exactly
the
configuration
of
those
data
centers
but
yeah.
E
E
So
another
aspect
you
know
the
cluster
management
is
one
thing,
then
there's
all
the
other
management
tools
that
you
need:
logging,
monitoring
and
so
on,
and
we
provide
those
capabilities
out
of
box.
We
love
heard
certain
partners
like
sis,
big
and
log
DNA,
to
help
them
not,
and
you
don't
have
to
worry
about
managing
those
those
tools
right.
It's
all
integrated
with
it,
not
so
one
of
the
interesting
aspects
that
came
in
from
the
the
kubernetes
or
our
IPAs
implementation
and
simplified
cluster
lifecycle
management
arena.
It
was
announced
yesterday
and
brad
has
talked
about
it.
E
It
automatically
updates
and
patches
and
does
all
those
kinds
of
things
and
for
the
kubernetes
clusters,
but
that
that's
a
quick
overview
of
the
public
thought
side
right
and
then
you
know
on
the
project
outside
we're
taking
the
same
upstream
kubernetes
packaging
got
together
with
you
know,
a
bunch
of
other
management
tools
that
you
need,
like
logging,
monitoring,
security,
alerting,
etc,
all
based
on
CN
CF
projects
and
providing
that
to
clients
that
want
a
on-premise
active
in
Eddie's
environment
that
they
can
manage
themselves
right.
They
have,
they
have
full
control
of.
E
You
know
how
kubernetes
r
is
deployed,
which
LS
it's
deployed
on
and
how
docker
is
configure
than
etc
right.
So
it
gives
you
that
that
full
control,
but
it
rather
than
you
know,
just
taking
all
the
open-source
components
and
trying
to
integrate
them
and
making
sure
that
they're
all
certified
and
checked,
and
you
know,
virus
scanned
and
and
all
those
consequently
scanning
and
all
those
kinds
of
things
we
do
not
for
you,
that's
kind
of
the
value
of
the
IBM
cloud
private
from
that
perspective.
D
E
So
we
provide
support,
obviously
for
the
IBM
middleware
right
for
the
open
source
middleware.
We
provide
kind
of
like
community
support
right,
so
we'll
look
at
your
issues
and
triage
them
and
work
with
the
community
to
fix
them
yep.
So
we
have
partnerships
right
with
the
open
source
vendors
that
are
backing.
Those
projects
to
you
know
take
those
those
were
plus.
E
E
I
didn't
mention
on
the
can
I
be
in
public
cloud
right
beyond
kubernetes.
We
also
provide
support
for
things
like
sto
and
key
native
right,
managed
services
for
that.
I
should
have
mentioned
that
before,
but
mention
it
now
and
for
those
things
we
also
I
sto
and
also
supported
on
the
private
cloud
side.
Okay,
just
a
little
bit
of
a
deeper.
E
The
the
architecture
of
the
solution,
so
you
know
from
the
infrastructure
point
of
view,
we
support
all
the
different
types
of
infrastructure,
including
power,
the
as
well
as
x86
and
various
storage
options
that
the
community
supports.
We
automate
that
deployment.
We
provide
a
terraform
based
deployment,
automation,
tool
for
the
infrastructure
management
and
then
we
support
all
the
container
management
capabilities,
including
the
image
registry
and
the
service
fashion,
and
so
on.
D
E
Yeah,
typically,
when
you
create
the
cluster
dust,
you
have
to
decide
between
a
single
node
or
an
HJ
deployment
right,
I.
Think
currently,
the
automation
tools
don't
allow
you
to
increase
it
foster
from
one
to
three.
So
you
have
to
decide
whether
it's
three
or
one
when
you
create
the
cluster
I,
think.
B
D
E
E
B
E
Know
the
IBM
provided
dashboard
as
well
that
kind
of
integrates
all
these
tools
together
and
then
you
know
you
have
all
supporting.
