►
Description
The vision of the vSphere Cloud Provider Interface (CPI) is to make VMware vSphere one of the best IaaS platforms out there to deploy BOSH and Cloud Foundry applications to. This is done by exposing several key features in vSphere, which brings a lot of value to users of BOSH. There have been many major improvements and feature additions since the vSphere CPI project ownership was transferred over to VMware at the end of 2017.
In this webinar, we briefly talk about BOSH and how it ties into the widely used vSphere CPI. We then do a deep dive into the architecture of the vSphere CPI, go over some of the key BOSH workflows, and look at what happens under the hood in the CPI. Then we give you a peek into what our roadmap looks like and present live demos.
Speakers: Yeshwant Babar, Neha Jain, Tushar Aggarwal, and Karim Elghamrawy
A
Hello,
everyone
and
welcome
to
today's
webinar,
we're
very,
very
excited
to
have
you
all
here.
My
name
is
Roslyn
I'm,
an
open
source
community
manager
here
at
VMware
and
I
will
be
written,
be
your
host
today.
This
is
a
webinar
series
where
we're
introducing
and
explaining
some
of
the
interesting
open
source
projects
that
stem
from
VMware
and
the
topic
of
today
is
a
project
called
the
vSphere
cloud
provider
interface.
We're
also
going
to
talk
about
Bosh,
which
ties
into
this.
If
you
have
any
questions
throughout
the
webinar,
please
add
them
to
the
Q&A.
A
You
will
see
a
button
down
down
below
in
your
zoom
window.
Please
use
that
for
any
questions
and
we
will
ask
and
answer
them
throughout
the
presentation.
Our
speakers
today
are
yeshwant
Babar,
Neha,
Jane,
Tushar,
Agarwal
and
Kareem
I'll,
gouge,
I,
hope,
I
pronounced
that
correctly
and
with
that
I'm
gonna
hand
it
over
to
Josh
one
who
will
start
the
presentation.
B
We're
gonna
be
talking
about
Bosh
and
VSEPR
today,
I'm,
a
barber
manager
here,
I,
am
we're
working
on
abortion.
Cpi
over
here
I
also
have
neha
the
char
and
curry.
With
me
for
the
engineers
working
on
abortion
CPI
for
the
day
we
are
going
to
be
talking
about
Bosch
and
what
viscosity
is
will
touch
upon
our
vision.
They
will
talk
about
all
the
things
that
you've
done
so
far
in
the
last
10
months.
They've.
B
We'll
take
a
look
at
how
the
director
watch
director
and
CPI
interact
with
each
other,
and
then
we
also
look
under
the
hood
to
see
what
actually
happens
underneath
we're
doing
keyboards,
and
we
also
have
bunch
of
demos
lined
up
hey
guys,
all
right.
So
what
is
Bosh
Bosh
is
a
very
popular
open
source
project
that
was
initially
created
to
deploy
cloud
foundry.
B
While
it
was
being
used
to
deploy
Cloud
Foundry,
they
realize
that
it's
Jedrek
enough
that
it
can
essentially
deploy
any
distributed
system
or
application.
That's
out
there
right
right
now,
Bosch
is
being
watched,
the
backbone
of
two
major
products
that
are
out
there,
which
is
PCF
and
PK
x,
which
is
sense
of
service
and
little
plum.
Pocket
watch
essentially
focuses
on
four
key
areas.
The
first
one
is.
It
really
simplifies
the
deployment
of
your
applications.
All
you
have
to
do.
B
Is
you
create
a
deployment
manifest
which
describes
how
you
want
what
your
application
will
look
like
and
just
provide
that
the
boss
and
boss
is
going
to
go
ahead
and
deploy
the
application
for
you?
Secondly,
it
also
manages
the
lifecycle
of
their
application.
It's
not
just
about
going
ahead
and
deploying
an
application
in
production,
but
you
know
there
are
so
many
things
that
happen
after
that.
You
want
to
release
patches.
B
You
want
really
is
a
great
the
latest
version
of
the
application
or
something
of
that
size
Bosch,
has
support
for
ruining
all
grades
which
in
America
application
with
minimum
per
node
on
that.
Thirdly,
it
also
focuses
on
monitoring.
It
can
wander
the
VMS
that
are
running
your
applications.
