►
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.
Cloud Native Apps with Server-Side WebAssembly - Liam Randall, Cosmonic
Server-side WebAssembly has the potential to increase security, extend application portability, and simplify cloud-native applications when operated in the Kubernetes ecosystem. This talk explores the pros and cons of different deployment models - embedded in a container, native execution, or embedded into other components. We will demonstrate the power that WebAssembly brings to even those projects hosted in traditional containers. This talk features a live build, compilation, deployment, and operation of reference applications. Featuring wasmCloud, Krustlet, and Envoy.
A
All
right
guys
we're
live
well,
let's
get
started
okay
well
good
morning.
Everyone
and
thank
you
so
much
for
coming
to
cloud
native
apps
with
server-side
webassembly.
My
name
is
liam
randall
and
I
have
been
in
technology
for
a
long
time.
I'm
a
serial
entrepreneur
and
investor.
I
focused
primarily
in
the
open
source
space.
I
did
prozek
and
then
os
query.
A
I
did
an
early
kubernetes
company
called
critical
stack
that
I
sold
to
capital
one
and
for
the
last
few
years
I've
been
obsessed
with
web
assembly
and
its
implications
for
the
cloud
I'm
a
hectaman
on
twitter,
I'm
on
all
the
slacks,
I'm
trying
to
be
very
accessible.
So
if
you
want
to
follow
up
with
me
afterwards,
don't
hesitate
for
a
moment
to
ping
me
directly,
so
we're
going
to
start
with
just
a
quick
little
summary
of
what
webassembly
is.
This
talk
is
not
designed
to
teach
you
all
about
the
details
of
webassembly.
A
Instead
we're
going
to
talk
about
how
you
can
use
webassembly
on
server-side
applications,
but
we
do
want
to
summarize
to
make
sure
we
bring
along
as
many
people
as
possible
with
us
along
for
the
journey.
So
webassembly
is
a
little
bit
like
some
things
that
people
have
tried
before
strike.
Stop
me
if
this
sounds
familiar,
you
can
write
your
apps
once
and
run
them
anywhere.
A
Of
course,
you've
heard
that
java
silverlight
flash
this
promise
has
been
made
many
times,
but
this
time
it's
different
with
webassembly
we
have
an
open,
wc3
standard,
that's
already
supported
on
the
devices
you
already
use
and
love
all
of
your
major
browser,
vendors,
it's
incredibly
efficient
and
fast,
and
it's
polygot
like
it's
not
like
java.
It
doesn't
come
with
its
own
programming
language.
It
allows
you
to
take
code
that
you
have
and
then
recompile
it
down
to
this
intermediate
format.
It
runs
everywhere.
What's
powerful
about
webassembly
is.
A
Is
that
these
little
engines,
these
wasm
runtimes,
can
be
tuned
for
different
circumstances.
So,
for
example,
you
might
want
to
have
your
code
running
on
big
silicon
in
the
cloud
and
you
can
have
a
webassembly
runtime
like
wasm
time.
That's
tuned
for
that,
or
perhaps
you
want
to
do
some
just
in
time,
compiling
or
ahead
of
time
compiling
out
on
the
edge,
and
you
want
to
use
it
a
runtime,
that's
tuned
for
low
power
or
low
memory.
A
A
Let's
talk
about
what
we
mean
when
we
say
cloud
native
when
we
talk
about
cloud
native,
we're
really
talking
about
this
third
epic
in
the
last
25
or
so
on
years
of
computing,
where
20
years
ago,
even
longer
than
that.
At
this
point,
25
years
ago,
people
started
to
realize
when
they
were
building
data
centers,
that
there
was
an
opportunity
to
be
more
efficient
and
virtualization
came
out,
and
that
really
was
accelerated
with
the
public
cloud
and
the
key
move
that
happened
there
is
is
that
the
security
boundary
moved
from
the
box
itself.
