►
From YouTube: VMware Cloud PKS (formerly VKE) Lightboard SmartCluster
Description
VMware Smart Clusters. Run Kubernetes without Managing Servers or Clusters. VMware Smart Cluster automates the selection of compute resources that constantly optimize resource usage, provide high availability and reduce cost.
A
A
Welcome
to
this
video,
my
name
is
Bosque
Saleh
I'm,
a
product
marketing
manager
within
VMware.
Today
we
are
going
to
talk
about
via
maccready's
engine
and
smart
clusters
via
McBurney's
engineer,
the
SAS
offering
from
VMware
that
allows
developers
and
IT
operators
to
consume
kubernetes
clusters
as
a
service.
Now,
specifically
via
my
kubernetes
engine,
offers
smart
clusters.
So
what
are
these
smart
alerts?
Does?
Let's
talk
about
them
now?
A
Vma,
kubernetes,
engine
or
vke
runs
on
a
public
cloud.
It's
multi
cloud
enabled
so
when
a
developer
comes
in
and
wants
to
deploy
applications
of
taking
care
of
a
kubernetes
cluster.
The
first
thing
that
they
do
is
they,
you
know,
go
to
the
vke
api
or
UI
and
ask
for
a
cluster,
and
that's
a
smart
cluster
that
we
are
talking
about
now.
What
do
we
really
mean
by
that?
As
developers
start
coming
in
and
they
go
to
the
vke
API
or
a
UI?
They
are
going
to
request
for
a
cluster.
A
A
Now,
once
you
have
the
region
and
the
name
of
the
cluster
vmi
kubernetes
engine
takes
over
at
this
point,
it
really
doesn't
want
the
developer
to
worry
about
what
would
be
the
cluster
size.
How
many
worker
nodes
do
you
need?
How
many
master
nodes
do
you
need,
etc?
It
just
takes
care
of
everything
from
that
point
onward,
including
what
instance
to
provision
to
make
optimal
use
of
resources.
So
at
that
point,
the,
depending
on
the
region
via
kubernetes
engine,
will
talk
to
that
public
cloud
is
endpoint
and
select
the
specific
region.
A
A
A
Now
that's
as
simple
as
that
BMA
comedy's
angel
will
then
give
back
the
developer
the
API
endpoint,
to
talk
to
this
cluster.
Now
at
the
start
of
a
smart
cluster,
they
are
no
working
notes
provisioned.
This
is
intentional,
and
the
reason
for
this
is
as
soon
as
a
cluster
is
created.
We
don't
start
provisioning
worker
nodes
until
a
debauch
load
is
deployed.
A
Now
it
will
provision
worker
nodes
as
needed,
which
means,
let's
say
in
my
app,
only
needs
a
single
worker
node
that
has
a
specific
CPU
and
memory
limit.
Depending
upon
the
yam
l
file
that
you
specified
for
that
application,
it
detects
the
worker
instances
needed
to
provision,
and
it
only
will
provision
that
many
worker
instances.
Now,
let's
say
another
worker,
you
know
application
comes
in.
That
needs
a
little
bit
more
space
than
the
work
on
that.
A
So
that
way,
application
developers
need
not
worry
about
the
size
of
the
cluster.
What
operating
system
do
I
need?
How
many
worker
nodes
do
I
need?
How
many
master
nodes
do
I
need
do?
I
have
high
availability
in
involves
within
VMware
kubernetes
engine
or
the
cluster
that
I'm
deploying.
All
of
that
is
taken
care
of
now.
Let's
say
the
developers
start
deploying
applications.
We
have
the
worker
nodes
and
master
nodes
working
for
them.
Then
you
know
as
more
and
more
applications
get
deployed.
We
keep
increasing
the
worker
node.
A
A
It
will
keep
monitoring
the
k8
scheduler
for
the
instances
that
are
deployed
the
port's
that
are
deployed
within
each
worker
node
every
10
minutes,
or
so
it
will
keep
checking
whether
the
load
across
all
these
worker
nodes
are
equally
distributed.
If
not
what
we
america
vanities
engine
will
do,
is
it
will
reprovision
the
pods
to
another
worker
node
and
will
reach
or
fill
the
install
a
specific
worker
node
that
way
the
cluster
is
automatically
scaling
out
and
scaling
in,
depending
upon
the
workload
that
the
cluster
has,
and
all
of
this
is
transparent
to
the
developer.
A
The
developer
is
not
really
thinking
about
any
of
this.
This
is
all
automatically
done
for
the
developer
by
vke.
Now,
once
the
apps
are
deployed,
we
also
are
scaling
them
accordingly.
Now
another
instance
of
kubernetes
version
comes
up
and
the
developer
wants
to
upgrade
their
cluster
to
that
specific
version.
So
we
want
to
now
upgrade.
A
We
can
eat
will
actually
take
care
of
the
upgrade
for
the
developer
in
a
rolling
fashion
for
all
the
moisture
and
work
notes
deployed
for
that
cluster.
What
we
can
eat
will
do
at
this
point
in
time
is:
let's
say
you
have
of
k8s
cluster,
that
has
version
1.9
for
each
of
the
nodes
in
it,
and
the
developer
wants
to
upgrade
these
to
1.10.
A
We
will
start
upgrading
or
applying
those
upgrades
in
a
rolling
fashion.
It
will
first
take
out
this
master
node
and
create
a
new
one
that
is
1.10
version,
join
that
work
in
a
master
node
to
the
cluster,
make
sure
that
all
the
clusters
are
in
you
know
in
a
good
state
and
then
move
on
to
the
second
master.
Similarly,
it
will
once
this
master
is
active.
It
will
delete
this
one
and
upgrade
this
one
to
a
newer
version
at
all
points.
A
Now
at
any
point
in
time,
let's
say
the
upgrade
fails
for
some
reason:
vke
will
actually
roll
back
all
the
instances.
True
the
original
version
of
the
Kuban
IDs
cluster
and
restore
the
backup.
You
know
in
case
something
was
lost.
So
all
the
upgrades
and
everything
is
taken
care
within
the
small
cluster.