It
has
an
agent
in
every
VM
that
boss
creates
which
speaks
to
the
director,
and
if
the
VM
stops
responding,
boss
can
go
ahead
and
restart
those
VMs
for
you.
It
can
also
monitor
in
applications
which.
B
A
B
Engineering
Bosch
has
a
construct
called
as
releases,
which
essentially
packages
all
your
source
code
dependencies
and
tools
in
one
release,
so
it
can
really
help
you
understand
in
every
release,
one
of
the
one
of
the
exact
contents
for
each
release,
but
it
also
is
agnostic,
there's
a
CPI
layer
underneath
Bosch,
which
helps
Bosch
to
talk
a
different
cloud
providers
and
we're
going
to
talk
about
CPI.
You
know
in
the
next
slide
so.
B
What
is
a
CPI
CPI
is
an
abstraction
between
the
boss,
director
and
Amaya's.
It's
the
thing
that
makes
Bosch
really
cloud
agnostic
right
and
the
way
it
achieves
that
is
by
defining
previous
self
methods,
which
are
essentially
an
interface
right.
So
Bosch
like
an
application.
All
you
care
about
is
creating
a
VA
rating
at
this.
B
Attaching
business
to
OBM
and
the
boss
expects
that
each
of
the
cloud
provider
implement
this
interface
so,
for
example,
or
create
disc
and
vSphere
would
look
like
a
VMDK
but
totally
different,
for
let's
say
AWS
or
GCP,
so
every
is
has
their
own
implementation
for
all
these
methods
and
it's
the
CPI
layer
which
allows
vash
to
be.
You
know
truly
diagnostic,
the
CPI
for
each
highest.
That's
out
there.
B
So
there's
a
CPI
for
AWS,
see
behalf
of
Google
as
your
OpenStack
and,
of
course,
we
sphere
or
motion
is
at
CPI
is
owned
by
the
individual
companies
except
AWS
and
Olivia
sequester
managed
by
every
right
now,
just
something
to
throw
out
there.
70
75
percent
of
Cloud
Foundry
customers
are
currently
on
vSphere,
so
BCA
CPI
is
one
of
the
most
widely
you
see
PRA
in
the
Bosh
community.
B
So
this
is
what
our
vision
right.
Our
vision
is
to
really
make
vSphere
the
best
I
asked
for
Bosh
to
deploy
applications
right.
We
really
want
to
bring
all
the
goodness
of
be
sphere
to
the
boss
and
our
foundry
work,
and
there
are
two
parts
to
this.
The
first
part
is
be
sitting
around
for
a
really
long
time
and
there
are
lots
of
features
in
vSphere
that
love
our
admins
and
developers
really
love
right,
for
example,
stuff,
like
datastore
clusters
or
storage
DRS,
which
have
been
around
since
500
or
5
one.
B
So
we
are
focus
is
also
the
instance
of
feature
paddy,
be
for
all
these
old
emergency
pis
and
bring
all
these
very
valuable
features
to
the
Cynthia
and
Bosch
one.
So
that's
the
first
part.
The
second
part
is
we
share
with
every
new
release
is
releasing
really
cool
features,
and
we
want
to
add
these
features
as
soon
as
possible
to
the
Martian
sepia.
Well
as
well.
B
An
example
might
be
b/c
of
6.5
and
support
for
BM
encryption
so
that
you
can
encrypt
VMs
or
a
vm
level,
and
you
don't
have
to
encrypt
the
whole
data
store
or
you
know
BM,
for
if
you
guys
haven't
heard
about
it,
it
uses
instant
own
technology
or
methodology
to
spin
up
VMs
on
the
fly.
So
it's
twofold,
so
we
want
to
gain
feature
parity
with
all
of
Eastern
releases,
but
we
also
want
to
introduce
new
features
in
vSphere
that
come
out
as
soon
as
they're
available.
B
B
A
B
Whole
n
cross
n
matrix
between
the
VCO
11
16,
so
we
have
a
pretty
large
footprint
when
it
comes
to
Glantz,
but
I,
guess
that's
something
which
brings
a
lot
of
value
and
I.
Think
for
us
to
certify
that
came
to
work
with
certain
we
scale
versions.
It's
really
important
so
that
my
foot
was
in
actually
maintaining
this
my
clients,
testing
across
different
versions
and
fixing
bugs.
B
But
then
we
also
as
I,
spoke
earlier.
We
add
new
features
as
well,
while
adding
new
features.