A
You
know
the
the
dell
server
the
ibm
server
up
to
the
cpu
and
there
were
a
lot
of
crazy
implications
in
that
and
then,
a
few
years
later,
we
sort
of
entered
the
epic
that
we're
in
now
with
containers
and
people
said
god.
You
know
I
used
to
build
these
images
and
it
would
have
you
know
raid
controller
firmware
in
it,
and
that
was
crazy,
and
why
am
I
doing
this
with
my
software
applications,
shipping
and
tire
operating
systems,
so
a
virtualization
layer?
A
This
picture
of
the
world
is
really
what
is
going
to
drive
the
next
20
years
of
change,
and
we
say
that
and
I
think
that,
because
containers
will
work
on
much
of
this
infrastructure
when
you
think
on
the
far
left
side
of
this
image,
we
have
the
central
public
cloud
and
on
the
far
right,
we
have
the
tiniest
devices
in
your
lives.
Tiny
watches,
refrigerators,
things
like
that.
A
I'd
point
out
is
your
web
browser,
which
is
probably
one
of
the
most
common
ways
that
you
interact
with
software
today,
never
mind
all
of
these
tiny
devices,
lower
power
and
memory
constrained
devices
these
are
are
not
these
will
not
work,
and
there
is
some
lower
bound
where
we
can
take
today's
technology.
So
what
are
the
problems?
What
are
the
additional
problems
that
we
have
with
today's
technology?
A
A
Think,
like
your
tesla,
you
want
real
time.
There
are
privacy
and
security
reasons
why
you
would
not
want
to
ship
personal
data
out
of
a
particular
environment.
There
are
times
when
you
want
liber,
limited
or
deliberate
autonomy,
think
like
drones,
or
things
that
go
offline.
Think
of
my
refrigerator,
if
it's
disconnected
from
the
public
cloud,
it
better
still
keep
things
cool
oftentimes
on
the
edge
is
where
the
data
is.
They
give
your
nest
doorbell.
It
doesn't
ship
all
of
that
data
back
to
the
central
cloud.
A
It
does
pre-processing
first
to
select
key
images
that
have
faces
and
other
things
and
then
ships
them
back
and
regulatory
reasons.
You
know,
gdpr
ccpa,
regula
regulation
is
an
invest,
move
right
now
we're
going
to
have
far
more
regulation
in
the
near
future
than
we've
ever
had
in
our
industry
in
the
past,
and
if
we
accept
that
the
future
of
applications
is
distributed,
then
we
also
accept
that
there
are
some
significant
problems
that
we're
going
to
have
to
deal
with
and
we're
going
to
talk
about
these
a
little
bit.
A
But
the
first
one
is
a
diverse
set
of
system
architectures.
The
second
one
is
is
that
the
application
architectures
themselves
will
be
different.
How
many
times
are
you
using
peer-to-peer
technologies,
for
example,
device
to
device
that
don't
even
transit
the
public
cloud
system
security?
How
do
we?
How
do
we
reuse
this
code
and
deploy
it
everywhere
and
keep
it
safe,
offline
use
and
machine
learning?
A
The
first
is
portability,
this
diverse
execution
environments
that
we
need
to
be
able
to
run
in
places
where
we
have
different
cpus
different
operating
systems,
or
even
neither
of
those
like
the
browser
where
we
just
sort
of
have
an
execution
environment.
The
second
big
reason
is
security
across
that
new
continuum
of
technology.
How
are
we
going
to
secure
this
and
lock
this
down?
Containers
are
scoped
to
a
linux
process.
A
How
do
we
take
a
security
model?
That's
scoped
to
a
linux
process
to
places
where
we
don't
even
have
linux
or
processes
for
that
matter,
and
the
third
big
reason
is
the
one
we're
going
to
spend
the
most
time
on,
because
I
think
that
this
is
the
opportunity
for
the
biggest
innovation
in
technology
period,
especially
when
coupled
with
webassembly
and
it's
the
tight
coupling
that
we
currently
make
between
applications
and
their
libraries
or
capabilities
that
they
use
we'll
tease
that
apart
in
a
moment,
let's
go
through
these
one
at
a
time.
