►
From YouTube: Application Networking Day Session #1: Future of Application Networking with Istio, Cilium & Envoy
Description
Featuring Idit Levine. The last 5 years have been all about building Cloud-native 1.0, deploying Kubernetes, understanding how to containerize applications, and managing daily updates. Those early successes are now leading to a new set of challenges in scalability, security and observability. We're now moving into the Cloud-native 2.0 era, which will require us to apply new technologies and architectures to solve bigger challenges. Learn how the evolution of Istio, Envoy and Cilium will play a critical role in this next stage of cloud-native applications.
A
A
A
There
is
a
few
things
in
solar
that
is
pretty
unique
in
this
ecosystem.
The
first
one,
which
is
really
really
important,
is
the
fact
that
honestly-
and
it
will
be
very
blunt-
we
have
a
lot
of
customers.
That's
really
really
helpful
because
we
saw
it
before
we
saw.
We
took
a
lot
of
a
team
and
basically
made
them
production
in
STL
in
Envoy,
with
Gateway
with
psyllium,
and
that's
exactly
what
what
is
very
strong
on
us.
A
The
second
thing
that
is
very
important
is
that
we
are
a
little
technology
leaders
in
that
in
that
air
project
and
the
last
one
is
that
we
also
not
enough
for
us
just
to
be
the
technical
leaders.
We
also
think
that
those
projects
are
really
really
challenging,
sometimes-
and
it's
very,
very
important
to
us
to
be
the
education,
because
it's
not
enough
to
give
you
the
bit
and
buy
that
you
were
running
there.
A
It's
also
very
important
to
us
that
we
will
know
how
to
operate
that,
let's
give
a
lot
a
lot
of
attention
to
some
of
the
adventures.
Therefore,
we
are
really
really
really
well
funded.
We
buy
over
billion
dollar
and
we
add
over
135
million
dollars
some
so
that's
kind
of
like
in
very
very
high
level.
Now
the
question
is:
what
are
we
trying
to
do
so?
A
You
know
we
are
all
here
in
Groupon,
which
means
that
we
basically
really
passionate
about
kubernetes
and
that's
where
everything
gets
started
and
that's
what
we
like
to
call
Cloud
native
1.1.0.
That
means
that,
basically,
you
know
you
started,
you
took
a
kubernetes,
it's
making!
You
really
really
more
government
very
fast
and
more
efficient
container
microservices,
but
in
the
nutshells
it's
also
introduced
quite
a
lot
of
problem
and
we
basically
we're
focusing
on
what
called
Cloud
native
2.0
and
Cloud
native
2.0
is
basically
the
motivation
of
the
API
management.
A
The
second
thing
that
is
really
really
important
is
you
know
we
all
about
the
clarity
we
all
about
eventually
consistent
and
therefore
we
are
not
a
big
believer
in
a
big
UI
that
you
click
in
and
basically
driving
configuration
in
production,
we've
way
more
believe
in
githubs.
We're
really
excited
about
you
know
now,
when
your
infrastructure,
there
is
a
lot
of
micro,
Services
everywhere.
That
also
means
that
you
don't
really
know
all
the
time
where
the
request
is
flying,
because
you
have
a
replication
of
each
of
those
micro
services.
A
A
The
last
one
is,
you
know,
multicluster
a
lot
of
people
that
we
know
started
with
one
cluster,
maybe
two
clusters
growing
growing
growing
to
hundreds
of
clusters
and
thousands
of
clusters,
and
the
question
is
that
how
do
you
manage
that
right?
You
can't
manually
go
to
each
of
the
classes.
I
mentioned
that
it
works
the
last
and
not
least,
you
know
new
technology
like
serverless
and
graphql.
We
think
that
it's
really
really
interesting
technology
and
it'll
be
very
nice
to
see
how
it's
going
to
work
with
the
current
technology
that
we
have.
A
So
this
is
where
solo
is
focusing
right.
So
basically
we
have
a
platform
blue
platform.
There
is
three
building
block
to
this
Platform
One
for
Gateway.
It
will
build
natively
on
top
of
STL,
so
it's
basically
taking
their
steering
grass
and
enhance
that
to
a
full-blown
API.
The
second
one
is
blue,
get
blue
mesh,
which
is
basically
a
distrove
as
steel
and
what
it's
doing,
it's
basically
enhancing
it
for
multi,
a
multi-cluster
behavior
as
rotation
and
other
tools
to
help
you
with
production
is
successful
and
the
last
one
is
blue.
A
So
you
know
I,
like
always
to
add
another
layer
here
and
that's
the
graphql
layer
calling
it
multiplexer
and
the
reason
is
because,
if
you're
looking
at
those
layer,
the
HTTP
one
is
basically
layer,
seven,
but
what
we're
doing
with
graphql
is
obstructive
right
HTTP.