We
really
keep
two
personals
in
mind.
Oh
one
persona
is
your
data
center
administrator,
which
is
your
VI
admin,
and
the
second
person
is
vSphere.
I'm.
Sorry,
sorry
developers
write
your
data
center.
Admin
is
responsible
for
managing
your
infrastructure
and,
if
someone's
deploying
Bosh
in
the
environment
environment,
they
really
need
some
sort
of
visibility
into
their
environment
as
well.
B
So
the
first
thing
that
we
address
was
reducing
the
number
of
permissions
that
Bosh
needs
to
deploy
on
top
of
this
year.
Prior
to
this,
release
botched
needed
some
really
extensive
permissions,
I,
don't
really
high
level.
You
know
we
asked
for
missions
at
the
data
center
and
the
root
level,
which
was
so
for
people
who
are
trying
to
describe
Cloud
Foundry
apart
in
a
multi-tenant
pattern
for
people
who
do
not
trust
their
developers
to
access
to
the
whole
V
Center.
B
A
B
Giving
him
some
visibility
right,
the
we
had
a
plug-in
sorry,
an
extension
for
we
in
the
V
scale,
UI
kabosh
what
it
did
was
or
it's
now,
let's
hear
your
developer
using
bar
should
be
spinning
up
a
bunch
of
VMs
and
all
these
VMs
are
not
pop
up
and
now
he's
here
and
your
benevolent
has
no
clue
what
these
VMs
are
and
if
they
don't
know
anything
about
it,
they
might
do
something
to
it
right.
So
if
they
start
just
powering
off
the
VM,
your
incur
some
downtime
applications
so
give
the
admin
some
visibility.
B
We
added
a
extension
which
will
let
the
admin
know
that
this
VM
was
not
well.
The
admin
starts
trying
to
make
any
changes
to
the
boss
different
VMs.
We
pop
up
a
warning
message:
saying
hey:
this
VM
was
not
deployed
by
you,
but
it
was
different
by
Bosch.
So
are
you
sure
you
want
to
delete
this
right
and,
and
things
like
that,
just
to
give
him
some
visibility
in
the
next
two
things.
We
are
also
planning
on
helping
admins
to
even
see
as
a
part
of
the
extension,
all
the
VMS
that
located
by
Bosch.
B
So
you
know
he
has
visibility
to
those
things
as
well,
along
with
that
now
addressing
all
the
other
developer
kind
of
stories
for
the
developer.
So
now
we've
added
support
for
storage
DRS.
You
know
so
plasterers,
we've
added
support
for
vm
groups
and
super
final
16
networks,
security
groups,
load,
balancers
and
you're,
going
to
see
demos
for
all
of
these
today
and
when
we
do
the
demos
we'll
dive
a
little
deeper
into
these
features
as
well.
Along
with
all
of
these
will
also
be
architected.
B
Along
with
contributions
to
the
CPI,
we've
also
been
making
some
contributions
to
Bosch
itself.
While
we
were
walking
on
PKS,
we
noticed
that
a
boss
can
manage
the
whole
lifecycle
of
your
storage
and
compute,
but
it
lacked
the
functionality
to
manage
the
lifecycle
of
a
phone
network
right.
So
watch
could
create
this
our
VMs
on
the
fly,
but
it
did
not
create
networks
on
the
fly
I'm
happy
to
announce
that
in
the
latest,
boss,
hoodies
the
functionality
that
we've
added
Bosch
is
now
available
in
the
in
alpha
stage.
B
D
D
D
D
D
D
D
D
D
D
Datastore
one
of
the
workflows
regarding
storage
box
here
is
stuff
that
we
have
later
on
in
the
webinar
another
point
that
I
would
like
to
highlight
these
of
network
resources
right
now
in
the
CPR.
We
assume
that
all
the
networks
are
connected
to
all
the
computer
resources.
We
do
not
base
our
selection
or
listen
on
the
basis
of
network.
D
Through
we
go
to
the
next
section,
I
want
to
describe
how
our
placement
engine
looks
like
so
a
policeman
in
this
is
basically
a
line
with
three
stages
in
you.
First
is
the
gathered
followed
by
filtering
and
then
sorry,
as
the
name
suggests,
gathering
is
just
basically
gathering
all
the
possible
and
clusters.
