►
From YouTube: Kubernetes UG VMware 20211104
Description
November 4, 2021 meeting of the Kubernetes VMware User Group. Recap of announcements and highlights from recent VMworld and KubeCon North America conferences
A
Hi
welcome
to
the
november
4th
meeting
of
the
kubernetes
vmware
user
group
on
today's
agenda.
We've
got
currently
one
item,
which
is
a
recap
of
recent
announcements
and
presentations
that
went
on
on
some
of
the
recent
conferences
which
were
vmworld,
maybe
the
associated
devops
loop
event
and
kubecon.
A
B
A
Okay,
great
so
the
the
first
thing
that
happened.
This
is
at
kubecon,
I
guess
technically.
Q
con
was
after
vmworld,
but
part
of
the
kubecon
presentations.
It
was
a
hybrid
conference
that
was
partially
online,
partially
recorded,
so
I'm
going
to
show
some
slides
that
were
presented
at
q
con
regarding
recent
updates
to
the
vsphere
cloud
provider.
A
They
were
recorded
in
advance,
so
technically
you'll
see
a
few
slides
here
that
indicate
some
of
these
things
are
roadmap
or
projected,
but
they
they
have
indeed
already
happened.
So
ignore
the
fact
that
they're
called
out
as
roadmap
and
one
of
the
devs
nicole
hahn
gave
this
presentation
at
kubecon.
But
there
have
been
some
recent
updates.
A
The
latest
version
has
some
zone
and
labeling
support,
improved
support
for
an
sxt
and
well
you
can.
You
can
read
it
here:
the
cloud
provider,
if
you're
not
using
a
distro
that
installs
it
as
part
of
the
distro
installation,
now
has
a
mechanism
to
be
installed
by
a
helm
chart.
You
know
this
might
be
useful
if
you're,
using
pure
upstream
open
source,
kubernetes
or
some
distro
that
doesn't
install
the
cloud
provider
for
you.
A
At
this
it
was
recorded,
so
I
accidentally
triggered
playback
of
the
recording.
So
the
other
subject
that
came
up
at
kubecon
was
the
vsphere
cloud
provider
moving
out
of
tree.
A
Given
that
the
there
have
been
no
new
features
added
to
that
entry,
there's
probably
plenty
of
good
reasons
to
move
anyway,
even
if
you're
not
forced
to
do
it
kind
of
the
only
situation.
I
can
see
that's
valid
for
clinging
to
life
on
that
old
legacy.
Thing
is
that
you're,
maybe
on
ancient
hardware
or
an
ancient
vsphere
release
that
you
haven't
upgraded,
that
can't
be
supported
by
the
new
out
of
tree,
but.
A
A
A
They've
had
an
aspirational
goal
of
this
at
one
time
even
happening
by
about
now,
but
when
they
actually
went
into
the
logistics
of
pulling
off
the
migration,
so
that
people
who
were
stuck
on
the
legacy
could
do
things
like
have
some
hope
of
you
know.
Moving
legacy
deployed
apps,
including
stateful
over
they
discovered
that
there
were
a
number
of
moving
parts.
They
had
thought
of,
so
that
this
this
has
been
delayed,
and
I
you
know
I'm
not
in
charge
of
this.
A
It's
walter
fender
of
google,
who's,
chair
of,
say,
cloud
provider
and
even
then
it's
somewhat
democratic.
I
think
you've
probably
got
a
year
and
maybe
longer,
but
you
probably
should
pretend
you've
only
got
a
year
because
you
don't
want
to
be.
You
know
doing
this
kind
of
this
thing
last
minute
and
if
you're
running
it
into
into
any
difficulties,
if
you
at
least
attempt
some
migrations
now
and
give
feedback,
you
know
there's
some
chance
that
your
pain
points
would
be
addressed.
A
But
if
you
sit
there
and
wait,
you
know
until
the
code
is
actually
getting
removed
out
of
the
out
of
the
tree.
C
A
C
A
Yeah,
I
don't
know
if
I
linked
it
in
this
deck.
I
I
whipped
this
deck
together
just
an
hour
ago,
but
the
sessions
from
kubecon
are
now
posted
up
on
youtube
by
the
cncf
and
the
actual
wording
of
what
the
official
story
is
on
this
migration.
A
I
would
say:
listen
to
what
walter
fender
said
during
the
sig
cloud
provider
session
and
I
can
track
that
down
and
post
I
I
can
send
it
to
you
and
put
it
in
the
agenda
notes,
because
I
don't
think
I
put
it
in
the
slide
deck,
but
it
would
be
useful.
That's
the
official
story,
you
know
not
listening
to
me.
He
made
a
statement
at
kubecon
and
I
think
that's
you
know
given
that
he's
chair
of
say
cloud
provider,
that's
the
one.