So
it
doesn't
really
make
sense
to
put
it
on
the
same
layer.
So
this
is
the
OSI
model.
There
is
the
you
know
the
hard
one
one,
and
then
there
is
all
the
others.
Each
of
them
is
very,
very
nicely
light.
A
You
an
open
source
project
right,
you
know
the
graphql
or
sto
is
taking
everything
from
layer
four
to
seven
and
psyllium
layer.
Three
and
four
I
think
that
what
we
are
doing
is
basically
obstructing
all
of
this
and
making
it
very
very
easy
to
consume.
So,
basically,
it's
a
platform
to
basically
driving
those
open
source
projects.
A
So,
as
I
said
three
building
block,
you
don't
have
to
use
them
all
right.
We
have
a
lot
of
people,
starting
only
with
the
match.
There
is
a
lot
of
people,
starting
only
with
the
Gateway
but
they're
beautiful.
It's
one
platform,
kind
of
like
like
in
engineering.
We
took
everything
that
is
common
between
these
things,
everything
that
related
to
multi.
You
know
to
the
piping
of
the
multi-cluster
everything
that's
related
to
you,
know
the
UI
to
the
CLI
and
basically
create
it
as
a
common
common
component.
A
Do
all
that
platform,
and
now
everything
is
basically
a
license
that
you
can
enable
so,
for
instance,
if
you're,
starting
with
the
Gateway
and
you
want
to
enhance
to
the
magic,
it's
just
a
license,
you
don't
need
to
redeploy
anything.
You
don't
need
to
do
anything.
It's
already
there,
which
is
pretty
pretty
unique,
you'll,
be
in
someone
really
really
excited
about.
A
While
we
were
doing
it,
I
was,
you
know,
we're
doing
it.
We
did
in
Google
and
so
on.
We
discovered
that
Google
is
doing
the
same
thing
in
in
by
themselves
and
we
decided
to
join
Air
Forces
eight
months
ago
and
we
just
announced
a
months
ago
Ambien,
which
is
basically
a
mod
in
service
in
SEO,
so
it's
sto
with
a
mode
that
is
basically
getting
rid
of
the
cycle.
It's
helping
you
a
lot
with
adoption,
because
it's
basically
you
can
adopt
it
a
gradually.
A
So
if
you're
looking
at
what
it
is,
basically
you
know
there
is.
There
was
some
complexity
or
not
complexity,
of
a
a
hiccup
in
actually
adopting
sto
before
and
with
the
current
sector
model,
and
what
we
did
is
basically
try
to
make
it
better.
We
did
it
by
basically
separate
layers,
four
and
layer.
Seven
and
again
there
will
be
a
lot
of
talk
about
it.
So
I'm
really
really
talking
a
high
level,
but
in
the
natural
it's
possible,
it's
helping
us
with
a
lot
of
stuff.
A
It's
helping
us
with
reducing
the
cost
right.
So
now,
you're
not
running
all
those
Pro
talks.
Yet
we
were
running
there,
one,
you
know
one
proxy
per
host
and
you
have
the
Waypoint,
which
is
basically
a
acting
as
a
layer.
Seven,
we
basically
did
some.
You
know
in
Solo
some
of
the
of
of
of
measurement,
of
how
much
it
will.
It
will
help
you
and
there's
quite
a
lot
of
reduction
and
that's
without
us,
actually
improving
it
and
fine-tuning
it.
I
think
we
can
get
even
better.
A
The
second
thing
is
Simplicity
of
operation
and
I
think
this
is
the
biggest
one.
If
you
ask
me
what's
big
about
anyone
is
this
is
the
fact
that
it's
easier
to
consume
no
more
in
it
container,
no
more
following
upgrade
no
more
CVS
upgrade
and
so
on.
Now
you
don't
have
proxy
everywhere
and
again,
you
have
a
lot
of
talks
here
about
that.
A
So
please
go
and
and
listen
to
them,
because
I
think
it's
going
way
more
detail
than
I'm
going
so
then,
actually
with
MB
on
what
we
try
to
do
and
wow
I
think
we
did
very
well
is
look.
We
can
change
the
the
security
security
have
to
be
at
least
as
equivalent
to
the
current
project,
but
I
think
that
what
we
achieve
on
top
of
it
is
reduced,
cost
is
Simplicity
of
operation
and
the
Improvement
of
performance.
A
So
I'm
basically
really
really
believe
that
this
is
the
future
of
STL,
though
today
sdo
is
by
default,
starting
with
a
sidecar.
I
really
really
believe
that
in
the
future
we
will
see
that
the
default
will
be
in
here
Now.
By
working
this,
we
actually
also
already
put
it
in
our
product.
So
we
are
working
right
now
with
our
customer
to
make
MBL
production
ready.
I
believe
that
in
three
months
we
will
have
customer
running
it
in
production
already
and
I
think
that
in
six
months
everybody
will
be
able
to
run
it
in
production.