D
Center,
like
all
the
person
leader
suit,
but
only
the
closest
and
data
stores,
then
having
specified
next
stage
is
filtering,
but
the
step
starts
by
a
filtering
single
place
in
real-time.
It
considers
various
filters
in
succession
a
filters
can
be
is
non-free,
space
memory
and
accessibility.
All
these
filters
are
currently
not
exposed
externally
to
the
user,
and
the
ordering
in
which
filters
are
applied
is
not,
although,
theoretically,
all
the
filters
are
required
to
be
mutually
exclusive,
but
sometimes
in
some
cases
some
filters
are
not.
D
So
now
it's
time
for
the
demo,
so
in
a
demo,
I
will
probably
demonstrate
how
how
CPI
load
balances
the
viens
across
clusters.
So
I'll
do
a
very
simple
experiment
in
which
I
will
start
a
single
instance,
and
then
I
will
increase
the
number
of
instances
to
do,
and
we
will
see
how
those
things
get
friends.
So,
to
start
with,
we
go
through
our
V
Center,
as
we
can
see
in
the
center.
We
have
a
data
center
with
a
toaster
one
movie
sequencer
to
each
of
the
cluster
set
to
next.
D
D
Another
one
is
I
scuzzy
CL
2,
dl
0,
which
is
a
iSCSI
data
store
only
accessible
from
yesterday
and
the
last
we
are
going
to
use
NFS,
which
is
a
shared
data
store
which
is
commonly
accessible
from
both
the
clusters,
and
the
last
is
the
network
resources.
So
we
have
like
two
and
a
60
switch
1
and
a
6
which
is
2,
which
are
big
networks,
but
for
this
experiment
we
are
using,
which
is
a
standard
switch
in
this
way
in
England.
D
In
the
mean
time,
we
were
deploying
the
boss
director,
which
is
finished,
and
we
are
uploading
the
release
and
stem
cells.
We
have
good
ingredients.
Stem
cells
are
like
the
post
or
a
power
of
TM
which
are
used
to
create
all
of
the
fields
on
the
moors
later
on.
The
jobs
and
specific
things
are
installed
on
them,
so
they
behave
as
applications.
D
Next,
we
are
going
to
look
at
a
cloud
config
dot,
iam,
this
general
or
JSON
file,
is
basically
describes
your
cloud
infrastructure
on
vSphere
how
it
looks
to
the
boss.
So
we
have
our
D
start
by
defining
a
easy
availability
zone
which
has
two
clusters.
Again,
we
see
a
cluster
one
and
the
vehicle
start
as
we
defined.
There
is
a
persistent
disk
type.
There
are
two
distance
default,
a
large
and
both
are
stored
on
shared
datastore,
0
1.
D
Moving
on
to
the
network,
as
we
saw
earlier,
we
have
a
network
which
uses
standard
via
switch
via
network
and
the
name
of
the
network
is
default
and
the
VM
configuration
we
have
to
read
so
beyond
computation
before
Dennard.
Again,
both
of
these
complications
are
using
datastore
SEC
will
be
a0,
+,
CL
lazy.
Now
these
data
stores
are
different
from
the
data,
so
that
we
mentioned
earlier,
because
these.
D
Now
we
will
open
the
deployment
manifest
of
the
zookeeper
release
that
we
are
going
to
deploy,
and
if
you
look
in
here,
you
will
see
that
we
are
going
to
deploy
this
reason.
Availability
zone
z1
with
one
instance,
which
means
like
we
will
be
deploying
this
VM
on
z1,
which
has
two
clusters:
cluster
1
and
cluster-
do.
D
D
So
we
can
see
the
director,
vm
and,
and
the
main
the
ends
are
present
on
the
cluster
1
and
we'll
go
back
to
the
rana.
Manifest
will
increase
the
number
of
instances
to
do
while
keeping
everything
same,
and
then
we
again
deploy
the
release
and
we
create
the
missing
beans,
which
is
one
more
VM
and
s.
We
navigate
back
to
the
vSphere
and
see
their
second
VM
has
been
created
on
the
startin.
It
should
be
we're
hoping
for.
D
D
You
know
is
there:
we
are
initiating
a
BMI
locator
with
vlsm
criteria,
the
criteria
is
barrier
disk
and
various
requirement
me
moving
forward.
We
see
that
all
the
filters
are
being
applied
and
the
storages
are
being
rejected
or
selected
based
on
we
respectives
and
finally,
we
can
see.
