►
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
Towards a Vendor Neutral Kubernetes - Andrew Kim, DigitalOcean
With over 50 Kubernetes Certified Platforms and Distributions, building Kubernetes with an open cloud philosophy has never been more important. This comes with a large set of technical and organizational challenges, from supporting the diverse range of features on the cloud (e.g. persistent volumes, load balancers, etc) to sustainably adding new providers into the ecosystem. In this session, Andrew, co-chair of SIG Cloud Provider, dives deep into what the community has been doing to move towards a vendor-neutral model with sustainability and portability as a top priority. He will discuss the migration of vendor-specific code out of Kubernetes core, architectural considerations of pluggable cloud provider features, the organizational challenges along the way, and the roadmap ahead.
To Learn More: https://sched.co/GrTl
A
A
Ones
and
distributions
that
want
to
run
into
programs
so
before
we
dig
into
some
of
those
topics
a
little
bit
about
myself,
I'm
one
of
the
co-chairs
of
Sigma
providers
and
I'm
a
software
engineer
and
ask
you
to
ocean.
If
you're
watching
the
reporting
of
us
talk
or
we
don't
get
a
chance
to
chat
at
UConn,
you
couldn't
reach
me
on
Twitter.
I
came
here
so.
A
A
A
A
A
So
the
stock
is
more
focused
around
how
we
want
a
standard.
A
standard
customized
prefer
Nettie's
so
that
it
can
run
on
any
operator,
so
public,
private
hydrant.
You
name
it
right.
The
goal
here
is
to
get
to
a
point
where
kubernetes
is
truly
cloud
agnostic
and
it's
hard
to
talk
about
this
without
talking
about
the
concept
of
kubernetes
of
providers,
because
really
the
term
provider
is
very
broad
and
it
could
mean
a
lot
of
things,
but
in
kubernetes.
B
A
B
B
A
Even
about
this,
without
talking
about
injury,
and
so
the
best
way
to
kind
of
dive
into
those,
two
concepts
is
actually
dig
into
so
the
history
behind.
Why
we
even
have
this
like
entry
of
energy
concepts,
and
so
when
we
talk
about
injuries,
a
providers
we're
talking
about
the
way
the
implementing
providers
from
the
get-go
right.
A
A
We
make
changes,
if
not
better.
We
have
to
actually
ship
together
with
the
rest
of
an
essentially,
we
realized
this
way
of
supporting
cloud
providers.
Wasn't
scalable
and
Tim.
Hawking
had
originally
suggested
this
idea
of
external
compromise
and
that
kind
of
kicked
off
the
work
for
us
to
start
pushing
all
the
necessary
changes
to
sustainably
support
new
providers
that
are
in
the
system
and
so
under
the
Tim's
destined.
We
built
this
new
architecture
that
supports
the
ability
to
plug
in
an
a.
A
A
Things
to
this
new
component,
we
have
a
manager
so
from
the
development
perspective,
the
hope
is
that
we
can
develop
test
and
release
all
the
individuals
other
letters
in
their
own
trees
and
then
that
allows
us
to
completely
decoupled
all
the
problems,
all
the
top
letter,
logic
from
programs,
and
so
this
was
the
angle
of
cat
and
vision.
But
we.
B
A
B
A
Can
integrate
on
your
features
and
bug,
fixes
it
significantly
faster
and
from
my
experience,
maintaining
the
digital
table
butter
and
we're
finding
that
this
is
really
beneficial
for
us,
because
we
can
add
those
features
without
having
to
wait.
You
know
three
months
of
excrement,
any
is
releases
or
if
it's
urgent,
like
that'll,
have
to
go
to
the
many
of
the
will
sing
attack
release
in
minor
event.
A
A
A
You
rather
mystery
out
here
between
yourself,
and
so
this
is
an
advantage
of
your
monetary
identity
tree
that
is
you're
more
likely
to
have
contributors
right.
If
you
commit
your
club
medical,
it's
in
that
communities
before
we
go,
you
know,
you're
gonna
have
more
people
in
your
code
in
trying
to
contribute.
You're
gonna
have
more
exposure
to
the
ecosystem,
but
then
advantage
other
treat
is
that
you
have
like
flexibility
in
the
ownership
record.
Maybe
you
want
scriptural
ownership
of
your
codebase
or
you
have
a
cadence
of
leases.
A
A
Way
of
integrating
different
ideas
that
we
have
like
a
tree
by
the
tree
is
inconsistent,
and
this
inconsistency
sometimes
makes
it
harder
for
us
to
reason
about
like
certain
architectural
decisions,
or
it
needs
us
to
add
features
that
may
not
address
the
concerns
of
all
the
bodies
and
things
and
too
often
we
find
ourselves
supporting
use
cases
or
one
provider
that
may
not
be
used
by
any
other
provider
right.
That's
all
that
having
this
ability,
consistency,
piles
on
technical
debt
over
time,
I
think
what's
really
important.
A
Is
that
there's
a
strong
community
acknowledgement
on
this
issue?
I
think
courses
all
kind
of
see
the
same
sort
of
problems
regarding
with
how
we
supported
outliers,
previously
and
I,
think,
is
that
no,
this
split
and
entry
nugatory
is
not
intentional
I
think
it
was
just
kind
of
you
know.
It
just
made
sense
to
put
providers
next
to
with
everything
else.
Burnett,
yes,
and
so
we
start.
