►
From YouTube: Keynote: Containerful Serverless - Roland Huss, Senior Principal Software Engineer, Red Hat
Description
Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon Europe 2023 in Amsterdam, The Netherlands from April 17-21. 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: Containerful Serverless - Roland Huss, Senior Principal Software Engineer, Red Hat
Over the last years, the serverless landscape was dominated mainly by platforms with a proprietary development and deployment model like the functions offerings from the big cloud players. But, in parallel, the container revolution changed how we package and run our applications in a standardized and portable way. So it stands to reason to combine both movements to get the best of both worlds. This session gives an overview of the rise of "containerful serverless" platforms and the role that Knative has played and plays in this evolution.
A
So
thank
you
and
welcome
to
canadifcon
I'm
happy
to
be
here
happy
to
welcome
you
all
and
thanks
a
lot
Evan,
it's
a
great
introduction
and
to
look
into
K
native
now.
We
want
to
go
a
little
bit
in
the
other
direction
and
would
and
look
like
and
see
how
Canada
fits
into
the
broader
landscape
of
serverless
and
fast
land
fast
platforms
actually,
but
before
I
start
I
want
to
clarify
what
I
do
mean
with
serverless
and
fast.
A
You
probably
know
that
serverless
and
files
are
probably
one
of
the
two
terms
which
are
the
most
diffuse
and
overviews
passwords
over
the
last
decade.
I
would
say
so.
My
definition
for
serverless
and
fast
is
like
that.
Serverless
is
a
deployment
platform
that
abstracts
away
the
infrastructure
while
fast
or
functional
service
is
a
programming
model
which
runs
on
top
of
serverless.
So
please
keep
that
in
mind,
while
I'm
presenting
the
the
next
slides,
because
I
might
intermix
both
of
them.
A
A
A
Here
we
go
and
kind
of
epoch,
so
we
start
with
something
which
I
called
prefas
or
something
like
that.
So
this
is
kind
of
the
transition
from
platform
s,
and
these
are
all
these
selection
that
I'm
going
to
show.
You
is
kind
of
my
personal
thing,
so
this
is
an
opinionated
approach,
how
I
timeline
the
history
of
serverless
and
fast,
but
you
see
that
AWS,
Beanstalk
and
Google
app
engine
work
first
kind
of
way
how
you
could
throw
over
your
code
to
the
cloud
and
then
run
it
somehow
right.
A
So
these
offerings
are
really
odd.
So
I
was
quite
surprised
that
that
old
they
still
exists
today,
of
course,
but
probably
in
a
different
format.
So
in
different
okay.
But
now,
let's
switch
over
to
the
to
the
next
epox
and
then,
of
course
they
are
the
big
bang
in
2014
with
AWS
Lambda,
so
Lambda
came
in
and
actually
Lambda
was
scripting
extension
for
S3
buckets
so
that
you
can
hook
in
your
business
code
when
something
changes
in
a
bucket
like
that.
A
But
then
it
turned
out
that
this
kind
of
glue
code
is
really
very,
very
useful
to
connect
Services
together,
and
this
and
the
success
of
AWS
Lambda
really
came
also
with
this
whole
set
of
services
that
are
around
of
and
that
Lambda
could
connect
to
them
and
in
in
the
wake
of
AWS
London.
There
are
a
lot
of
other
stuff
has
been
emerged
and
there
has
been
some
tooling,
like
serverless
serverless
framework,
which
adds
a
nice
user
experience
on
top
of
AWS
and
also
other
platforms.
A
But
AWS
Lambda
is
one
of
the
the
most
prominent
ones
and
then
also
some
indication
for
this,
of
course,
but
a
lot
of
the
competition
competitors
of
the
hyperscalar
space
Also
created
functions
offerings
a
little
bit
later.
So
this
is
a
really
big
to
edit
quite
fast
after
AWS
Lambda,
and
the
problem
actually
is
not
that
they
are
that
you
have
the
choice.
Now
that
you
can
really
choose
the
platform
that
you
want
to
run
your
functions
on,
it's
just
that
they
are
all
kind
of
proprietary
formats.
A
So
functions
as
I
said,
is
a
programming
model,
so
it
defines
a
function.
Signature
and
these
are
kind
of
all
different
offerings
and
they
kind
of
compete.
But
if
you
stuck
select
one
then
you're
stuck
to
the
ecosystem
and
then
also
to
the
services
that
that
you
want
to
connect.