Let
me
create
a
VM
with
on
the
tooth
and
cluster
and
that
a
duster.
D
D
Satisfaction
problem
in
scoring
in
the
end,
as
we
saw
all
the
relative
score
placements,
are
pick
one
by
one
and
until
the
beerbot
discretion
succeeds,
it
might
be
our
case
in
certain
scenarios
that
team
creation
fields,
you
can
lack
of
resources
and
you
might
want
to
retry
the
creation
of
another
and
then
it's
a
generalized
framework
which
allows
developers
to
create
their
custom
resource
pipelines,
for
instance,
the
piece
and
GPUs
like
we
did
for
selection
myself
with
this
I
will
complete
my
demo
and
one
thing
last
thing
before
I
go.
We
just
demonstrated
right
now.
A
D
B
B
A
B
B
Actually
start
out
testing
the
alpha
bills,
at
least
a
month
in
advance,
so
that
when
the
Beast
and
GA
is,
we
already
have
support
for
these
new
visa
versions.
When
out
on
the
table
it
G
is
so
yeah.
The
baby
of
bcog
is
the
version
we
have
support
for
that
less
than
less
and
we
for
the
most
part,
cool.
D
Also,
we
keep
the
badges
on
our
on
our
github
repository.
There
are
edges
ensuring
the
sting
of
pipelines
and
it
can
be
used
on
what
latest,
by
playing
something
and
which
pipelines
are
we
passing
so
as
I
show
you
the
screen,
we
are
supporting
life
cycles
without
seven
tests,
which
is
six
or
seven
Michigan,
and
within
a
sixty
two
or
three.
C
So
I'm
going
to
walk
you
through
some
of
these
key
features
of
CPI
like
we
improved
agustawestland,
syllables
I
talked
more
about
them
as
we
go
through
the
demo,
so,
let's
switch
over
to
the
demo
first,
since
demo
takes
a
little
bit
of
time
to
deploy.
This
is
a
recorded
demo
and
it
was
the
same
keys
with
the
shot
he
recorded
it
because
it
takes
time
to
do
more,
live
goodbye.
C
As
part
of
this
demo,
I've
already
deployed
my
director
I've
already
unleashing
a
suit
people
released
when
this
is
blood,
which
has
also
be
nothing
there.
Let's
take
a
look
at
the
cloud
campaign
as
part
of
this
demo.
The
first
thing
that
they're
gonna
look
at
is
B
and
blue,
as
you
can
see
in
this
stock
and
think
I
have
specified
of
the
mCP
IBM
Co
I'm,
using
open
network,
which
is
a
sexy
in
this
case,
to
deploy
this
case.
C
If
you
see
I've
specified
to
cervicals
seek
a
cool
one
and
Cynthia
move
to
as
part
of
this
cloud
country
and
a
security
group
colors
and
its
group
2
as
we
go
through
the
deployment.
I
talk
more
about
them,
and
the
last
thing
is,
if
you
can
lead
us
to
us,
for
this,
VM
type
I
specify
the
details
of
cluster
as
well
as
a
regular
leader,
sir.
So
in
this
one
deployment,
I'm
going
to
talk
about
the
envelope
as
well
as
server
boots
as
well
as
data
source
process,
all
three
of
them
together.
C
So
we
uploaded
our
top
10
fake,
and
now
we
start
applying
this
release.
Well,
the
waves.
Well,
the
deterministic
contains
less
sleeping
of
their
armies
and
right.
So
if
you
don't
trust
someone
and
if
you
go
and
look
at
Yemen
whole
scope,
they
know
we
are
also
listed
over
here.
The
way
you
can
use
VM
or
whole
scope
that
CP
is,
you
can
create
a
VM
glow,
along
with
that
it
can
create
whole
scope
and
define
VL
musu.
C
C
If
you
don't
have
an
existing
me,
loop
secured
we'll
go
ahead
and
feed
of
the
improve
and
add
V
M,
so
it
and
after
that,
you
can
create
a
VM
host
soul
ow
and
apply
that
your
3d
games
look
at
and
a
sexy,
which
is
the
opiate
Network
that
the
shot
was
talking
about
in
this
I've
created
two
syllables.
So
if
you
want
to
use
your
balance
of
each
other
than
sexy,
you
need
to
create
your
load.
Balancer
watch
ourselves
and
sell
before
you
start
using
them
with
the
CPI.
