►
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: VMware SIG - Fabio Rapposelli & Steve Wong, VMware
Join the SIG VMware introduction session to learn our mission, recent accomplishments and discuss future work. We will also focus on how new contributors can get involved in the SIG. Kubernetes has the concept of a Cloud Provider, which is a standardized module which allows Kubernetes to run on various platforms which might have different implementations of networking, storage, and node management. Kubernetes is in the process of moving to a new “out of tree” cloud manager architecture to allow cloud providers to have independent feature and patch release cycles. Learn how the vSphere cloud provider is evolving to meet this new model. This session will outline how to get started when you deploy Kubernetes on vSphere infrastructure, on-prem or in a public cloud. Material covered is applicable to kubeadm deployments, and when using Kubernetes distributions on vSphere infrastructure.
To Learn More: https://sched.co/GrbA
A
So
welcome
everybody.
If
you
want
to
move
up,
it
looks
like
we
didn't
come
close
to
filling
in
the
room,
but
it's
up
to
you
yeah.
Why
don't
you
flip
it
so
I'm
Matt
guy
he's
the
other
guy.
We're
co-chairs
of
the
vmware
sig
were
engineers
with
the
cloud
native
apps
business
unit
at
vmware,
I've
been
working
on
kubernetes
before
I
joined
VMware,
so
I
became
active
in
the
kubernetes
storage
sake
back
in
2015
Fabio,
so.
B
I've
been
a
VMware
for,
like
6
years
now,
I've
been
mostly
involved
with
open
source
projects
when
it
was
at
VMware,
so
I
started
working
and
everything
had
to
deal
with
containers
back
in
2014,
so
docker
vagrant,
more
recently
kubernetes,
and
so,
if
you'd
seen
fused,
kubernetes
or
vagrant
or
docker
on
any
of
our
VM
or
platforms.
There's
a
good
chance
that
you
saw
the
code,
I
wrote
during
my
years
at
DM
were
so
and
now
we're
steve
and
I
were
chairing
this
SIG
vmware.
B
A
A
You
know
there's
a
concept
in
kubernetes
of
an
abstraction
layer
called
a
cloud
provider
that,
in
theory
gives
you
a
situation
where
you
can
be
portable
to
multiple
clouds,
both
multiple
public
clouds
and
on-prem,
and
have
an
app
be
unable
to
tell
where
it's
running
and
just
work
effortlessly
in
all
these
places.
Now
I
wish
I
could
say,
that's
just
gonna
magically
happen,
but
you
have
to
build
that
abstraction
layer
and
as
kubernetes
adds
features
you
know
you
want
to
take
advantage
of
the
features
without
breaking
that
great
abstraction
layer.
A
How
you
join
the
first
step
is
to
join
this
Google
group.
Some
of
these
things
are
gated,
where
you
have
to
join
the
group
in
two
to
be
able
to
edit
documents
in
some
cases
see
the
documents.
Well.
The
reason
for
this
gatekeeping
is
just
to
protect
you
from
spam
and
there
have
been
occasionally
unfortunate
incidents
in
not
in
our
group,
but
I've
heard
in
others
of
trolls
going
in
there
and
just
you
know,
putting
essentially
graffiti
and
group
planning
documents,
and
we
don't
want
that
going
on.
So
you
do
join
the
group.
A
You
can
join
the
slack
as
well.
I
think
your
first
step
I
recommend
the
mailing
list
just
because
it's
easier
to
search
for
old
stuff
slack
is
great
for,
inter
give
like
I'm
it's
2:00
in
the
morning.
I
might
hope
somebody's
online
and
help
me
out,
but
you
know
beyond
a
point:
there
can
be
so
much
traffic
in
there
that
that
isn't
easily
searched
that,
for
many
purposes
going
in
there
in
the
mailing
list
is
a
better
experience
in
my
opinion,
but
you're
welcome
to
do
either.
So
that's
it
next.