I'd
go
with.
A
A
You
know
the
out
of
tree
ready,
but
the
issue
is
that
somebody
will
pull
a
switch
and
actually
remove
the
code
out
of
the
main
kubernetes
distro,
and
then
it's
no
longer
an
option
at
that
point
and
that's
that
session
was
actually
kind
of
an
interesting
one
to
look
listen
to
because
it
went
over
kind
of
the
details
of
actually
the
amount
of
code
that's
being
removed
is
pretty
big,
and
these
are
bits
that,
if
you're
shuffling
them
out
to
you
know
on-prem
edge
locations
and
stuff
you're
paying
for
the
bandwidth,
transmission
and
the
storage.
A
And
you
know
in
a
way
it's
sad
that
kubernetes
as
a
container
orchestrator
was
plugging
the
microservices
versus
monolith
story.
Well,
in
some
in
some
ways
it
was
a
monolith
itself
so
that
you
might
be
deploying
on-prem
on
vsphere,
but
still
dragging
along
the.
You
know
the
the
bits
associated
with
cloud
providers
or
amazon
google's
cloud
azure,
whatever
else
might
be
in
there,
and
it
will
be
nice
in
some
sense
that
that
goes
away
in
the
future.
C
What
what
are
you
guys,
seeing
as
the
pain
points
being
people
been
seeing
in
those
migrations?
I
think.
C
That
kind
of
separates
that
out
to
its
own
component
right.
A
Yeah,
I
I
think
it's
this.
You
know,
storage
panes,
because
the
rest
of
your
pods
don't
really
have
that
much
associated.
If
you
did
move
and
then
ties
to
legacy
software
like,
like
you
just
said
that
you're
on
old
vsphere
releases
and
for
most
people,
I
think
most
people
who
are
on
old
vsphere
releases
also
have
they
have
a
reason
for
being
on
the
old
ones,
which
is
often
that
they're
also
on
old
hardware
and
it's
nice
to
say
you
could
move
up
to
7-0
up,
update
2
or
update
3
the
latest.
A
But
you
know
I
do
understand
that
there
are
some
pieces
of
legacy
hardware
that
simply
can't
be
supported
by
the
latest
vsphere
releases,
because
there
are
some
storage
drivers
and
network
drivers
that
dropped
out
of
out
of
support.
Some
of
them,
maybe
can
still
run
even
unsupported,
but
some
of
them
won't
even
install.
A
I
know
I
was
sort
of
in
it,
but
I'm
talking
for
me
it's
it's
running
my
home
lab
where
you
know,
given
that
it's
a
home
lab,
I'm
cheap,
I
don't
run
out
and
buy
new
hardware
every
year,
so
I'm
on
older
stuff,
but
even
in
production
there
are
places
that
end
up
with
hardware.
I
think
at
this
point
to
be
in
that
position.
A
A
There
is
also
there's
also
been
a
new
csi
driver
update.
That's
come
out
actually
two
of
them
in
about
the
last
month,
so
the
2.22
was
bug
fixes
for
people
on
the
2.2
series
and
the
2.3
has
some
new
features
and
some
some
changes.
I'm
not
going
to
go
over
them
all
here,
but
if
you
click
click
that
link
on
the
docs
for
b
each
of
these
you
can,
you
can
see
the
list
of.
What's
there,
I
think
in
the
2
3
it
supported
a
new
file
system
type
and.
A
On
what
the
other
one
was,
but
there
were
a
couple
new
features.
Well
here,
let
me
I
guess
I
can
open
the
dock
and
see
what
comes
up.
Oh,
it
supports
xfs
and
raw
block
volume.
Support
is
now
an
alpha
feature
so
that
that's
what
came
out
in
that
2.3.0
of
csi,
then
at
vmworld,
the
new
7-0
update,
3
release
of
vsphere
came
out.
The
number
of
bug
fixes,
including
some
security
fixes
there
are
some
new
devices
supported.
A
I
think
I
glanced
over
the
list
and
I
think
a
lot
of
the
new
devices
that
are
being
supported
are
in
the
realm
of
storage,
but
there
are
also
deprecation
warnings
on
some
legacy
devices.
So
that's
been
a
recent
trend
that,
as
every
update
comes
out
some
of
the
older
storage-
and
I
don't
know
if
in
this
case
any
of
the
network
nics
dropped
off
the
map.
A
But
if
you're
on
older
hardware
you,
you
definitely
should
take
a
look
at
the
release
notes
before
blindly,
assuming
that
you
can
update
and
I've
posted
a
blog
post
there,
that
kind
of
explains
it
in
a
little
more
detail
in
writing.