C
So
in
this
case,
Sita
actually
does
not
go
ahead
and
create
the
advance
analogous
Allah
or
some
oppose.
It
expects
them
to
be
present.
The
court
had
once
you've
created
yourself
good.
We
can
specify
them
in
your
camp
conflict
like
I,
showed
you
earlier
and
see.
If
I
will
go
ahead
and
add
vm's
to
those
circles,
your
server
pool
can
be
static
or
dynamic
and,
as
you
can
see,
Sipapu
want
a
static
and
see
people
to
is
dynamic.
C
C
C
Care
to
see
clear
beyondblue,
so
this
will
be
M
the
first
created
by
sepia
and
both
the
wheels
were
added
to
this
remember.
The
next
thing
is:
we
cannot
look
at.
He
does
no
class
authorities,
so
we
specified
by
beta
so
plaster
and
our
local
leaders
well
as
part
of
our
data,
so
property
in
cloud
can
pay
see.
They
will
use
the
placement
algorithm
that
Tricia
talked
about
and
will
pick
the
best
treatment
option
from
the
given
stories.
Theta.
So
cluster
are
the
leaders
that
are
provided
in
this
case.
We
expect
this.
C
We
planted
back
beta
so
plaster
that
Liam
provider.
So
if
you
click
at
one
of
those
beers
or
it
thicker,
the
leaders
of
person
correctly
you'll
see
both
the
VMS
were
created
inside
the
stochastic
raster,
which
has
lead
us
so
I
stuff
in
here
on
ds4,
so
are
good.
We
also
feed
it
inside
the
status
of
processor.
C
Now
we'll
go
and
look
at
one
of
the
vs
and
look
at
the
data
source
that
it
is
using.
If
you
click
about
the
am
whatever
be
their
source,
you
will
see
that
it
is
using
two
dealers
nose
once
I.
Still
see
CL
1
TS
one
and
another
one
is
0
and
D
s,
for
the
second
data
store,
belongs
to
the
Vita,
so
cluster,
that's
where
our
ephemeral
disk
and
will
be
imbalance
of
these.
The
first
data
store
is
where
the
position
is
for
this.
C
We
must
read
it
so
this
is
also
something
specifically
cloud
property
size
specified
and
my
best
type
that
might
this
should
be
patient
is
specific
leader.
So
let's
go
and
look
at
the
companies,
that's
under
the
same
until
you
get
it
under
this
type.
I
have
specified
if
he
does
not
to
be
extra
seasoning
on
ds1.
That's
why
my
process
and
then
start
treats
for
that
yeah
now
next
we're
gonna
look
at
these
are
balloons
where
the
beams
were
added,
so
we
have
two
circles
and
once
I
click
up
the
chief
specified
over
here.
C
So
if
you
come
and
look
at
NXT
and
go
to
see
their
new
bar,
which
is
the
membership
type
for
this
is
Scotty,
so
my
both
the
ends
were
added
to
the
stubble
the
nexus.
We
look
at
the
table
too.
This
is
the
dynamics
of
a
pool,
so
this
is
tied
with
and
stood
one.
So
there's
going
to
look
at
NS
group
one
and
then
let's
go
one.
We
see
that
they're.
The
two
members
were
added
to
the
same
scope.
C
So
if
you
go,
if
you
see
over
here
that
two
effective
members,
if
you
click
on
that,
we'll
see
what
the
VM
to
add
it
did,
the
secretary
also
that's
part
of
the
NSX
signification.
We
have
specified
a
separate
security
group
called
a
description,
so
your
wheels
will
also
be
added
to
this
technical
support.
The
games
are
added
to
this
secretary
as
well.
C
E
All
right,
so,
let's
switch
gear
a
little
bit
and
talk
about
another
open
source
contribution
that
we
did
not
just
to
the
CGI
but
pretty
much
to
the
Bosch
ecosystem,
and
that
is
the
Bosch
networking
feature.
What
do
I
mean
by
the
Bosch
networking
feature?
So
if
we
look
at
Bosch
just
to
give
some
context,
usually,
when
use
box
use
Bosch
to
automate
the
creation
of
VMs
or
the
creation
of
storage
on
any
is,
but
there
was
no
solution
for
automating
networking.