A
So
this
first
challenge
is
portability.
We
are
on
the
way
to
50
billion
internet
connected
devices
by
2030..
That
is
a
phenomenal
number
and
collectively
this
represents
as
much,
if
not
more
computational
power
than
we
have
in
the
cloud
itself.
If
this
is
the
new
cloud,
then
we
have
a
whole
new
set
of
challenges
here.
But
what
I
would
emphasize
is
is
that
yes,
there's
a
story
around
arm
and
power,
pc
and
mips,
and
all
of
the
long
tail
of
risk
chips
that
you
find
out
in
today's
ecosystem.
A
But
this
is
also
a
problem
in
data
centers,
think
of
the
amd
graviton
64-bit
arm
processors
that
they've
launched
when
we
build
containers
today
we
make
an
implicit
assumption
and
we
target
a
specific
cpu
architecture
so
another
case
where
containers
don't
fail
and
if
you
suddenly
have
to
support
dozens
or
hundreds
of
different
architectures
having
an
intermediate
format
like
webassembly
is
a
huge
win.
So
for
this
first
big
challenge,
I
think
it's
clear
webassembly
brings
a
lot
to
the
table
and
it's
something
that
can
be
considered.
Let's
go
to
the
second
big
challenge.
A
What
about
the
security
of
these
things
now
this
idea
of
capability
driven
security
is
not
new.
Any
of
the
linux
networking
you
know
geeks
in
the
audience,
have
probably
used.
You
know
pseudo
setcap
before
to
give
privileges
to
snort
or
shirakata
or
bro,
or
something
to
monitor
a
network
interface.
But
it's
the
idea
that
the
children
no
longer
inherit
the
permissions
of
the
parents
and
webassembly
really
embraces
this
capability,
driven
permissions
in
that,
if
you
want
to
access
a
file
or
a
folder,
then
you
have
to
be
specifically
granted
to
it.
A
Webassembly
is
denied
by
default
and
you're,
becoming
increasingly
familiar
with
this
think
five
or
ten
years
ago,
when
you
installed
an
application
on
your
phone
it
just
installed
today.
It
actually
prompts
you
on
a
line
by
line
basis
for
access
to
the
video
camera
to
the
microphone
and
the
other
capabilities
that
same
model
is
coming
to
our
cloud
software
and
then
we're
going
to
extend
that
so
for
this
second
big
problem,
I
think
webassembly
is
another
great
fit.
It
brings
us
a
more.
It
brings
us
a
more
portable,
more
granular
security
model
that
runs
everywhere.
A
That's
feels
like
a
win-win.
It
runs
at
near
native
speed.
So
let's
carry
this
forward
a
little
bit.
What
about
this
third
problem?
You
know
in
in
french
there's
this
wonderful
word
elan
and
it's
to
translate
it
directly.
I
think
that
means
something
like
momentum,
but
it's
so
much
more
than
that.
It's
that
moment
right
when
the
bird
has
got
its
wings
flapped
and
it's
about
to
take
flight
or
when
the
dancer
is
spinning,
it's
her
leg
faster
and
faster
and
faster
and
picking
up
momentum
right,
that's
the
elan,
and
what?
What
really?
A
I
think
is
a
big
opportunity
to
ride
this
webassembly
excitement
is
to
consider
this
tight
coupling
here,
because
this
tight
coupling
that
we
have
acts
as
an
anchor
that
holds
us
in
place
if
you're
building
for
a
specific
cloud
provider
and
you're
using
specific
cloud
provider
services.
Who
cares
how
portable
your
code
is,
if
you're
stuck
in
that
particular
location?