Of
course
you
can
connect
with
the
Google
Cloud
function
or
the
AWS
services,
but
at
the
end
you
kind
of
decide
first,
which
platform
to
use,
and
then
you
stick
to
that.
A
So
this
was
this
state-of-the-art
at
the
beginning
of
2018
and
then
total
in
parallel
pollution
happened,
which
is
the
race
of
Docker
and
the
containers
of
docker
I.
Don't
have
to
tell
you
about
Dr
Patoka,
it
just
remixes
a
set
of
existing
Linux
Technologies
together
and
adds
on
top
of
that
my
gist
of
Docker
itself,
and
then
we
have
kubernetes,
then
a
little
bit
later
in
2014.
Also
don't
talk
about
kubernetes
so,
but
you
know,
of
course,
actually
it's
on
top
of
daughter
and
for
great
orchestration
stuff.
A
So
so
we
have
this
now,
these
two
streams
of
serverless
offerings
in
the
cloud
and
then
also
Docker
and
kubernetes
and
and
Docker
really
took
over
the
I.T
World
in
storm.
So
it
was
a
big
kind
of
the
dominant
deployment
platform
for
your
applications
and
then
of
course,
people
started
to
combine
those
so,
but
even
in
the
in
the
hyper
scale
that
they,
the
offerings,
were
a
little
bit
different
and
separated
like
that.
But
then.
A
Framework
that
that
one
on
top
of
kubernetes
but
then
also
be
some
combinations
like
AWS,
that
used
as
a
back
end
and
gives
you
the
impression
of
an
infinite
size,
Docker
demon,
something
like
that
and
the
same
is
true
for
sensors.
So
it's
really
abstracts
away
the
infrastructure
that,
in
that
case,
that
you
really
think
about
that.
There's
a
talk
of
demon,
but
you
do
not
operate
that.
A
A
A
What's
up,
okay,
then
of
Canada,
if
you
know,
has
been
initiated
by
Google
but
has
been
backed
by
many
vendors
from
the
very
beginning.
So
there
was
VMware
IBM
redhead,
but
the
players
in
all
those
based
some
products
on
top
of
K
native.
But
the
good
thing
is,
and
even
these
products
are
running
as
managed
Services,
some
of
them
some
are
based
on
internal
or
any
kubernetes
distributions
like
Reddit
on.
A
Okay,
so
yeah,
but
the
good
thing
is
that
if
you
run
one
application
with
Kinetic
of
one
of
these
platforms,
you
can
then
move
them
on
to
another
platform.
Of
course
the
devil
lies
in
the
details:
it's
not
automatically
possible,
always,
but
at
least
it's
much
better
than
all
the
awarded
Gardens
that
you
have
seen
before.
A
They've
also
offers
some
something
similar,
so
they
like,
for
example,
Asia,
container
apps
or
AWS
everyone
they
all
in
the
deal
is
also
simplified
application
deployment
model
that
which
makes
kind
of
famous
that
it
really
makes
easy
to
bring
your
application
with
the
cloud
as
well
as
really
a
very
sophisticated
and
capable
Auto
scaling
mechanism.
So
all
these
is
already
happened
there.
A
A
A
Sorry
then
we
have.
What
I
would
could
see
is
really
that
the
kubernetes
control
plane
itself
could
be
based
on
a
serverless
model,
especially
if
it
for
optional
features
of
kubernetes,
so
that
you
do
not
have
to
pay
something
for
features
that
you
are
not
using
on
kubernetes
and
saved
I
think
this
would
be
a
nice
way
to
explore
and
subsequent
code
stop.
This
is
really
kind
of
the
dream
that
we
all
have
somehow,
but
it's
still
not
possible,
so
it's
possible
in
certain
cases,
but
it's
hard.
A
Let's
say
like
this,
and
finally,
no
not
finally
but
schedule
cereal
is
a
core
kubernetes
feature.
This
would
be
also,
of
course,
I
would
say
something
we
should
could
strive
for
and,
finally,
I
think
that
a
hybrid
serverless
are
running
serverless
on
multiple
clusters,
with
a
single
control
plane
that
combines
anything.
It's
one
of
the
the
hottest
topics
these
days
as
well,
and
with
that
I
say
thank
you
and
yeah,
and
so
sorry
for
the
audio
issue.
Center
thanks
foreign.