E
So
basically,
if
you
want
to
create
a
deployment
on
box
with
you
make,
what
you
usually
do
is
that
you
go
ahead
to
your
is
that
you're
using
and
then
you
manually
have
to
create
the
network,
and
then
you
use
Bosh
to
create
a
deployment
and
in
your
deployment
manifest,
and
your
cloud
confidently
reference
this
network
that
you
manually
created
before,
which
is
a
very
painful
process,
most
of
the
times.
Sometimes
you
just
want
a
network.
You
want
the
deployment
and
you
want.
E
Added-
and
this
feature
actually
requires
contribution
to
the
CPI
contribution
to
Bosch
contribution
to
the
Bosch
CLI,
so
it
was
a
major
contribution
that
way
we
provided,
and
we
basically
do
the
creation
of
the
network.
Now
we
do
the
deletions
of
the
network
automatically
we're
working
on
network
updates
as
well,
so
pretty
much
all
the
network
lifecycle
is
part
of
this
feature
that
we
are
implementing
okay.
So
what
is
a
solution
that
we
provide?
E
In
the
case
of
least
here
we
go
to
NSX
T,
and
then
we
create
a
bunch
of
routers
and
switch
create
the
network
based
on
the
configuration
that
the
user
provides
same
thing
with
the
late
network
and
of
course
we
we
needed
to
also
be
able
to
talk
to
this
new
CPI
methods
from
Bosch,
which
required
also
some
changes
in
Bosch
cos.
We
need
to
keep
track
of
the
state
of
the
network's
that
you
already
have.
You
need
to
be
able
to
list
the
network's
and
so
on
and
so
forth.
E
E
To
do
this,
we
need
to
create
a
t1
router
and
then
connect
this
deployment
to
the
stealin
router
and
then
connect
the
stealin
router
to
the
t0
router,
and,
basically
that
that
thing
over
here
that
is
connecting
all
the
VMS
is
the
switch
that
is
connecting
all
these
pins
and
we
assume
that
the
director
is
on
a
different
network
connecting
to
another
t1
router,
so
there's
connectivity
between
these
two
parts
over
there.
So,
let's
assume
we
want
to
use
bot
to
have
this
deployment.
E
So
if
we
wanted
to
do
that
with
Bosh,
that
does
not
support
managed
networks.
What
we
needed
to
do
is
we
basically
have
to
go
to
NSX
T,
the
n60
manager
and
manually
create
all
these
components
that
I
just
discussed.
So,
for
example,
the
first
thing
we
need
to
do
is
create
a
t1
router,
give
it
a
name
and
then
connect
this
t1
router
to
the
teacher
or
router
T's,
your
router.
E
E
Another
thing
that
we
need
to
do
for
the
t1
router
is
to
basically
enable
route
advertisements
so
that
the
t0
router
can
know
about
the
existence
of
feet.
One
router
and
all
IP
addresses
that
are
defined
there.
Once
we
create
the
t0
router
and
the
t1
router.
We
also
need
to
create
the
switch,
so
we
create
the
zookeeper
network
switch
and
then
another
thing
we
need
to
do
is
connect
this
switch
to
the
T
t1
router.
So
we
create
a
logical
ports
between
in
the
t1
router
and
connected
to
this
switch.
E
Okay,
so
with
that,
we
also
give
the
IP
address
that
we
discussed
before
the
172
IP
address
to
be
the
IP
address
of
the
port.
So
with
that,
all
the
components
that
we
need
for
the
network
is
already
there.
So
the
next
step
is
to
basically
just
go
to
the
deployment
manifest
and
deploy
the
zookeeper
deployment
that
we
needed
to
deploy.
Well,
we
have
to
reference
this
network
in
the
cockpit
we
have
to
tell
Maj
like
hey.
We
have
this
network
in
the
IOSH
right
now
and
I
want
to
create
a
deployment
on
that
network.
E
That
I
just
manually
created-
and
here
is
the
network
section
in
the
car.
So
this
is
the
config,
and
here
is
where
we
basically
define
the
network
that
we
created
and,
as
you
can
see,
if
you
look
at
the
cloud
properties
over
here,
we
basically
give
the
name
of
the
networks.
The
logical
switch
that
we
won
the
deployment
to
be
connected
to.
E
And
then,
after
that,
we
created
employment,
manifests
so
to
create
a
deployment
of
margining
two
things
you
need
to
have
a
cloud
con-5
already
in
bar,
and
then
you
have
to
create
a
deployment
manifest
for
the
actual
deployment
that
you
want
to
deploy.
