►
From YouTube: Keynote: And Here We Go: Dual-stack Networking in Kubernetes - Lachlan Evenson, Microsoft Azure
Description
Don’t miss out! Join us at our next event: KubeCon + CloudNativeCon Europe 2022 in Valencia, Spain from May 17-20. Learn more at https://kubecon.io The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects.
Keynote: And Here We Go: Dual-stack Networking in Kubernetes - Lachlan Evenson, Principal Program Manager, OSS at Azure, Microsoft Azure
A
Hello,
I'm
lockheed
evanson
with
microsoft
azure.
If
this
is
your
first
kubecon
cloud
native
con,
welcome
dual
stack.
Work
for
kubernetes
began
in
late
2017
and
if
you're
at
the
last
in-person
kubecon
in
san
diego
autumn
of
2019,
you
may
be
wondering:
wasn't
there
already
a
keynote
about
dual
stack?
Are
we
actually
living
in
a
time
warp,
because
that
would
explain
a
lot
spoiler
alert,
no
time
warp?
No
time
travel.
A
A
A
Back
then,
a
kubernetes
service
could
only
have
a
single
ip
family
so
to
create
a
dual
stack
service.
You
needed
to
create
two
services,
one
ipv4
and
one
ipv6
turns
out
end
users
did
not
want
to
do
this,
finding
that
out
early.
Well,
that's
the
beauty
of
alpha
sure,
minimal
change
to
the
minimal
change
to
this
services.
Api
was
the
fastest
way
to
allow
dual
stack,
and
only
after
exploring
that
path
did
we
fully
realize
that
the
complexity
tax
on
users
was
way
too
high.
A
A
That
brings
us
to
the
second
lesson.
The
kubernetes
project
is
vast
and
it
seems
hard
to
make
big
changes.
The
process
is
too
hard.
It's
just
too
much.
The
reality
we
experienced
is
that
yes,
change
is
hard
but
possible
reach
out
to
those
not
so
horrible
geese
wandering
around
in
the
community.
A
Community
collaboration
is
the
strength
of
kubernetes
we
iterate
and
learn.
We
make
big
changes
in
the
case
of
the
services
api,
the
dual
stack
pr
pull
request.
We
had
eleven
thousand
and
eighty
eight
lines
added
and
three
thousand
four
hundred
and
thirty
two
lines
of
code
removed
from
the
networking
substrate
of
kubernetes
itself.
A
This
was
definitely
look
ma,
no
hands
moment
riding
on
a
bike,
hoping
that
we
don't
fall
over
with
a
lot
of
eyes
on
alpha.
We
were
ready
to
go
to
beta
and
I
say
beta
you
say
beta,
I
say
tomato
you
say.
Tomato
and
1.21
beta
features
in
kubernetes
are
actually
on
by
default.
Don't
tell
your
mum
and
hey
the
networking.
Infrastructure
has
dual
stack
where
the
feature
gate
is
enabled
on
a
given
cluster
or
not,
and
surely
the
code
was
perfect
from
the
get-go
yeah
right
yeah.
A
Of
course
not
the
community
found
and
fixed
bugs
together
and
the
dual
stack
feature
is
slated
to
go
stable
in
the
upcoming
1.23
release
round
of
applause
for
the
community.
Thank
you,
the
third
and
final
lesson:
it's
really
hard
to
land,
an
enhancement
in
a
single
release,
but
you
don't
have
to
pay
attention
to
the
release
schedule.
A
The
timing
for
when
you
land
your
cap
and
your
code
matters
you'll,
need
a
production,
readiness
review
and
testing
and
docs
the
release
team.
Is
us
it's
you
ray.
It's
me
with
a
little
upfront
planning.
We
can
all
be
that
much
happier,
it's
worth
remembering
why
we
impact
embarked
upon
this
journey
in
the
first
place,
I
p
addresses,
are
not
infinite,
believe
it
or
not.
Public
ipv4
addresses
have
been
largely
exhausted
and
when
you're
talking
about
the
flat
network
address
topology
of
a
kubernetes
cluster,
even
ipv4
addresses,
are
going
to
run
low.