A
So
these
are
the
capabilities
that
you
would
use
to
satisfy
all
sorts
of
things
in
your
app
if
you're,
building
an
application
to
calculate
loan
interest
or
you're
building
some
social
media
scoring
or
you
know
any
kind
of
business
logic
that
you
want
to
put
into
it.
You
have
to
typically
couple
that
with
a
large
amount
of
non-functional
requirements
in
the
enterprise,
that's
just
extreme
and
you
have
to
start
coupling
it
with
security
and
logging
and
other
sorts
of
tools
and
the
dirty
secret
of
enterprise.
A
Application
development,
especially
on
the
server
side,
is,
is
that
the
vast
majority
of
the
application
code
base
is
someone
else's
code.
It
is
nearly
criminal
that
in
today
modern
cloud
native
app
development.
You
start
each
new
application
with
20
000
lines
of
someone
else's
code,
and
there
are
some
devastating
impacts
to
that.
The
private
studies
that
I've
seen
in
some
enterprises
was
even
higher
as
many
as
as
much
as
99
of
their
template.
A
Some
java
spring
boot
application
was
non-functional
requirements
that
they
were
managing
and
those
non-functional
requirements
act
as
a
weight
that
holds
you
in
place
there
now
across
this
microservice
lifecycle.
When
you
accept
all
of
those
non-functional
requirements,
you
end
up
with
some
pretty
devastating
impacts.
The
first
is
is
that
you
have
an
inability
to
move
things
from
design
quickly
out
of
poc
into
production,
because
there's
completely
different
assumptions
that
are
made
that
are
non-functional.
A
So
this
tight
coupling
that
we
see
at
the
top
layer
really
becomes
one
of
the
biggest
challenges
that
we're
facing
in
modern
cloud
native
development
and
with
web
assembly
today
and
its
adoption
to
meet
the
first
two
challenges
there.
It
comes
a
huge
opportunity
as
well
to
solve
for
the
third,
so
I'm
one
of
the
co-creators
of
an
open
source
cnc
framework.
A
That's
called
wasmcloud
and
it's
both
compatible
with,
but
not
dependent
upon
the
current
cloud
native
stack
and
we're
going
to
look
at
a
couple
demonstrations
of
running
things
all
together,
while
also
running
for
the
edge,
and,
as
I
mentioned
before,
this
is
not
the
only
approach.
There's
microsoft
has
put
out
a
wonderful
application
layer,
a
tool
that
is
directly
aligned
to
cloud
native.
It's
a
sort
of
depends
on
containers
in
the
ecosystem,
so
it
would
meet
the
first
left
half
of
the
cloud
native
world,
but
doesn't
solve
for
the
far
right
half
here.
A
A
So
now
this
isn't
even
a
new
idea.
It's
a
little
different
to
start
thinking
from
the
top
down
to
start
at
the
business
logic
and
work
your
way
down.
But
to
me
this
is
the
next
obvious
step
in
the
continued
epics
of
technology
and
when
you
think
about
the
landscape
today,
we
already
have
layers
that
are
doing
this
for
us
just
in
a
different
domain.
A
For
example,
on
top
of
kubernetes
we
could
layer,
istio
and
istio
would
just
remove
some
of
the
network-related
concerns
such
as
traffic
routing
and
role-based
access
control,
and
on
top
of
that
we
could
even
use
dapper,
which
will
give
us
this
application
layer
concerns
like
the
key
value
store,
but
wasm
cloud
takes
it
even
further
and
really
turns
your
business
logic
into
these
composable
actors.
It's
more
of
a
shift
and
we
couple
it
with
both
elixir
on
the
back
end
for
extreme
scale
and
for
and
with
gnats.
A
So
you
get
a
networking
layer
on
the
front
end.
So,
let's
slide
into
our
first
demo.
Now.
This
is
an
incredible
demo
that
completely
surprised
me.
It
came
out
of
nowhere
and
I'm
only
going
to
give
you
the
summary
of
this
one,
because
the
full
talk
is
definitely
worth
your
time.