That
was
by
bob
plankers,
and
I
think
he
also
has
a
if
you
don't
want
to
read
the
blog
and
prefer
video
blog
plankers
also
has
a
video
form
of
that
blog
post
up
on
youtube.
A
So,
oh
okay,
so
I
do
have
a
link
to
that
session,
recording
at
kubecon.
So
the
second
one
shown
here
was
by
say
cloud
provider,
and
that
has
detail
on
what
the
expectation
is
for
the
calendar
on
when
you're
going
to
be
forced
to
get
to
to
drop
the
entry
cloud
providers
and
then
miles-
and
I
had
a
presentation
on
using
gpus
with
kubernetes
pods
hosted
on
vsphere.
A
A
It
creates
a
network-based
pipeline
over
to
the
node
that
has
the
physical
gpu
and
for
many
types
of
workloads
there
is
a
cost,
because
you
know,
if
you
had
a
physical
gpu,
your
data
movement
would
happen
over
a
pci
express
lane,
and
in
this
case
it's
happening
over
an
ethernet
link.
A
A
A
You
know
there
are
a
lot
of
scenarios
where,
with
machine
learning,
the
training
stage
maybe
uses
full
gpu
cards,
but
subsequent
execution
might
need,
might
benefit
from
a
portion
of
a
card
but
be
unable
to
use
the
full
car,
and
this
bit
fusion
technology
lets.
A
You
allocate
resources
in
units
of
fractional
gpus
or
you
know
individual
cuda
cores
and
that
that
can
be
nice
for
a
runtime
execution
of
machine
learning
workloads
where
you
could
have
a
bunch
of
kubernetes
pods,
sharing
a
gpu
each
gpu
card,
each
consuming
a
fractional
part
of
it,
and
that
would
be
something
perhaps
a
little
difficult
to
pull
off
in
a
pure
physical
scenario.
A
So
anyway,
the
link
to
that
presentation
recording
is
there
the
decks
for
these
are
up
on
the
kubecon
skid
site.
A
A
Well,
I
don't
have
the
link
in
this
slide,
but
it
it
at
every
kubecon
conference.
They
have
used
some
service
called
skid,
s-c-h-e-d,
dot,
com,
yeah
and
you
can
find
the
schedule
and
then,
if
you
click
on
a
session,
the
speakers
are
supposed
to
upload
the
decks
and
for
both
of
these
sessions.
A
A
And
the
vmworld
sessions
are
all
publicly
available.
You
do
have
to
create
a
free
vmworld
account
to
play
them,
but
they're
out
there,
and
this
link
will
search
for
vmworld
from
2021,
with
an
additional
filter
of
kubernetes.
Now
I'll
caution,
you
that
this
group
is
about
you
know
pure
upstream,
open
source
kubernetes.
A
So
some
of
those
sessions
were
in
the
context
of
vmware's
commercial
distributions,
but
I
think
it's
fair
to
say
that
in
many
cases
the
things
covered
with
regard
to
you
know,
what's
available
best
practices
etc,
are
more
along
the
lines
of
use
cases
so
that
you're
likely
to
find
useful
content
in
those.
Even
if
nominally
the
session
was
sort
of
you
know,
vmworld
isn't
the
kubecon.
So
you
know
it.
It
is
a
commercial
conference.
A
So
if
commercial
pitches
offend
you,
you
probably
shouldn't
go
look
at
these
videos,
but
I
still
think
that
for
many
people
there's
value
to
be
found
in
these
sessions,
just
because
they're
speakers
giving
talks
yeah
about
kubernetes
in
general
and
the
fact
that
they
maybe
used
a
commercial
distribution
is
sort
of
orthogonal
to
the
the
main
message
being
delivered
in
this,
and
I
believe
that
the
number
that
filter
will
return
you
20
or
more
sessions,
so
there's
a
fair
amount
of
content
out
there
available
at
that
length.
There.
E
Are
also
some,
there
are
also
some
see
from
the
code
connect
which
which
are
on
youtube
and
not
on
the
vmworld
site
on
vmware
code,
which
were
you
know
more
whatever
I
mean
some
other
stuff
and
some
directly
related
to
kubernetes
and
not
tan,
zoo
specific
as
well.
A
A
Okay,
great
and
if
anybody
caught
any
sessions,
either
at
kubecon
or
at
vmworld
and
wants
to
talk
about
things,
they'd
recommend
or
discourage
people
from
wasting
their
time
on,
go
ahead
and
chime
in
here.
Also,
if
you
want
to
recount
anything
that
occurred
at
these
conferences.
Well,
I
don't
vmworld
was
pure
virtual,
so
I
don't
know
that
this
happens
to
a
great
extent,
but
kubecon
had
a
lot
of
interesting
call
it
the
hallway
track
stuff
going
on
that.