A
B
So
right
now
the
sink
is
being
created
earlier
in
2019,
so
we're
thinking
we're
first
meeting
was
around
March
this
year.
So
it's
a
pretty
new
sig,
obviously
we're
focusing
on
everything
that
has
to
do
with
VMware.
So,
as
you
can
imagine
you
know,
most
of
our
efforts
are
trying
to
get
to
make
kubernetes
run
best
on
vSphere,
which
is
I.
Guess
or
you
know
the
largest.
B
You
know
install
base
out
there
when
it
comes
to
VMware
platforms.
Are
we
also
looking
after
things
like
mini
cube
for
fusion
and
workstation,
so
of
desktop
hypervisors
and
so
on,
particularly
we're
focusing
on
this
for
projects
right
now,
so
we're
sponsoring
these
four
projects?
Club
provider
and
CSI
are
kind
of
going
handyman
and
I'll
explain
that
shortly
and
now
we're
we're
joined
the
cluster
API
working
group
recently
that
was
recently
formed.
B
There's
a
lot
of
Basel
run
cluster
API
and
there's
actually
a
session
on
that
later
today,
which
I
will
recommend
you
to
join
and
we'll
have
a
slide
at
the
end,
with
all
the
details,
so
speaking
of
auditory
components
and
CSI
and
a
cloud
provider.
So
this
is
a
like
a
brief
overview
of,
what's
going
on
inside
kubernetes,
so
kubernetes
originally
originally
built
as
like
a
single
repo
standalone
project
where
you
had,
everything
was
checked
inside
the
same
repo,
which
is
kubernetes
kubernetes,
and
this
is
included
code.
B
That
is,
you
know,
vendor
dependent,
so
it's
Club
provider
dependent.
So
in
our
case
we're
talking
about
the
vSphere
law
provider.
There's,
obviously
you
know
it.
Yes,
there
is
OpenStack,
there's
many
of
those
club
providers
that
started
early
on
in
the
kubernetes
history
that
have
been
part
of
the
original
kubernetes
core
repo.
This
is
obviously
not
something
I
can
scale.
You
know
as
the
kubernetes
project
ones,
so
around
cubanelles
1.6
timeframe.
B
It
was
a
proposal
to
have
the
club
providers
being
removed
from
from
the
core
codebase
into
their
own
discrete
repos,
and
this
was
the
was
how
in
the
club
provider
interface
was
born.
So
originally
we
had
different.
You
can
see
from
the
from
the
schema
here
we
had.
We
used
to
have
cubelet
running
the
coal
provider
code.
Now
we
have
a
something
called
cloth.
B
Controlling
manager
that
is
running
and
a
clock
controlling
manager
is
something
that
is
no
longer
part
of
the
core
kubernetes
repo
and
in
our
case,
what
we
had
and
I
just
flip
him
back
one
sec.
We
used
to
have
a
club
provider
that
was
taking
care
just
not
just
of
the
compute
part,
but
also
the
storage
part,
and
this
is
because
it
was
you
know,
growing
organically
inside
the
project
and
there's
features
were
added
and
you
interface
matters
we're
at
it.
B
We
were
just
adding
on
top
of
that
when
this
auditory
move,
you
know
over
time,
was
refined.
There
was
a
decision
in
splitting
the
two
concerns.
We
can
two
separate
efforts.
One
is
cloud
provider
for
everything
has
to
do
with
compute
and
then
there's
CSI,
which
is
a
different
effort
related
to
storage,
so
we're
basically
speeding
out
from
what
is
the
current
entry
implementation
into
two
separate,
discrete
auditory
components.
One
of
them
is
CSI,
and
one
of
them
is
the
call
provider.
B
B
You
have
your
application,
you
have
your
definition
and
you
can
just
deploy
it
at
samyama
on
top
of
any
cloud
provider
and
expect
the
club
provider
to
perform
those
function,
that
you
request
them
in
your
yellow
file,
transparently,
because
there's
a
there's
a
driver,
there's
a
club
provider
that
performs
that
for
you,
so
it
was
seen
before
club
provider
used
to
handle
part
of
the
volume
functions.