You
know
a
variety
of
of
middlewares,
whether
it's
cloud
native
or
containerized
middleware,
you
know,
or
whether
it's
for
a
blockchain
or
app
servers,
and
so
on,
data
workloads
and
a
lot
of
the
DevOps
tools
like
jenkins
and
those
kinds
of
things
all
are
enabled
on
on
this
platform
right
and
you
can
connect
out
to
public
cloud
service.
We
support
the
service
broker.
E
Open
service
broker
model
allows
you
to
connect
to
other
services
that
are
not
running
in
kubernetes,
but
are
want
to
be
consumed
by
applications
running
in
kubernetes.
So
we
support
you
know
things
like
even
for
your
existing
mainframe
environment
or
things
like
that.
If
you
want
to
connect
to
that
and
consume
services
from
z/os
or
or
Z
Linux
support
for
C
Linux
IBM
cloud
private
can
actually
run
on
on
C
Linux.
So
we
support
all
the
Linux
is
so
that's
kind
of
the
private
cloud
side.
E
Let
me
talk
briefly
about
the
multi
cloud
management
capabilities,
so
you
know
the
reality
is
that
you
have
multiple
clusters
today,
right
you're,
not
just
gonna,
have
one
cluster
is
even
you
know,
for
a
small
environment
or
for
the
environments,
test,
environments
and
production
environments.
But
then
you
know
there's
a
lot
of
considerations
around
from
your
application
point
of
view.
E
Isolation
right,
multi-tenancy
might
be
one
reason
for
creating
multiple
clusters
of
different
lines
of
businesses.
Don't
want
to
see
each
other
or
interfere
with
each
other
and
so
on.
You
have
data
considerations
right,
so
some
clusters,
don't
have
to
be
Jia,
got,
have
to
be
separate
right
to
meet
data
residency
requirements.
E
Other
things
like
you
know,
cost
right.
Maybe
there's
you
know.
Certain
providers
at
certain
times
are
providing
their
infrastructure
or
their
containers
containers
services
at
lower
cost
and
others.
So
they
may
decide
to
go
with
one,
but
then
you
know
for
a
period
of
time
you
may
run
it
on
another
and
then
load
considerations
for
your
applications.
F
F
G
F
B
B
F
E
Think
that
is
definitely
an
item
on
our
roadmap.
We
are
doing
work
even
the
current
release
in
terms
of
better
supporting
and
it
multi-tenancy
is
not
just
about
kubernetes
itself
right.
There's
multi-tenancy
for
your
infrastructure,
there's
multi-tenancy
for
your
management
tools
right
there,
logging
monitoring
and
so
on.
So
we're
doing
some
of
that
work
to
kind
of
enable
that,
from
the
end
stock
perspective,
right
multi-tenancy,
another
aspect
of
multi-tenancy
is
kind
of
things
like
resource
allocation
between
them.
So
things
like
photos
and
so
on.
E
E
In
the
context
of
the
public
cloud
we
implemented
multi-tenancy
in
a
way
one
of
the
models
is
kind
of
running
kubernetes
in
kubernetes
right,
where
you
leverage
kubernetes
to
spin
up
the
demasters
of
other
kubernetes
clusters
right.
So
that's
an
effective
model
that
we
found
on
a
public
cloud
side
to
deal
with
the
scale
of
clusters
on
the
public
cloud
we
have
to
manage.
E
You
know
tens
of
thousands
of
clusters
right
and
creating
masters
for
each
of
them,
and
you
can't
give
them
the
same
API
server
right
so
to
two
different
people
can't
share
the
same.
So
you
have
to
have
a
separate
API
server
for
each
of
them.
But
then
you
don't
want
to
provision
physical
machines
or
VMs
for
each
of
those
right.
So
you
need
to
find
a
way
isolating
those
different
tenants
giving
them
their
own
API
server,
while
saving
costs
on
your
hardware
as
a
common
provider.
E
E
And
we've
had
a
few
two
or
three
iterations
since
that.
So
today
the
multi-cloud
manager
allows
you
to
connect
and
manage
super
Nettie's
clusters
wherever
they
are
right
on
your
on-premise
environment,
right
in
the
IBM
cloud
on
Amazon
on
Microsoft
or
Google
right.