So,
as
you
can
see
here
in
the
instance
of
groups,
the
zookeeper,
let's
go
back
so,
as
you
can
see
here
we
have
the
instance
group.
The
zookeeper
deployment
is
basically
referencing,
the
zookeeper
network
that
we
have
defined
in
the
cloud
conflict.
Now
then,
after
that,
we
just
deploy.
E
E
All
right,
so
everything
I
just
mentioned-
was
the
status
of
boss
before
our
feature
now
I'm
gonna
discuss
what
we
actually
changed
so
now
we
have
these
beer
and
in
a
16
that
doesn't
have
any
network
all
we
need
to
do.
In
this
case
we
don't
have
to
go
and
manually
create
the
networks
beforehand.
All
we
need
to
do
is
basically
just
put
everything
in
the
cloud
convict.
So,
let's,
let's
take
a
look
at
how
the
cloud
config
will
look
like
in
this
case.
E
So
here's
the
network
sections
in
the
cloud
con-5
after
our
change,
pretty
much
everything
is
the
same,
except
for
this
managed
part
here
where
we're
basically
telling
much
hey
Basch.
This
is
a
managed
network,
so
please
go
ahead
and
create
this
network
in
the
ayahs.
And
if
you
look
at
the
clock
properties
over
here,
it
kind
of
changed.
We
basically
need
to
tell
boss
some
information
about
the
nsx
t
network
that
we
have
so
we
we
give
the
edge
cluster
ID.
E
We
give
the
t0
router
an
e,
the
transport
zone
ID
and
a
switch
name
which
is
which
is
optional,
but
that's
pretty
much
all.
We
need
to
do.
We
didn't
go
ahead
and
read
the
network
beforehand.
We
just
changed
the
con
thing
and
then,
if
you
look
at
the
deployment
manifest
afterwards,
it's
exactly
the
same.
E
E
E
E
And
the
VMS
are
connected
to
that
logical
switch
so
pretty
much
the
same
state
that
we
had
before,
but
instead
of
manually
creating
the
networks
it
was
just
automatically
created
for
us.
So
this
part,
where
we
create
networks
and
delete
networks,
is
already
there
in
an
alpha
state
in
bosch,
we're
also
working
on
the
updated
network.
Part
if
you
already
have
any
system-
and
you
want
updated,
we're
working
on
that
as
well.
A
A
B
B
So
most
of
the
ISS
out
there
already
I've
been
finishing
in
vSphere.
We
did-
and
we
should
add
this
feature
for
vo
encryption
in
6.5
prior
to
that
you
did
have
encryption,
but
that
was
at
the
old
eNOS
11
right.
So
an
admin
data
store
with
the
new
feature.
You
can
add
support
for
that
one
so
that
you
can
encrypt
a
VM
and
disk
attached
to
that
VM
at
a
per
VM
level.
So
your
next
book
on
that
we
also
have
open
storage
policies.
B
A
B
Gpu
side
of
things
over
here,
we
try
to
just
make
sure
that
the
VM
lands
on
hosts
which
have
GPUs
attached
to
them.
So
we're
doing
that
book
as
well,
and
some
cosmetic
changes
in
the
visa.
Tell
you
why,
right
now,
when
you
create
a
VM,
all
the
VMS,
the
names
and
they
get
are
just
essentially
just
a
UID.
A
B
Do
not
really
denote
anything
in
terms
of
what
these
VMs
do,
so
we
will
try
to
add
a
prefix
or
somehow
try
to
change
the
name
of
these
VMs,
so
that
is
well.
They
didn't
know.
You
know
which
Institute
screwed
up
our
applications,
but
they
belong
to.
So
this
is
the
kind
of
stuff
planning
to
work
on
the
next
release.
B
B
A
I,
don't
see
any
more
questions
right
now,
so
I
would
like
to
thank
you
all.
Thanks
to
one
who's
been
presenting
today,
there's
been
really
really
useful.
A
lot
of
new
information
and
it's
great
to
see
where
the
be
spear
CPI
has
has
evolved
throughout
the
past
year
here,
and
it's
really
really
interesting
to
see
the
future
roadmap
of
this
project
here.
I
want
to
thank
everyone,
who's
been
attending
as
well.
This
will
be
posted
up
on
the
cloud
native
apps
YouTube
channel
shortly
here.