Now
it's
no
longer
the
case.
So
there's
no
longer
storage,
any
storage
functions
inside
the
club
provider
interface.
B
Everything
is
handled
by
CSI
it
also
animals,
so
Club
provider
also
handles
the
availability
still
and
support.
So
is
again:
you
can
specify
availability
zones
inside
your,
your
pods
pack
and
the
club
provider
will
handle
that
for
you
at
the
club
provider
layer,
so
the
status
of
cloud
providers,
external
call
providers
in
kubernetes
has
been
office
since
one
six,
one
six
and
it's
being
moved
to
bed
at
some
point,
I
remember
exactly
what
version
moved
it
to
better
and
it's
currently
still
better
as
of
113,
which
was
just
released.
B
B
When
it
comes
to
storage,
that
was
mentioned,
the
CSI
provider,
CSI
is
yet
another
interesting
effort
and
steve
has
been
part
of
this
for
for
some
time.
So,
if
you
want
to
you
know,
add
some
some
flavor
to
my
explanation,
please
feel
free
to
chime
in
CSI
was
basically
created
with
the
intent
of
abstracting,
completely
abstracting
storage
functionalities
from
any
container
Orchestrator
that
will
implement
the
CSI
interface.
So
it's
not
something
that
is
strictly
connected
with
kubernetes
kubernetes
is
implementing
CSI
and
so
I.
As
far
as
I
know,
mesos
is
also
implementing
see.
A
Even
within
kubernetes,
you
could
have
a
back-end
support.
Multiple
independent
kubernetes
clusters
and
I
think
kubernetes
embraced
this
because
they
wanted
to
move
certain
parts
like
the
cloud
providers
and
the
storage
plugins
out
of
tree.
One
of
the
real
reasons
that
maybe
not
might
not
be
obvious
for
why
they
want
to
get
it
out
of
the
comments
or
space
is
that
you
can
have
scenarios
where
they
do
a
kubernetes
release,
which
they
do
quarterly.
A
A
bug
comes
out
in
one
storage
provider
out
of
30
of
them
and
it
needs
a
patch,
but
you
have
to
rebuild
and
read
retest
all
of
the
coburn,
a
suite
in
order
to
get
that
out
the
door
and
maybe,
in
the
meantime,
some
other
pull
requests
have
been
merged.
So
it
basically
ties
the
release
cycle
where
you
couldn't
patch
a
storage
related
thing
that
only
affected
one
of
30
storage
plugins
without
releasing
the
whole
of
kubernetes,
and
that
just
is
not
optimal.
So
by
pushing
it
out
of
the
tree.
A
B
And
the
same,
you
know
the
same.
The
fact
you
have
independent
release
trains
for
all
these
components.
So
it's
obviously
the
same
thing
that
is
happening
with
a
cloud
provider
that
I
mention
before.
So
everybody
is
on
a
different
release:
schedule
where
we
have
a
single
interface,
that
is
your
contract
with
the
kubernetes
and
kubernetes
eventually
will
run
on
its
own
schedule,
while
all
the
various
components
will
run
on
their
own
separate
and
the
the
the
interface
the
API
is
gonna,
be
the
only
thing
connecting
the
two
so
the
status
of
CSI
in
kubernetes.
B
It's
actually
very
interesting
because
the
CSI
spec
just
recently
reached
100.
So
there
were
the
version
of
the
CSI
specification,
rich
maturity,
so
the
first
GA
or
stable
version,
and
that
CSI
spec
was
actually
added
to
and
supported
and
move
to,
GA
stable,
also
in
kubernetes
113.
So
kubernetes
113
is
actually
including
CSI
as
a
stable
feature,
and
is
it
my
understanding
that
this
is
going
to
be
the
suggested
way
of
handling
storage
volumes
inside
kubernetes?
From
now
on,
from
113
on
yeah.