We
support
all
those
options
and
you
have
centralized
dashboard
and
visibility
into
the
utilization
of
those
resources
and
and
the
details
of
the
underlying
printer
kubernetes
resources,
and
you
can
search.
B
E
It
also
supports
a
model
of
compliance
and
configuration
checking
model
where
you
can
ensure
that
all
these
clusters
are
configured
according
to
your
specification
in
terms
of
network
policies,
pod
security
policies
etc,
and
we
provide
a
application
model.
That
is
an
extension
of
the
kubernetes
sick-out
capabilities
right.
That
allows
you
to
create
an
application
which
consists
of
multiple
components
right
that
can
be
deployed
across
different
clusters
and
gives
you
visibility
into
those
applications
by
automatically
general
for
fun,
dashboards,
for
the
components
of
those
applications
and
presenting
dr.
E
tivity
users,
and
all
these
capabilities
are
implemented
as
extensions
to
the
kubernetes
api
right,
so
they're,
custom
resources
that
we've
added
that
we're
working
with
the
community
we
align
with
whatever's
there,
and
example
in
cube
federation.
We
leverage
the
cluster
registry
and
then
we
add
on
certain
fields
and
annotations
so
that
it's
compatible
with
you
know
what
other.
F
B
E
D
B
E
And
manage
and
deploy
workloads
and
those
environments,
one
of
the
other
important
things
we
think
will
come
down
is
the
notion
of
scheduling,
scheduling
management
so
on.
So
one
of
the
aspects
that
we
have
is
a
you
know
in
the
high-performance
computing
world,
there's
a
lot
of
scheduling,
capabilities
and
resource
management
capabilities.
E
We're
looking
at
brigham
got
in
into
kubernetes
right
and
we
have
an
initial
version:
technical
preview
that
allows
you
to
use
an
HPC
scheduler
inside
kubernetes
as
a
pod,
scheduler
right,
and
you
basically
get
the
benefits
of
some
of
the
policies
like
fair
share
and
reservation
and
prioritization
in
a
kubernetes
native
way
right.
So
you
don't
have
to
change
or
modify
your
applications,
and
it
gives
you
that
capability.
C
E
D
E
D
E
This
so
we
did
introduce
some
custom,
a
custom
resource
to
allow
you
to
specify
parallel
jobs
right
because
kubernetes
doesn't
have
the
concept
of
a
parallel
job
where
all
the
pods
need
to
be
Co,
scheduled
together
right
and
we're
working
with
the
other
communities
like
to
keep
and
so
on.
To
say
you
know
what
what
this
thing
should
be,
but
basically
it's
exposing.
You
know
in
a
kubernetes
native
way,
the
policy
engine
of
the
so
go
ahead.
Sorry.
G
E
A
A
In
progress
Richard
this,
he
is
actually
taking
care
of
this
part
as
I
say
he's
not
here,
but
he's
working
on
it.
We
are
with
the
target.
So
this
it
done
open
source
by
July,
August
timeframe.
So
I
will
cloud
provider
interface.
That's
your
interface
between
companies
and
IBM
cloud,
and
you
know
typically,
as.
A
They
don't
need
to
really
do
anything
with
this
thing.
These
IPS
is
taking
care
of
them
on
this
side,
however,
you
know
having
it
open
source.
They
can
always.
You
know
trade,
any
bug
the
thing
out
there
or
a
new
feature
or
anything
all
right
and
the
folks
were
not
using
ideas,
but
they
are
interested
in
you
know
having
their
own
cubicle
store
on
IBM
cloud.
They
can
use
it.
It's
mainly
right
now.
The
bigger
thing
is
a
load
balancer.
So
it's
manages
to
what
you
like.
Using
keep
tip
slightly
and
two
versions
out
there.
D
A
A
So
this
is
the
new
architecture
I
mentioned
in
the
beginning
of
top
right.