I
found
interesting.
D
I
I
thought
vmworld
was
a
lot
better
than
last
year.
B
D
It
still
being
a
virtual
conference,
they
really
they
really
tried
to
yet
to
give
that
kind
of
hallway
feeling
to
it.
They
they
tried
to
keep
their
slack
channels
active
and
they
had
a
lot
of
kind
of
additional
stuff
going
on
around
the
around
sessions,
but
I
thought
the
most
most
the
value
was
was
with
the
live
kind
of
workshop
session,
so
they
had
a
lot
of
pre-recorded
content.
I
mean
that
stuff.
I
don't
see.
D
You
know
that
that
stuff,
I
don't
see,
has
the
added
value
I
mean
you
can
always
watch
it
later
and
it's
free
content
so
that
I
got
most
of
the
most
the
value
out
of
out
of
the
the
tech
plus
pass
because
it
was
you
know
you
do
all
the
meet
the
expert
sessions
and
those
were
actually
those
were
some
of
the
best
sessions,
because
you
know
you
actually
get
to
talk
to
people.
It's
a
smaller
group.
You
actually
get
into
quite
a
lot
of
detail.
A
Yeah,
unfortunately,
I
think
I
did
some
of
those
myself
and
they
weren't
being
recorded
so
for
people
who
missed
it,
I'm
not
sure
how
much
of
an
opportunity
there
is
to
catch
up
on
what
went
on
in
those
but
yeah.
I
had
the
same
experience
where
I
think
people
are
learning
to
do
these
online
conferences
a
little
better
and
even
people
are
learning
how
to
participate
in
them
more
effectively
too.
D
Well,
I
really
only
focus
on
on
the
live
sessions
and,
like
you
said
when
recorded,
I
think
I'm
not.
I
just
not
to
forget
like
that
on
the
monday,
at
the
devops
loops
kind
of
side
conference,
the
pre-vmworld
conference,
which
was
focused
mostly
on
on
on
devops
and
cloud
native
stuff
that
had
some
some
very
excellent
sessions
in
there
as
well.
I
believe,
I
believe,
all
of
those
were
recorded,
but
they
were
more.
The
kind
of
the
traditional
traditional
and.
E
D
D
Well
well
worth
re-watching
also
because
they
they're
also
slightly
higher
level.
They
were
more
like
you
know
what
what
are
they?
What
are
the
trends
currently
around
monap's
development
or
the
trends
around
devops?
What's
the
kind
of
the
the
current
thinking
out
there,
I
thought
those
were
because
I'm
always
into
that
high
level
story.
I
thought
those
were
very
valuable.
Some
of
those.
A
Okay
and
for
any
of
those
interactive
types
that
weren't
recorded,
perhaps
if
somebody
wants
to
make
a
request,
I
could
try
to
track
down
speakers
to
repeat
the
experience
here
in
a
user
future
user
group
meeting.
If
it's
of
interest
and
kind
of
relevant
to
the
subject
of
kubernetes
on
vsphere.
A
Okay
with
that
said,
I
think,
okay,
that's
that's
the
end
of
what
I
had
prepared
here.
So
at
this
point
this
was
the
recap
on
our
agenda.
If
anybody's
got
any
other
subjects,
the
forum
is
open
for
discussion
here,
also
I'll
throw
out
any
anybody.
Who's
got
requests
for
content
at
future
meetings
include
december
or
later.
C
A
E
I
don't
know
steve
what
falls
within
the
guidelines
of
the
user
group
here,
but
with
tce
being
open
sourced
and
being
a
opportunity
on
vsphere.
I
don't
know
if
that
falls
into
a
possible
subject
to
bring
up
here
or
if
we
try
to
be
completely
distribution
agnostic
in
this
group.
Even
if
it's
not
a
commercial
offering.
A
They
have
released
a
free,
open
source
distribution
of
kubernetes,
that's
based
on
the
same
technology
used
in
the
tonzo
commercial
products.
So
you
know,
like
I
said
it's
same
technology
used
in
the
tanzu
commercial.
It
does
have
some
aspects
that
are
a
little
different,
some
of
those
tonzu
commercial
products.
Obviously
they
come
with
commercial
support
and
the
community
edition
does
not,
but
the
tonsil
community
edition
is
also
going
to
be
used
as
a
leading
platform
for
evaluating
alpha
level
features
and
technology
ahead
of
what
gets
into
the
commercial
product.
A
A
A
If
you
want
to
learn
more
on
the
kubernetes
slack,
there
is
a
channel,
a
slack
channel
created
for
it
and
you
can
go.
They
have
their
own
community
meetings
and
their
own
github
repo.