This
was
showed
at
wasm
cloud,
I'm
sorry
a
wasm
day
cloud
native
wasm
day
yesterday,
and
this
was
a
group
called
red
badger
and
they
did
this
on
behalf
of
a
large
financial
service
and
to
help
set
the
stage
a
little
bit.
A
This
organization
had
a
number
of
service
interruptions
with
some
of
the
public
cloud
outages
over
the
last
year
and
they
have
their
regulators
asking
them.
When
can
they
go
multi-cloud
so
what
they
do
in
the
demo?
The
source
code
is
linked
right
here.
If
you
want
to
go
ahead
and
grab
a
quick
link,
you
can
spin
this
up
yourself,
but
I'll
set
up
the
scene
a
little
bit
further
in
this
demo.
A
A
A
A
It
doesn't
have
all
the
components
in
the
microservice
to
complete
it,
but
the
sort
of
the
technology
here
is
going
to
auto
fail
right
over
to
aws
and
as
we
watch
the
request
come
in
here.
This
really
is
a
very
interesting
demo.
So
I
kind
of
wanted
to
summarize
this
one
for
you
and
encourage
you
to
watch
this
when
it
hits
youtube
in
two
weeks,
13
days
from
today,
we'll
tweet
this
out
off
of
our
accounts.
A
A
The
cncf
nats
essentially
gives
us
the
ability
to
create
these
super
super
nodes
and
leaf
clusters
here
so
in
when
you
can
satisfy
a
request
locally,
these
requests
are
being
routed
locally
and
in
that
previous
demo,
we
were
running
on
kubernetes.
By
the
way
I
forgot
to
mention
that
detail.
My
apologies
so
we're
running
on
two
different
kubernetes
distributions
in
the
previous
demo,
but
at
the
application
layer
we
now
have
the
ability
to
do
this
auto
discovery
and
resolution
outside
of
our
local
leaf
nodes.
A
If
we
can't
meet
our
make
our
constraints
there,
what
it
means
to
you
as
a
developer
is
that
you
no
longer
have
to
think
about
clouds.
You
no
longer
have
to
think
about
topology.
In
fact,
your
business
logic
is
completely
separated
from
your
topology
and
your
topology
can
change
at
runtime
we're
going
to
demonstrate
that
in
a
live
demo.
We
do
in
just
a
second
here
so
in
this
next
demo.
What
we're
going
to
do
is
we're
going
to
do
something
similar
to
that
except
we're
going
to
deploy
running
code
to
people
in
the
audience.
A
So
taylor,
are
you
connected
to
the
wi-fi,
so
taylor
up
here
is
is
running
a
wasm
cloud
as
well,
and
what
we're
going
to
do
is
we're
going
to
demonstrate
that
same
ability
to
fail
over
here.
Okay,
so
all
right.
So
this
is
my
local
computer
here
and
I'll
just
explain
the
dashboard
a
little
bit
and
there's
all
of
course
works
via
api
and
through
a
command-line
cli
called
wash.
Everything
in
the
webassembly
world
is
some
sort
of
a
wah
joke.
You
know
it's.
A
This
is
the
washboard
there's
wash
there's
all
sorts
of
wad
jokes
that
are
out
there
and
we're
essentially
running
the
java.
A
pet
store.
Has
anybody
ever
written
the
java
pet
store
before
no
one
all
right?
This
is
like
a
when
you
get
into
java
programming.
This
will
be
one
of
the
sort
of
demos
that
the
books
always
use.
It
starts
simple
and
basic,
and
it
gets
increasingly
and
increasingly
complex
here
and
logically,
what
we
have
here
is
an
api.
That's
fronting
three
other
different
actors
that
are
here
now.