A
B
A
A
A
B
A
But
again,
we
really
need
to
be
thinking
of
this
migration
production
for
anybody,
and
if
we
look
at
kind
of
like
what
is
the
fallback
mechanism
right
need
to
plan
for
this
transitional
period,
where
we
can
run
both
the
core
parts
of
River
days
and
the
part
of
communities
in
either
integer.
So
this
is
this
select
users
within
the
same
communities
version.
B
A
Going
to
introduce
a
change
in
stage,
we
really
need
to
have
a
solid
and
it
gets
even
more
complicated
when
we
bring
storage
into
the
picture.
You
know,
we're
kind
of
the
proper
liners
out
are
also
moving
these
storage
plugins,
which
require
their
own
micro
scale.
So,
in
this
case
of
migrating
plugins
to
CSI,
we
need.
A
A
Want
to
do
this
with
up
walking
or
breaking
development
for
others.
The
last
thing
you
want
to
do
is
to
form
a
code
base
and
have
to
maintain
the
entry
other
treated
not
to
cubicles
channels
and
cigarette,
so
we're
trying
to
shift
more
responsibilities
of
a
system
to
these
external
plugins.
But
we
still
want
to
have
everything
under
like
one
single
source
period
and
when
we.
A
Smooth
transition
into
splitting
code
within
the
main
tree
out
to
their
owners,
so
there's
really
two
key
things
that
you
can
do
once
you
stage
elastic
rubber.
Ducky's
first,
is
that
you
can
actually
vendor
what's
staged
as
if
as
if
it
weren't
external
people
using
some
books.
So
from
the
perspective
of
the
main
tree,
it
would
be
as
if
you're
importing
this
external
package,
even
though
you're
still
on
the
same
entry
so.
A
A
A
A
lot
of
common
libraries
that
Clank
Oh
injection
area,
and
so
by
doing
this,
we
can
fill
both
the
entry
comm
providers
by
importing
what
we
have
in
the
stage
directories
all
without
changing
that
source
of
truth
by
we're
still
developing
in
the
state
directories.
But
then
we
can
still
build
our
controller
manager
of
the
external
switch
are
just
clone
directly
on
stage
and
so.
A
For
contributions
looking
for
ways
to
get
involved-
and
you
love
looking
with
go
defense
because
I-
don't
we
all
do
this
is
kind
of
this?
Is
the
issue
to
look
at
the
shoot?
Six,
nine,
five,
eight
five!
That's
what
that
meant.
So,
once
you
break
you
break
out
those
dependencies,
we're
looking
create
those
external
modules
and
we're.
A
A
A
A
A
A
Like
bomb
credit
solutions
like
they'll
care
about
that
right
and
then,
ideally
future,
if
you
want
to
be
in
a
place
where
we
keep
the
changelog
to
only
contain
what
its
development
or
what
is
considered
the
core
or
whatever
at
ease,
and
so
like
penguin,
I
do
this
were
able
to
actually
do
that
end.
If
a
proper
ladder
wants
to
change
is
making
just
blink
out
from
those
the
core
change
lock
and
tell
redirect
users
to
the
change
log
of
based
on
providers.
A
A
B
A
A
A
A
A
A
The
conditions
right,
some
providers-
will
let
you
shut
down
a
note,
but
some
providers
don't
anything
about
like
regions
for
systems.
But
some
providers
have
read
in
some
types
and
provide
examples,
and
we
have
a
lot
of
lightning
assumptions
around
what
should
be
supported
or
we
have
the
proper
things
where.
A
A
A
That
were
forcing
providers
to
gentlement,
but
also
mean
we
can
actually
like,
go
down
this
path
yet
because
we
have
to
break
a
lot
of
those
assumptions
we
have
and
select.
In
this
ideal
teacher,
we
move
everything
out.
We
can
start
having
meaningful
conversations
around
like
what
this
rate
might
look
like,
and
so
one
thing
that
I've
learned
to
working
this
effort
is
that
the
code
and
architecture
is
only
one
side
of
the
story.
There's
a
huge
community
aspect
of
this
effort
that
we
really
need
to
keep
a
strong
focus
on.
A
Row
or
16
document
or
like
the
actual
lava
fires
in
China
and
so
a
lot
like
our
technical
issues,
there's
a
tough
balancing
act
there,
and
so
one
of
the
things
that
were
struggling
with
this-
that,
just
like
we
can
at
every
single
provider
like
we
can't
just
keep
adding
a
safety,
because
that's
not
I
think
we
did.
This
we'd
have
more
cloud
providers,
things
that
we
do
like
other
states
right
so.
B
A
Facebook
were
exploring
what
the
new
governor
structure
and
relating
to
this
model,
where
you
want
a
single
state
representing
the
state
of
public
letters,
and
we
want
cloud
providers
joining
this
one
thing
projects,
and
there
are
many
motivating
factors
for
us
to
do
this.
Preventing
six
brawl
is
definitely
a
big
one,
but.
A
B
A
A
A
So
if
you
want
to
learn
more
about
some
of
the
work,
we're
doing
and
state
of
all
these
efforts,
I
highly
recommend
to
a
provider
intro
that
Christmas,
that's
tomorrow,
a
laid-back
young
and
if
you
really
want
to
contribute-
or
you
really
just
want
to
dig
deep
into
some
of
the
technical
challenges
that
we
have
in
there
sink
I
recommend
come
to
the
deep
dive
which
is
gonna
be
on
Thursday
and
with
that
I
want
to
say.
Thank
you
for
listening
to
my
talk.