So
you
can
learn
more
about
it
there.
Let's
see,
I
think
the
slack
channel
is.
A
E
A
That,
actually,
I
think-
and
it
really
aligns
well
with
this
group,
because
people
running
kubernetes
on
vsphere
often
are
yeah
well
they're,
almost
always
in
an
on-prem
scenario
and
technically
you
could
run
vsphere
in
the
amazon's
cloud,
but
I
think
most
of
the
people
coming.
This
group
are
more
of
interest
in
running
it.
A
On-Prem
and
sometimes
the
air
gap
is
voluntary
when
you're
on-prem,
sometimes
it's
a
temporary
outage
where
you're
faced
with
you
know
a
short-term
air
gap,
whether
you
like
it
or
not,
and
the
kind
of
things
you
do
for
air
gap
can
often
be
relevant
to
people
who
don't
intend
to
don't
prefer
to
run
air
gap
but
face
the
issue
from
time
to
time.
So
yeah,
that's
a
great
idea.
I'll!
Try
I'll
try
to
line
something
up
along
those
lines.
A
D
Yeah,
well,
you
just
triggered
me
as
well
scott,
because
the
what
I
found
interesting,
what
I
find
interesting
about
about
about
the
the
vmware
distributions
and
ntc
is,
is
this
adoption
of
this
this
packaging
standard
around
the
carville
toolkit?
I
mean
still
being
relatively
new
to
to
this
this
world.
D
Part
of
it
is
like
well,
that's
a
major
driver
behind
the
carvel
stuff,
at
least
so
it's
becoming
one
as
you,
as
we
can
kind
of
notice
in
the
in
some
of
these
projects,
so
that
that
I'd
love,
I'd,
love,
someone
with
a
little
bit
more
knowledge
about
broader
knowledge,
about
the
differences
between
those
those
different
ways
of
delivering
software
into
kubernetes,
to
maybe
contrast
those
different
different
approaches
so
that
that's
that's
something
I'm
I'd
like
to
to
know
more
about,
but
another
idea.
D
I
had
it's
actually
something
that
I
would
like
to
do
a
session
about
at
some
point,
because
kind
of
a
hobby
horse
for
me
is
is
availability
around
kubernetes.
So
we've
talked
about
this
on
this
at
this
user
group
before.
But
how
do
you
map
availability
constructs?
You
know
that
are
native
to
the
kubernetes
world
in
the
cloud
native
world
onto
you
know,
onto
cloud
provider
logic
in
this
in
our
case
vsphere,
and
how
do
you?
D
How
do
you,
you
know,
make
those
two
things
work
together
really
well
and
that
that
and
that
again
that
that
can
be
mostly
distribution
agnostic?
Obviously,
my
experience
is
going
to
be
on
the
on
the
vmware
tkg
side,
but
these
principles
they
apply
generally
I'd
love.
I've
never
actually
seen
anyone
do
a
session
on
that
or
wrote
or
write
a
decent
blog
post.
A
About
that,
yes,
there
actually
were
sessions
on
that
at
kubecon's,
going
back
three
to
four
years,
and
then
it
sort
of
dropped
off
the
map,
and
so
much
has
changed
since
then
that
I
think
it's
perfectly
valid
to
go
revisit
that
what
hasn't
changed
since
three
to
four
years
in
kubernetes.
A
Yeah,
when
you
get
to
availability
there's
you
know
how
do
you
make
it
resilient
so
that
you
never
took
an
outage
in
the
first
place,
then
taking
it
to
the
next
stage,
where,
let's
assume
you
took
an
outage
in
spite
of
your
best
efforts,
how
do
you
get
back
to
where
you
were,
or
how
do
you
fail
over
to
an
alternate
region?
You
know
it's
a
pretty
rich
topic
that
I
think
would
go
beyond
a
single
meaning,
potentially.
E
And
especially
with
the
you
know,
cpi
csi
migration
from
cloud
providers
where
all
of
that
changes
on
vsphere
the
way
that
it's
done
from
the
entry
model,
even
people
that
are
using
that
today
there
are
different
capabilities
that
exist
today
within
the
vmware
solutions
for
kubernetes
or
vmware
integrations
with
kubernetes
yeah.
That
makes
it
relevant.
C
One
of
the
things
I
was
going
to
bring
up
was
something
that
I
don't
even
know
if
it
works
the
same
in
the
outa
tree,
but
like
an
issue
we
have
right
now
is
if
using
the
entry
cloud
provider.
C
E
A
E
Won't
the
cpi
still
needs
that
connection
to
vcenter
in
order
to
configure
itself
and
to
get
back
the
provider
id
so
that
still
exists.