A
These
actors
have
capabilities;
they
have
lego
blocks
on
the
bottom
and,
what's
great
is
that
you
can
plug
them
into
lego
blocks
anywhere
so
locally
they're
plugged
into
an
http
web
server
and
into
a
postgres
database,
and
that's
what's
happening
here
with
these
link
definitions
now
I
can
also
see
that
there
are
some
other
hosts
here
that
are
sitting
on
the
on
the
network
and
taylor.
What
is
your
id?
A
A
Okay,
so
he
said
that
there's
nothing
but
wasm
cloud
running.
That's
on
his
computer.
You
can
see
that
taylor
isn't
isn't
run
anything.
So
let's
go
ahead
and
start
part
of
the
application
here
on
taylor's
computer
and
we'll
go
ahead
and
start
vets,
which
is
a
piece
of
this
app
and
I'm
just
going
to
go
and
kick
that
off
and
I'll
give
you
a
demo
of
what
this
app
actually
looks
like
if
you're
not
familiar
with
it.
A
First
diane
edwards
add
visit
there.
We
go
all
right,
so
we
can
see.
Snoopy's
go
has
been
to
the
vet
now.
What's
really
powerful
here
is,
is
that
this
is
all
part
of
this
somewhat
amorphous
mesh
now,
so
you
can
see
taylor's
running
a
part
of
the
vets
app
and
if
I
come
back
to
my
computer
right
here,
you
can
see
that
I'm
running
it
all
as
well,
and
we
can
it.
You
know
beyond
using
the
ui.
We
could
also
you
know
curl
this.
You
guys
probably
can't
see
that
very
well.
A
If
you
can,
let
me
in
the
back
know
when
the
text
is
big
enough.
Great.
Thank
you
so
much
great,
so
I
can
just
curl,
and
I
can
see
that
I
could
hit
that
api
now.
What
would
happen
if
we,
you
know,
delete
this,
and
this
is
the
you
know
the
the
live
demo
right.
So,
let's
see
what
happens
when
we
delete
it,
so
I'm
no
longer
running
that
piece
of
this
application
here
locally
and,
let's
see
fingers
crossed.
A
Think
about
what
we
just
demonstrated
there
simply.
We
have
two
completely
disconnected
clouds,
there's
no
vpn
between
our
between
our
devices.
Here
right,
I
could
be
running
on
kubernetes
or
not.
I
could
be
on
the
edge
or
not.
I
couldn't
even
define
the
edge
to
be
honest
with
you
right.
There
is
no
definition
of
that
of
that
term
anymore,
because
there
is
no
real
definition
of
cloud
anymore,
because
all
of
the
old
boundaries
have
simply
fallen
to
the
wayside.
A
A
A
A
However,
they
now
can
stay
in
their
own
swim
lane
and
stay
out
of
ours.
This
is
a
huge
opportunity
in
in
cloud
native
computing,
and
what
I
would
want
to
emphasize
is
is
that
there
are
multiple
approaches
to
this.
You
know
I'm
demonstrating
in
wasm
cloud
today,
because
that's
the
technology
that
I
work
on,
but
dapper
is
also
moving
in
this
direction
as
well.
A
So
with
our
last
couple
of
minutes
here,
let's
talk
about
some
of
the
things
that
are
working
together
to
make
this
happen.
So
wasn't
cloud
is
a
cncf
project.
We
started
it
about
two
years
ago
and
it
is
a
stack
of
components
that
give
you
the
ability
to
develop
very
simply
and
down
at
the
bottom
layer.
It
starts
with
web
assembly,
which
I
like
to
think
of
as
just
a
virtual
cpu.
A
You
know
when
you're,
when
you
compile
your
code
to
webassembly,
there's
no
virtual
operating
system
here
now
there
are
some
operating
system-like
constructs
on
the
way
called
wazi
and
when
those
come
along,
you'll
sort
of
have
the
ability
to
you
know
open
sockets
and
do
some
of
those
other
things,
but
right
now.
This
is
just
simply
like
a
box
that
you're
locked
into
and
what
wasn't
cloud
does.