A
So
they're
suggesting
it
is
the
road
forward.
However,
I'd
like
to
point
out
that
kubernetes
has
deprecation
rules.
So
at
the
current
point
in
time,
there's
no
gun
to
your
head,
saying
you
just
because
kubernetes
got
released
last
week,
you
have
to
switch
over
to
it.
I'd
anticipate
the
entry
plug-in
is
going
to
be
alive
and
well
for
at
least
a
year
right,
Fabio
yeah.
B
A
A
Haven't
been
in
any
enterprise
space
for
a
while
I
know
that
if
I
were
in
the
shoes
of
somebody
using
it,
I
would
jump
on
some
new
storage
plug-in
release
a
week
after
it
came
out,
but
that's
up
to
you,
you
have
the
option
of
doing
it
either
way.
There
are
some
storage
providers
that
maybe
don't
have
the
option
because
they
sort
of
have
lockdown
new
entry,
storage,
plugins,
so
you'll
find
some
solutions
out.
A
There
might
have
to
go
to
CSI
ASAP,
but
on
the
VMware
platform
you're,
not
looking
at
that
situation,
you're
going
to
have
a
pretty
wide
window
where
you
either
one
will
work.
I,
don't
know
that
it's
true
that
you
can
run
them
both
simultaneously
but
then
again,
I.
Don't
know
why
you'd
want
to
either
right.
Fabio
say
it
again:
I
don't
know
that
it's
possible
to
run
the
entry
and
CSI
simultaneously
in
the
same
cluster,
but
I'm
not
sure
why
you'd
want
to,
but
so.
B
You
can
actually
run
kubernetes
with
a
mixed
version,
so
you
can
run
kubernetes
with
an
external
cloud
provider
while
maintaining
the
entry
cloud
provider
for
storage,
there's
a
flag
in
cubelet.
You
can
specify
to
do
that
or
you
can
do
something
like
the
opposite,
like
using
CSI,
but
rely
on
the
entry
for
everything
that
has
to
do
with
compute.
It's
kind
of
you
know.
It's
probably
not
suggested,
but
I
understand
that,
but
I
didn't.
A
Even
do
it?
Okay,
there
you
have
it
one
thing:
I
will
caution
you
with
these
changes
is
beware.
This
applies
to
all
of
kubernetes.
They
should
put
you
know,
sell-by
dates
on
blog
articles
and
as
these
kind
of
architectural
changes
take
place,
google
search
can
readily
find
you
blog
articles
on.
This
is
the
way
you
should
do
this
that
are
a
year
or
two
years
old
and
they're,
getting
somewhat
treacherous.
A
So,
if
you're,
a
newbie
to
using
kubernetes
on
the
vSphere
platform,
just
be
on
the
lookout
and
be
a
little
skeptical,
if
that
how-to
is
more
than
a
year
old
in
some
cases
the
sell-by
on
these
are
it's
probably
ought
to
be
six
months,
but
that
isn't
unique
to
this
I
think
it's
you
that
that's
a
universal
thing
anytime.
You
have
a
project
with
as
much
velocity
as
kubernetes
with
you
know
the
the
thousands
of
just
commits
per
month.
This
is
what
happens
and.
B
You
know
when
it
comes
to
the
status
of
these
two
projects
inside
of
cig,
we're
actually
happy
to
announce
that
we
just
tagged
the
very
first
version
of
both
the
CSI
provider
for
Reece
fear
and
the
club
out
of
tree
club
provider
plugin
yesterday.
So
we
have
a
zero
one,
zero
version
out,
which
is
the
first
one.
It's
just
kind
of
you
know
decently
complete
as
an
implementation.
So
we're
you
know
suggesting
you
if
you're
running
kubernetes
on
vSphere,
you
want
to
try
it
out
you're
welcome
to.
B
Feedback
is
extremely
welcomed
like
if
you're,
not
if
you
cannot
like
commit
to
the
time
to
actually
help
out
in
the
project,
even
if
you're
trying
out
I
mean
this
is
the
future
for
a
you
know,
running
kubernetes
on
vSphere
when
it
comes
to
you
know,
platform
integration.