There
is
work
upstream
to
change
that
into
a
in
cluster
api
upstream,
meaning
not
a
vmware
solution,
but
a
subset
of
distributions
and
in
cluster
api.
There
is
a
work
to
move
to,
what's
called
a
para
virtual
module
model
where,
in
the
management
cluster
api
cluster,
you
would
have
access
to
the
vcenter
and
didn't
need
that
from
your
workload
clusters
or
your
actual
application
clusters.
C
Yeah
and
the
interesting
thing
is
right
now
and
I
I
feel
like
they
may
have
fixed
this
in
a
newer
kubernetes
version,
but
in
the
current
version
we're
running
if
it
stays
that
way
long
enough,
the
cubelet
just
it
it
stops.
Trying
to
start
the
controller
manager
stops
trying
to
start
some
of
those
control,
plane,
pods
and
never
tried
it
even
it
just
times
out
it
just
like
I
give
up
so
yeah.
C
E
In
the
cubelet
code-
and
you
have
to
restart
cubelet
in
order
for
it
to
try
again,
yes,
yes,
exponential
back
offs
are
always
fun.
C
So
that
that
kind
of
goes
back
to
maybe
some
of
the
air
air
gapping
that
you
were
even
talking
about
is
essentially
we
have
some
of
our
clusters-
air
gapped
from
the
vcenter
and
so
they're,
not
always
air-gapped,
but
if,
if
they're
remote-
and
they
can
be,
then
air,
gapped
and
they're
expected
to
work.
But
if
one
of
those
nodes
restarts
during
that
situation,
then
it
it's
stuck
like
that.
C
A
A
You
know
just
get
the
cluster
to
tolerate
incidents
and
then,
if
you
go
a
step
further
and
run
app
workloads
on
top
of
that,
there
are
tools,
you
know,
like
valero
level,
things
for
achieving
resilience
up
up
at
an
app
level
and
I
think
they're,
two
different
subjects
with
different
techniques
for
addressing
the
issue,
but
they're
both
very
relevant,
maybe
trying
to
commingle
them
in
one
talk
is
a
little
too
much
and
potentially
confusing
anybody
got
thoughts
on
that.
E
Yeah-
and
I
think
it
definitely
should
be
split
into
you-
know
different
talks,
I
think
there's
the
infrastructure
side
and
how
you
deal
with
infrastructure
level
of
kubernetes
and
air
gap,
and
then
there's
also
the
more
I
guess:
developer
devops,
whatever
the
applications
in
air
gap
in
actually
utilizing
kubernetes
and
air
gapped.
Once
you
have
a
kubernetes
cluster,
I
think
our
two
separate
domains
that
both
deserve
you
know
an
entire
session,
possibly
just
because
they're
so
large-
and
you
know
there's
so
much
to
talk
about
there.
Uh-Huh.
A
Earlier
on
in
this
discussion,
I
know
a
few
of
you
brought
up
the
whole
packaging
thing
with
carvel,
but
I
just
want
to
make
sure
that
everybody
on
this
call
even
knows
what
carvel
even
is
because
yeah,
it's
relatively
newer,
if
you
do
just
say
so
now
and
stop
me.
But
you
know,
carvel-
is
an
attempt
to
elevate
packaging
of
apps
and
system
tools
beyond
what's
available
in
helm.
A
It
allows
you
to
publish
things
in
a
in
in
a
system
that
offers
opportunities
for
being
a
little
more
opinionated
to
keep
you
on
a
safe,
glide
path
and
it
is
being
used
in
tanzu,
but
it
is
an
open
source
project
that
is
supported
by
the
cnn
in
the
cncf
in
infrastructure,
and
I
think,
a
few
of
those
things
scott.
I
I
wanted
to
give
you
a
shout
out
for
some
of
those
packages
you
created.
A
E
E
E
I
think
what
is
it
I
think
it's
like
57
gigabytes
for
the
last
four
versions
of
every
helm,
chart
there,
but
completely
open
source
and
works
on
any
I've
used
it
on
eks
anywhere.
I've
used
it
on
tanzu
community
edition.
I've
used
it
on
commercial
offerings
from
vmware,
and
I've
used
it
on
a
kind
cluster
as
well.
So
it's
I
would
say
agnostic.
The
only
thing
I
don't
know
if
it
works
on
is
openshift
because
that's
its
own
beast,
but
yeah.
E
So
I
I
definitely
think
that
carvel
would
be
an
interesting
conversation
to
have
here,
because
image
package,
especially
will
be
huge
for
air
gapped.
D
Yeah,
dear
scott
and
steve,
I'm
interested
in
your
thoughts
on
this.