Is
it
comes
along
and
it
brings
a
framework
for
your
application.
A
It's
an
application
framework
that
gives
you
a
box
of
legos
and
takes
care
of
a
lot
of
the
hard
underlying
tasks.
For
example,
when
you
want
to
move
data
out
of
your
box
out
of
webassembly
into
the
guest
operating
system
and
back
it
does
all
that
translation
for
you,
even
with
complex
data
types,
so
it's
designed
for
extreme
scalability,
both
horizontally
and
vertically,
as
you
just
saw
right
now,
I
could
spun
up.
A
I
could
have
scaled
my
acres
locally
to
you,
know
100
actors,
or
I
could
scale
horizontally
out
and
out
and
across,
and
we
leverage
this
built-in
networking
layer
called
nats
also
open
source
in
order
to
get
that
done.
So
this
is
extremely
scalable
and
composable
and,
of
course,
signed.
It
also
inherits
the
capability
driven
model
that
a
webassembly
starts
with
now.
On
top
of
that,
we
essentially
maintain
a
library
of
hot
swappable
capabilities.
A
Those
are
those
lego
blocks
like
web
servers
and
database.
You
know
driver
type
things
like
sql
servers,
message
queues,
but
these
capabilities
are
very
easy
to
add
and
we
have
a
new
developer
experience
with
wasmcloud
1.50,
which
we
just
launched
last
week
that
give
you
wizards
in
order
to
scaffold
these
things
out
and
get
them
built
out.
A
A
That's
that's
been
developed
over
the
last
20
years
to
implement
things
like
whatsapp,
initially
and
discord
it's
incredibly
scalable
and
powerful,
and
then
we
combine
this
with
nats,
which
is
the
magic
layer
that
just
enabled
that
seamless
connection
between
these
two
leaf
nodes
here
when
the
vets
app
could
satisfy
its
request
locally
here,
just
like
we
did
in
the
aws
gcp
demo.
It
did
so,
but
when
it
couldn't
it
bid
the
contract
out
over
to
taylor's
computer
and
then
pulled
it
back,
and
that
is
incredibly
amazing
for
cncf
projects.
A
It's
self-healing,
it's
self-forming,
there's
no
service
discovery
and
we
just
adopted
a
new
technology
called
jet
stream.
That
sort
of
gives
you
a
local
cache
of.
What's
on
the
network
that
that
works
there
now
when
we
even,
I
just
want
to
mention
that
when
we
think
about
what
this
topology
and
what
this
technology
enables
it's
going
to
enable
some
pretty
crazy
things.
A
Just
as
I
was
able
to
seamlessly
offload
part
of
my
microservice
and
make
topology
detail
with
taylor
here,
you
can
do
that
across
clouds
as
well.
There's
no
reason
that
you
can't
have
part
of
your
application
running
in
one
cloud
and
part
of
your
application
running
in
another
cloud
or
on
the
edge
itself.
A
A
This
is
the
sort
of
summary
view,
though,
of
where
I
think
we're
going
for
the
next
epic
of
technology
and
when
we
look
at
this
world
view,
it
really
involves
this
crisp
layer
here
in
separation
between
your
applications
and
what
sits
underneath
them,
your
business
logic
and
everything
else,
and
it
really
starts
to
meet
the
sort
of
needs
that
we
have
today
around
tiny
composable,
portable
business
logic
that
we
can
take
with
us
anywhere
in
a
way
that
the
current
epic
of
technology
just
doesn't.
So
when
we
think
cloud
native.
A
I
just
want
to
remind
you
that
you
know
webassembly
is
neither
you
know,
web
nor
assembly,
but
cloud
native
is
really
kind
of
getting
undefinable,
as
well
as
those
boundaries
kind
of
fall
away
on
the
edge.
Thank
you
so
much
for
your
time
today.
My
name
is
liam
randall
wasmcloud.com
and
then
for
more
information
about
our
company.
It's
cosmonic.com.