So
earlier
the
entry
was
KCM,
the
cumulus
cloud
controller
magic,
cube,
respect
of
magic
ACM,
where
you
know
the
code
which
you
can
see
here.
This
is
actually
the
architecture
which
is
on
the
Kuban
is
done
only
thing
we
already
here
is
the
IBM
site,
but
in
three
everything
was
under
the
chief
controller
manager,
and
you
know
like.
A
You
figure
out
that
that's
not
the
way
to
post
everything
our
power
of
three,
so
the
only
cloud
connector
point
here,
as
you
can
see,
is
a
cloud
control
manager
in
free
people.
Imagine
and
every
node
also
has
certain
codes
to
get
the
information
about
node
and
zone,
so
that
thing
is
also
gone
in
this
new
architecture.
A
B
C
So
very
recently,
Guan
Yalu
and
his
team
in
China
put
together
very
quickly.
In
fact,
a
cluster
API
provider,
IBM
cloud
I'm
gonna-
will
have
a
link
in
here
I
think
for
it
I
encourage
you
all
look
at
it.
The
beauty
of
it
is
it's
still
small
I've
looked
at
the
code,
so
it's
easy
to
kind
of
poke
around
and
kind
of
learn.
There's
a
lot
of
stuff-
and
you
know
you'd-
be
surprised
how
much
of
the
core
kubernetes
concepts
make
their
way
into
this
code.
C
Senior
DS
controllers,
it's
even
used
in
the
new
customized
stuff,
which
is
how
you
can
instead
of
parameterizing
templates
into
your
and
your
you
know,
using
helm
scripts
or
what
have
you
or
using
your
config
files
that
you
can
use
customized
to
tweak
all
the
config
files,
but
what
what
the
cluster
API
provider
does?
This
whole
approach
is
everything
else
in
kubernetes
is
declarative
right.
C
You
can't
use
to
all
the
declarative
models
and
somebody
said
well,
you
know
what
would
it
be
great
if
we
had
a
declarative
model,
three
own,
to
declare
how
to
provision
your
cluster?
So
that's
what
it's
filling
in
and
you're,
seeing
all
the
cloud
providers
provide
their
their
their
cluster
API
provider
plugin,
and
so
it's
a
really
neat
structure.
You've
got
the
main
project.
Is
the
cluster
API,
Provost
or
API,
and
all
the
cluster
API
provider
projects
then
pull
a
lot
of
the
same
stuff
and
use
it.
C
So
it's
like
a
nice
plugin
extension
model
and
if
you
basically,
you
can
see
how
all
this
works,
you're
gonna
have.
Some
mammals
will
show
here
on
the
thing
of
what
is
the
declarative
describing
the
cluster,
but
you've
got
your
target
cluster
and
that's
the
the
cluster
we
intend
to
create
and
manage
so
you
can
be
creative
cluster,
an
IBM
cloud
or
AWS
ORS
or
a
Google
cloud
or
VMware.
C
B
C
There's
usually
either
mini
cube
or
kind,
but
you
start
out
with
that
little
small
cluster
and
from
there
you're
gonna
create
and
provision
onto
the
onto
the
actual
target
cluster.
And
then,
when
everything's
done,
the
bootstrap
cluster
is
going
to
go
away.
They
provide
their
own
community
tool,
the
cluster
CTL
and
basically
it
works
like
all
the
other
CTL
tools,
you're
used
to
a
couple
different
flags
and
provide
a
couple.
C
B
C
C
It's
not
it's
just
a
little
interesting
how
the
provider
code
uses
some
of
the
stuff
from
the
base
project,
so
it's
a
little
bit
kind
of
funky,
but
there's
not
too
many
see
Arnie's.
The
controllers
aren't
and
it's
kind
of
a
fun
project
to
learn
a
whole
bunch
about
how
kubernetes
works
in
the
small
thing
of
these
projects,
where
there's
not
a
lot
of
code,
so
I'm
going
to
encourage
you.
If
you
want
something
to
go,
look
at
look
at
it,
go
ahead.