So
this
is
what
you
should
be
looking
at
if
you're
running
kubernetes
on
vSphere
today,
obviously
we're
not
suggesting
to
run
this
in
production
today.
B
But
if
you
have,
you
know,
Devon
test
clusters,
or
just
you
know
your
sandbox,
where
you
play
out
with
with
kubernetes
in
vSphere,
you
know
just
try
it
out,
even
just
giving
us.
You
know
some
feedback
back,
you
know
creating
a
feedback
loop
where
people
are
actually
using
this
and
a
day-to-day
basis
is
extremely
important.
You
know
for
an
effort
of
for
the
status
of
this
project,
yeah.
A
So,
if
you're
an
end-user
definitely
come
to
that
meeting,
if
you
have
specialized
interests,
you
can
drill
down,
so
the
cloud
provider
is
that
abstraction
layer
for
running
on
the
vSphere
hypervisor,
so
I'd
say
that
those
meetings
tend
to
be
more
developer
focused
but,
like
all
of
kubernetes,
there's
nothing
hidden
here,
it's
all
open
source
you're
welcome
to
come
and
get
involved
in
those
meetings
as
well.
The
cluster
API
thing
is
something
of
specialized
interest.
A
Where
you
know
this
is
what
composes,
VMs
and
kind
of
bootstraps
them
up
as
you're
deploying
kubernetes
on
top
of
vSphere
infrastructure.
It
might
be
of
great
interest
if
you
there
are
commercial
distributions
that
might
manage
this
kind
of
thing
already
as
part
of
a
commercial
distribution.
However,
if
you
are
building
a
commercial
distribution
or
you're
choosing
to
go
with
the
pure
upstream
kubernetes
out
of
open
source
and
kind
of
roll,
your
own
onto
your
own
data
centers,
this
stuff
could
be
of
great
value
to
you.
A
So
it's
an
API
for
driving
the
the
bootstrap
and
afterward
you
might
want
to
see
Doug
here
and
if
Lok's
in
the
audience,
we
have
a
few
people,
I
notice
in
the
audience
who
are
very
active
on
that
project,
so
hang
around
and
meet
them
after
this
session
breaks
up
on
github.
This
is
where
the
source
lives
you're
free
to
do
the
open
source
convention
of
logging
issues.
A
B
Just
want
to
highlight
that
we
have
three
meanings:
we're
currently
running,
so
we
have
our
bi-weekly
meaning,
which
is
like
the
main
SiC
meaning
bi-weekly
on
Thursday,
it's
right
after
the
community
meeting.
So
if
you
feel
a
tanning
community
meeting-
and
you
have
another
out
of
the
spare
forty-five
minutes,
join
our
our
call
when
you
join
the
Google
group,
that
was
on
the
first
slide,
you
will
have
access
to
the
meeting
agendas
and
all
the
recordings
that
you
can
see
in
the
link
here.
B
So
there's
a
if
you
have
any
interest
of
discussing
any
time
know
topics
that
has
to
do
with
VMware
and
kubernetes.
Just
you
can
log
in
there
just
add
your
name
to
the
discussion
list,
anything
to
the
agenda
put
whatever
you
won
there
and
we'll
make
sure
to
bring
that
up.
Yeah.
The
next
meeting,
yeah.
B
Also,
if
you
have
anything
you
want
to
demo
we're
also
very
eager
to
see
you
know,
people
running
VMware
and
kubernetes
or
maybe
using
you
know,
summoner
tools
to
manage.
You
know
life
cycle,
the
the
VMS
when
it
comes
to
kubernetes,
were
you
know,
we'd
love
to
have
you
do
a
demo?
We
actually
have.
If
you
look
at
the
recordings,
we
have
some
demos
usually
running
around.
You
know
all
the
projects
that
were
in
you
know
we're
sponsoring
just
want
to
mention
that
we
had.