The
do
you
I
mean
I
I
I
don't
understand
the
caval
tilke
as
deeply
as
as
suddenly
as
a
scot,
but
the
it
seems
to
it
seems
to
address
some
some
limitations
around
helm,
mostly
and
then,
and
then
I
was
kind
of
already
familiar
with
with
some
of
the
overlay
stuff
that
ysd
does
because
of
the
customized.
D
I'm
just
wondering
if,
if
these
are
things
that
the
wider
community
outside
of
the
vmware
ecosystem,
you
know
people
you
know
are
looking
for
alternatives
to
to
that.
You
know
how
how
how
broad
is
the
problem?
How
wide
is
the
problem
and
how
much
of
a
chance
does
a
project
like
like
the
carvel
tools
have
of
breaking
out
of
the
vsphere
ecosystem
when
it
comes
to
to
packaging
stuff
for
kubernetes.
A
D
Well,
so,
but
just
just
that
point
steve,
because
it's
it's
interesting,
you
say
that
the
problem
that
I
have
from
I'm
I'm
very
much
inside
the
vsphere
ecosystem
right.
So
so
all
I
kind
of
know.
Well,
I
know
a
bit
more
now,
but
I
love
this
initial
stuff
is
coming
to
me
through
the
lens
of
of
what
people
are.
What
vmware's
developing
in
need
of
tanzu.
A
D
So
I
I
don't
know
what
else
is
going
on
and
and
the
the
so
I'm
very
much
aware-
and
I
see
this
in
a
bunch
of
different
projects,
all
the
all
the
open
source
projects
that
vmware
is
involving
themselves
in
is
very
much
to
serve
the
needs
of
getting
of
tanzu,
but
you
know
broadly
in
all
kinds
of
different
ways,
so
I'm
not
automatically
assuming
that
that
they're
pushing
a
particular
project
or
a
particular
technology
stack,
because
it
would
benefit
the
community
generally
or
kubernetes
generally,
because
I
know
that
there's
a
big
commercial
interest
in
what's
driving
them
a
lot
of
these
projects.
D
So
so
that's
why
I
would
like
to
someone
with
a
bit
with
a
broader
experience
of
the
kind
of
community.
A
view
of
this
to
to
explain
you
know
to
kind
of
contrast.
Those
things
you
know
is
is
because
I've
seen
a
few
examples
where
vmware
seems
to
be
doing
something
very
interesting
and
innovative.
No
one's
heard
of
it
outside
of
the
v
sphero,
pivotal
ecosystem,
yeah,.
E
E
And
I
can
tell
you
from
my
experiences
I
deal
with
a
lot
of
public
cloud
kubernetes
as
well
from
you
know,
the
you
know
hyperscalers
in
their
managed
kubernetes
as
well
as
well
as
branch
or
openshift
and
other
distributions,
and
it's
definitely
an
issue
that
exists
in
other
places.
E
There
is
you
know
the
partial
issue
of
you
know
in
the
end,
who
came
first
to
the
game,
so
helm
came
first.
So
that's
what
most
people
are
using
customize
is
built
into
cube
ctl.
E
So
a
lot
of
people
are
using
that
because
they
can
just
do
ctl
apply
minus
k,
but
the
issues
with
both
of
them
are
prevalent,
and
you
see
that
by
customers
and
users
in
environments
that
are
not
just
tanzu,
you
see
that
in
any
kubernetes
environment
and
as
such,
the
carvel
tool
said,
I
think
definitely
has
a
lot
to
offer
in
those
worlds
and
we're
actually
utilizing
it
in
non-tanzu
clusters
today.
So
it
definitely
is
getting
usage
there.
Is
it
as
big
as
I
think
it
should
be?
E
The
answer
is
no,
but
it's
getting
there
more
and
more
people
are
using
it
these
days
outside
of
tanzu,
and
it's
just
very
new,
which
is
why
it,
I
don't
think
is,
is
as
prevalent
in
the
community
as
it
will
be.
I
think
in
the
not
so
far
future.
E
A
Often
these
major
problems
attract
the
fact
that
a
lot
of
people
feel
a
pain
point
often
causes
two
or
three
solutions
to
the
pain
point
to
come
about,
and
then
you
have
to
go
through
something
I
mean
this
goes
all
the
way
back
to
the
early
days
of
kubernetes.
When
you
had
the
orchestrator
wars
of
kubernetes
versus
mesos
versus,
I
don't
know
swarm
and
a
few
others
that
were
in
the
mix.
D
One
of
the
things
that
impressed
me
about
the
tc
release
was
that
that
there
seems
to
have
been
a
very
big
effort
to
actually
put
all
of
the
community
building
blocks
in
place
from
day
one
right.
So
they
have,
you
know
like
dedicated
community
manager
and
they
have
dedicated
evangelists
people.
You
know
in
that
role
promoting
that
project.