D
C
So
you
get
the
boot
trip
cluster
running
and
there's
resources
there,
and
then
you
would
and
then
it
takes
care
of
moving
them
to
your
target
cluster,
in
addition
to
obviously
using
the
actuators
to
get
code
up
and
running
so
like
on
OpenStack,
it
uses
like
gopher
or
whatever,
to
start
up
the
VMs
and
non-abs.
It's
gonna
do
something
whatever
it
takes
to
provision
in
our
cloud
as.
D
C
Really
differentiated
right,
what
does
it
mean
to
get
something
up
and
running
on
AWS
versus?
What
does
it
mean
to
get
something
running
up
on
IBM
cloud
using
soft
Larry?
That's
what
have
you
so
so
that
that
code
is
kind
of
work
it's
differentiated,
but
with
that
that's
really
the
heart
and
soul
of
those
actuators.
There's
two
of
them
take
a
picture.
Go
ahead!
That's
where
you
can
find
all
the
fun
code
and
then.
B
C
C
C
B
C
C
Move
those
from
the
blue
strap
cluster
to
the
target
cluster,
get
those
up
and
running,
and
then
you
connect
the
last
step
destroyed
or
bootstrap
cluster
right
here.
Take
a
picture
of
this.
This
is
where
the
close
of
the
the
code,
the
quan
yas
team,
did,
is
out
there
really
cool
stuff
I,
encourage
you.
Let's
get
some
people
contributing
to
it
and
helping
out
even
myself
when
I
get
my
cookies
free
time.
That's
what
I
want
you.
C
B
C
Was
really
efficient,
Jason
pronounced
it
Razzie
folks
I
was
talking
to
they
said
they
pronounced
it.
Raizy
I,
don't
know
you
know
it's
one
of
the
two
I'd
better
go
with
Jason,
just
called
Razzie.
But
again
our
IBM
cloud
team
has
10,000
active
clusters
that
they
manage
right.
So
at
any
given
time,
there's
10,000
active
clusters.
They've
got
a
small
team.
They
are
obviously
very
aware
of
tools
like
Jay
and
some
what-have-you,
but
they
really
felt
like
they
needed
their
own
tools
to
make
this
work.
Their
tools
are
what's
key
with
them.
Obviously
designed
for.
B
C
C
B
C
A
Yeah
I
mean
this
is
quick,
I
guess
really
here,
but
we
just
talked
about
the
sake
right.
This
war,
the
Furies
and
IBM
cloud,
considering
managing
deploying
kind
of
things,
maintaining
you
know
being
part
of
the
sake
following
the
seek
you've
got
the
you
follow
the
munition
happening
like
a
as
an
ICP
side
and
multi-cloud
manager
from
the
cumulus
prospecting.
Also
other
ciencia
projects
right,
it's
another
projects
and
you
know
we
would
love
folks
to
contribute.
A
A
We'll
definitely,
you
know,
make
see
a
lot
more
things
you
know
do
better
in
the
future,
so
please
contribute
so
if
you'll,
just
just
reach
out
to
any
of
us
on
the
slack
channel
can
have
to
get
started,
and
we
have
now
the
sub
projects
where
people
see
things
as
very
explaining
that
if
you
I
provided
IBM
cloud
side
and
the
core
interface
things
which
is
coming
up
with
that,
thank
you,
but
we
still
have
you
know
we
still
have
30
more
minutes
in
this
marathon.
So
lot
of
time
for
discussions
in
Australia.
D
E
B
B
D
E
C
C
If
you
assume
things
are
immutable,
which
means
it's
kind
of
simplest,
if,
if
I
could
just
treat
the
machines
as
things
that
are
gonna
go
away
and
then
I'll
start
up
a
new
one
with
a
new
version
but
Kalen,
it's
sounding,
like
you've,
already
been
dealing
with
what
some
people
call
the
more
complex
case,
which
is
no
I,
will
make
it
allow
it
to
be
mutable
and
I
will
actually
be
doing
the
upgrades
in
place.
If
I
understood
you
correctly,.