B
B
B
And
so
we
we
didn't
focus
on
cluster
API
on
this
presentation,
and
this
is
just
because
there
is
a
very
nice
presentation
on
cluster
API
on
vSphere
in
AWS
coming
up
today,
so
I
just
wanna.
You
know
besides
the
the
list.
Here
we
wanted
to
give
you
a
couple
of
pointers
on
sessions
you
attend
this
week.
So
today
we
have
a
3:40
p.m.
we
have
use
of
cluster
API.
To
deploy
clusters
on
pram
and
in
public
law
and
on
pram
is
VMware.
B
A
Then,
tomorrow
there's
the
cloud
provider
sig.
So
this
is
the
overall
concept
of
implementing
this
abstraction
layer
so
that
your
kubernetes
workloads
can't
tell
what
cloud
they're
on.
So
what
you'd
expect
at
that
meeting
is
every
cloud
provider
shows
up
so,
if
you're
an
end-user
doing
multi
cloud
or
maybe
you're
a
developer,
trying
to
do
a
kubernetes
distro
yourself
that
could
be
hosted
anywhere.
That
isn't,
you
know
locked
in
to
some
one
particular
location
or
kind
of
technology
that
would
be
of
interest
to
you.
A
We've
got
the
intro
to
the
cloud
provider
and
then
the
deep
dive.
Then.
Finally,
the
VMware
cig
has
a
deep
dive.
That's
going
to
cover
advanced
features
like
using
availability
zones
that
are
supported
in
the
vSphere
hypervisor
through
DRS,
the
Numa
support,
features
and
advanced
resource
management
that
can
be
performed
using
vSphere
as
an
adjunct
to
what's
already
built
into
the
stack
of
kubernetes,
your
container,
runtime
and
Linux.
So
all
of
those
have
some
forms
of
resource
governance.
A
If
you
elect
to
use
them
where
you
would
deploy
pods
that
disclose
minimal
requirements
and
maximal
cut
off
limits
for
things
like
CPU
cores
memory,
but
it
turns
out
that,
because
some
hypervisors,
like
vSphere,
have
been
around
with
a
10-year
history
of
covering
those
kinds
of
issues
that
they
have
advanced
features
that
you,
what
that
would
allow
you
to
potentially
benefit
from
engaging
both
of
those
simultaneously.
So
this
deep
dive
is
going
to
get
into
that
subject
and.
A
A
Okay,
we
were
asked
if
we'd
like
to
have
sega
leadership
outside
of
vmware
right
now,
both
Fabiano
ir
vmware
employees,
but
we
actually
would
like
to
see
a
user
voice
there.
So
if
you're,
a
user,
I'm
not
sure
I
think,
potentially
even
another
vendor
that
wants
to
get
involved
in
actively
lead
or
drive
the
vmware
saying
we're
open
to
that.
So
you
know,
I
think,
we're
we
don't
want
this
to
be
vendor
dominated.
A
B
I
mean
where
everything
we're
running
so
far
has
been.
You
know
extremely
transparent,
and
you
know
we're
trying
not
to
be
too
vendor
oriented,
obviously
also
because
most
of
the
people
involved
are
actually
working
on
open
source
hundred
percent
of
their
time.
So
there's
no
like
company
involvement
when
you
know
when
it
comes
to
that,
but
we're
actually
actively
looking
for
somebody
not
from
vmware
to
step
up
as
a
sig
leader
and.
A
I
might
get
killed,
but
I'm,
not
in
VMware
Sales
I'm.
An
engineer
I
might
get
killed
for
saying
this,
but
my
mission
and
I've
been
told
this
in
my
job
description
you
can
stand
even
though
VMware
has
some
kubernetes
distros.
My
job
is
to
support
them
all.
So
if
you
want
a
rep
run,
Red
Hat
open
shift
on
top
of
vSphere
we're
doing
nothing
in
this
cloud
provider.