A
So
I'm
personally
really
really
excited
about
that.
So
this
is
what
we
are
doing
right.
Basically,
we
have
three
building
block
the
Google
Network,
the
blue
mesh
and
the
blue
Gateway.
We
have
an
edge
on
like
Blue,
Portal
and
Group
Rock
URL,
but
that's
basically
what
we
have,
but
there
is
a
few
things
that
is
very,
very
different
in
Solo
than
other
other
other.
That
I
think
that
it's
very
important
to
highlight.
So
we
are
here
in
the
conference
and,
as
you
know,
there
is
a
lot
of
projects.
Okay
right.
A
The
real
question
is
that
when
will
we
win-
and
we
don't
really
know
right-
I
mean
there
is
new
ones
the
journey
to
the
ecosystem
every
day.
I
think
what
solo
is
really
really
good
at
is
recognize
the
winner
more
so
we're
working
on.
You
know
we
from
the
get-go
started
with
with
kubernetes
we
adopted
NY.
Six
years
ago
we
started
with
SEO
there's
people
here
in
the
in
the
company
that
actually
worked
with
SEO
before
it
was
announced.
A
The
second
thing
is
that,
once
we
discover
those
projects
that
we
believe
will
be
the
winners,
we're
really
good
in
basically
leading
you.
So
you
know
we
have
a
lot
of
people
here,
that's
sitting
in
there
in
the
crowd.
That
is
basically
the
leaders
of
of
those
Community.
We
are
contributing
to
Android
to
SEO
and
honestly
anything
that
we
need
to.
A
We
have
two
of
the
TOC
member
of
the
SEO
Community,
there's
five
in
total
six
soon
and
yeah,
and
the
last
one
as
I
said,
is
it
we
continue.
You
know,
working
together
with
the
community,
to
basically
make
the
project
the
project
better.
Are
we
not
stopping
there
right,
so
we
discovered
them
becoming
a
leader
there
and
then
we're
pushing
the
boundary.
So
if
you
see
what's
special
about
solar
is
that
we
always
try
and
increase
the
boundary.
A
If
this
is
the
you
know
what
everybody
is
selling
today
right,
what
we
calling
blue
mesh
right,
it's
basically
something
that
I
was
talking
about
in
2018
before
people
knew
what
service
mesh
is.
So
that's
something
that
that
we're
working
on
and
honestly
got
wrong
for
a
long
time
until
we
fixed
so,
but
we
knew
that
this
is
the
problem
that
you
guys
will
have.
The
second
one
is
web
assembly
for
extending
of
the
envoy.
A
The
Dead
one
is
everything
that
related
to
ebpf
ambient
and
so
on,
so
really
really
important
to
us
to
to
push
the
boundary
now
I'm
almost
but
done.
But
one
thing
that,
as
I
said,
I
think
it's
extremely
unique
for
us
is
the
fact
that
we
already
done
it
We
Dine
with
hundreds
of
customers
and
that
I
think.
Where
is
our
strength?
We
we
successfully
have
a
lot
of
companies,
they're
running
right
now
in
production
and
enhancing
and
and
and
and
and
and
extending
their
environment
drastically.
A
But
as
I
said,
it's
not
important,
it's
not
enough
and
what
is
really
really
important
is
also
to
educate.
So
this
is
part
of
Education
right.
You
will
hear
today.
A
lot
of
people
from
the
community
is
talking
about
education
and
teach
you
about
all
those
projects.
And
besides
that,
and
everything
for
free,
you
can
come
and
basically
try
to
sell
a
workshop.
It's
all
for
free.
You
can
learn
a
lot
they
have
ends
on
and
then
also
can
certify
if
you
want
again
zero
money.
A
So
let
me
finish
with
kind
of
like
a
personal
note:
when
I
started
the
company,
there
aren't
a
lot
of
people
in
the
company.
It
was
me
and
a
few
engineers,
so
I
was
the
person
who's
supposed
to
enable,
and
basically
my
job
was
to
go
and
evangelist
and
I
think
that
the
smartest
thing
that
I
ever
did
in
seoula
is
understanding.
That
solo
is
not
me,
and
this
is
very,
very
important.
The
many
they
understood
that
solo
is
not
me
and
it's
not
Christian,
and
it's
not
late,
but
it's.
A
How
amazing
is
the
people
working
here,
I
mean,
and
you
will
see
them
all
here,
but
honestly
they
are
all
a
plus
plus
Technologies
and
I'm
so
proud
to
be
their
peer
and
working
with
them.
So
they
are
all
here
right
and
then
bring
them
here
as
usual
to
the
front.
Listen
to
them
talk
to
them,
learn
for
them
and
just
tell
us
how
we
can
help
So
yeah.
Thank
you.
So
much
I
really
hope
you
enjoyed
today
and
the
rest
of
the
comments.