You
know
they're
putting
it
big
on
the
radar
trying
to
attract
a
lot
of
people
which,
in
turn,
is
going
to
help
carvel
a
lot
right,
yeah
and
a
bunch
of
other
stuff
around
around
around
tcc.
A
But
if
people
can't
find
support
and
get
their
questions
answered
and
have
a
feeling
that
they
can
get
listened
to
with
regard
to
issues
and
feature
requests,
it's
going
nowhere
in
the
long
run,
and
I
I
I
think,
if
it's
a
project
that
has
a
lot
of
vmware
contribution
that
that's
currently
viewed
as
a
necessity
on
anything
that
we're
going
to
get
involved
in.
A
But
it's
nice
to
hear
your
feedback
that
that
that's
the
perception
that
you
see
that
taking
off
in
the
right
direction,
at
least
okay.
Well,
I've
got
some
great
ideas,
then,
for
future
meetings
of
this
group,
I'm
not
sure
I'm
going
to
be
able
to
put
together
that
whole
availability
thing
as
quickly
as
first
week
of
december,
but
I
can
try
and
robert
if
you
want
to
follow
up.
I
I
know
you
you
mentioned
that
you
might
be
interested
in
presenting
so
you're
welcome
to
that
either
solo
or
if
you
want
I'll
volunteer.
A
So
if
you
don't
want
to
do
that,
solo,
let
me
know
too,
and
I
can
either
recruit
other
participants
or
try
to
help
out
myself
with
as
long
as
you
have
the
understanding,
I'm
not
claiming
to
be
the
subject
matter
expert
there,
although
I
do
have
a
fair
amount
of
experience
with
regard
to
backup,
in
particular,.
D
Yeah,
it's
no!
It's
just
it's
just
an
idea.
I
have
I
mean
I
I
I'm
actually
there's
a
there's,
a
vmware
user
group,
the
dutch
vmware
green,
is
coming
up
in
december
and
they've
asked
me
to
do
something
for
it.
I'm
still
trying
to
figure
out
what
whether
I'll
just
rehash
my
tanzy
for
dummies
and
update
it
or
do
something
on
tce
or
something,
but
that
that
idea
of
doing
a
proper
one
about
availability.
I've
been
thinking
about
that
a
long
time,
but
it
will
also
require
quite
a
lot
of
preparation.
I
think.
A
A
D
A
Right
yeah,
you
put
the
same
amount
of
effort
in.
I
do
the
same
thing
where,
if
you're
going
to
do
a
presentation,
maybe
you
can
get
a
blog
post
out
of
the
same
amount
of
effort
and
and
also,
if
you're,
going
to
one
of
the
major
conferences
like
kubecon.
I've
always
found
it
useful
to
practice
a
presentation
at
a
meetup
or
something
first,
where
you
don't
have
as
many
eyeballs
and
yet
actually
probably
can
get
more
interactive
feedback
at
something
like
a
meetup.
D
A
D
A
time
for
it,
but
but
if
you
wanna,
you
wanna
give
a
shot
at
that
topic.
I'd
love,
I'd,
love
to
see
yeah.
A
Yeah,
like
I
say,
if
you,
if
you're
interested
in,
maybe
even
if
it's
next
year
doing
that
joint
either
with
me
or
help,
I
can
help
recruit
others
to
participate
in
it.
That
might
actually
be
a
decent
panel
session
for
a
conference
as
well.
Maybe
we
can
put
something
together.
E
A
Okay,
that
sounds
great
in
fact,
like
I
said,
I
really
appreciated
some
of
those
packages
you
put
together
and
they're
of
general
interest,
like
your
cube
vip.
So
even
you
know,
there's
the
topic
of
general
carvel
as
well
as
some
of
the
actual
working
examples
of
how
you
can
utilize
it
to
do
some
good
in
the
world.
I
I
would
love
to
have
you
do
a
session
like
that.
E
Yeah,
definitely
more
than
happy
to
do
that
and
you
know
discuss
how
to
do
that,
and
I
just
built
one
for
generic
helm
instead
of
just
bitnami,
and
things
like
that.
That
you
know
are,
in
the
end
helpful
for
development
cycles.
Outside
of
you
know
specific,
you
know,
vmware
open
source,
you
know
things
as
well.
A
Okay,
that
sounds
great
I'll,
follow
up
with
both
of
you,
then,
on
dm'ing,
you
on
the
kubernetes
slack,
if
that's
a
good
way
to
keep
in
touch
with
you,.
A
A
Okay,
I'll
take
silences
no,
and
thanks
for
attending
it
was
I
I
ended
up
learning
a
lot
from
the
audience,
as
I
usually
do.
So,
thanks
for
attending
bye.