You
know
it's
our
job
to
make
that
work
and
work
well
same
thing
for
mesosphere
DCOs
or
any
other
kubernetes
distro.
A
B
So
just
review
the
questions
any
and
it's
about
you
know
implementing
load,
balancer
support
on
a
cloud
provider.
That's
an
interesting
question.
Actually
we
have
an
issue
open
to
explore
options
in
implementing
the
load
balancer
on
vSphere.
There's
a
lot
of
challenges
to
do
that,
obviously,
because
there's
no
load
balancer
facility
and
vSphere,
but,
for
example,
there
is
when
you
run
vSphere
n-double,
yes,
because
there's
LB
that
works
with
yes,
so
there's!
Actually
it's.
A
Easier
to
do
in
a
public
cloud
because
usually
the
public
cloud
provider
is
going
to
have
one
and
only
one,
but
once
you
go
on
Prem
I
would
contend
that
you
don't
want
to
lock
out.
You
know,
declare
that
their
one
load
balance
or
rules
of
them
all
that
wouldn't
be
a
good
thing
to
do
so,
but
in
the
cloud
providers
say
they
are
talking
about
both
load
balancer
in
ingress
and
they're.
Both
things
that
often
are
one
answer
in
public
clouds,
but
multiple
when
you
drop
on
for.
B
Him
we
would
actually
love
to
hear
how
people
are
using
today,
because
we
have
you
know
we
have
some
our
ideas
on
that,
but
we
were
exploring.
You
know
various
options,
because
there's
I
heard:
there's
people
like
using
things
like
nginx
to
do
like
the
front-end
load
balance
right,
they're,
not
obviously
integrated
with
you
know.
A
A
A
Well,
I
think
when
you're
running
kubernetes
in
a
public
cloud
you
use
whatever
solution
the
public
cloud
provider
kind
of
puts
a
gravity
field
on
when
you
go
to
on
Prem
there's
a
decade-plus
history
of
various
commercial
load
balancers
and
some
open
source
ones,
yeah
anything
from
H
a
proxy
to
network
equipment.
Vendors.
A
You
know
at
VMware,
I,
don't
think
that
company
wanted
to
play
any
favorites,
and
my
experience
is
that
once
something
got
used
on,
Prem
people
aren't
really
wild
about
switching
so
I.
My
initial
impression
is:
we
ought
to
support
them
all,
but
it
shouldn't
be
a
huge
effort
to
just
make
it
work
in
kubernetes
how
we
pull
that
off.
A
I,
don't
know
if
you
want
to
come
to
our
cig
and
start
composing,
an
engineering
group
to
kind
of
put
an
abstraction
layer
around
load
balancers
for
on
Prem
I
think
we'd
be
very
happy
to
participate
in
that
or
on
the
other
hand,
if
you
are
with
a
load,
balancer
vendor
and
you
want
to
come
and
give
a
presentation
on
how
to
make
it
worth.
Work
with
kubernetes
on
prem
on
vSphere
we'd
be
open
to
giving
you
a
slot.
A
Yeah
yeah,
so
just
to
repeat,
he
pointed
out
that
with
NSX
there,
that
is
a
for
those
who
don't
know.
That's
a
VMware
product
I
hate
to
push
a
product
because
it
looks
biased,
there's
other
products
as
well,
but
yeah
there
are
solutions
that
would
just
work.
The
key
is,
can
you
make
it
automatic,
so
you
didn't
even
have
to
look
lift
your
little
pinky
finger
to
just
have
it
work
now.
B
A
Think
maybe
we
are
getting
the
cut
off
here,
but
we'll
be
hanging
around.
So
if
you
want
to
ask
us
questions
right
after
this
feel
free
to
take
this
to
the
hallway
or
I'm
in
a
number
of
other
sessions,
I'll
be
here
all
week
track
me
down.
If
you
can't
find
me,
ask
ask
them
to
find
me
in
the
VMware
booth
or
something,
and
somebody
should
be
able
to
find
me
all.