►
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
Keynote: Envoy Update – Matt Klein, Senior Software Engineer, Constance Caramanolis, Software Engineer, & Jose Nino, Software Engineer, Lyft
To Learn More: https://sched.co/JLgh
A
Good
morning,
so,
I'm
not
sure
if
everyone's
heard
of
envoy
but
well
rectify
that
soon
enough.
Now
this
is
what
lifts
topology
looks
like
three
and
a
half
years
ago.
The
reason
why
I
may
seem
familiar
not
only
because
one
of
us
or
anyone
who's
familiar
which
has
presented
lifts
topology
in
the
past,
but
this
network
topology
is
actually
pretty
familiar
and
is
actually
there's
variations
of
this
apologies
seen
across
the
industry
and
some
of
the
issues
that
were
seen
with
this
topology
and
that
we
experienced
personally
at
lift
was
an
example
of
the
elbe.
A
This
is
a
lack
of
observability.
I
know
their
issue
too.
Is
that
there's
numerous
technologies
right
like
if
I
were
to
pull
the
entire
room
right
now
and
ask
for
everyone's
favorite
language?
Obviously
it's
C++
right.
We
would
get
different.
Just
like
ever
ask
for
different
of
what
everyone's
favorite
proxy
back
in
the
day
was.
We
would
get
different
answers
so
a
lot
of
some
of
those
issues
that
I
mentioned
and
the
other
issue.
Other
issues
that
we'd
experienced
were
actually
the
forcing
function
in
defining
envoys
mission.
A
B
Taking
that
mission
statement
as
our
North
Star,
we
shifted
the
paradigm
into
the
service
mesh,
more
specifically
a
service
mesh
where,
alongside
every
single
micro
service,
an
envoy
sidecar
was
running.
Alongside
with
the
service
mesh,
we
were
able
to
flow
all
the
network
traffic
through
an
envoy
instance,
but
this
allows
us
to
do
is
to
see
a
lot
of
improvement
in
our
dimensions,
but
specifically
three
dimensions
that
are
very
important
one.
We
have
the
same
configuration
language
in
all
of
our
topology
number.
Two.
B
We
have
the
same
network
guarantees
at
every
hop
of
the
network
to
defend
our
micro
services
from
the
common
pitfalls
of
distributed
systems
and
three
when
problems
do
occur
because
they
do
occur.
We
have
the
same
of
severability
metrics
at
every
hop
of
the
network,
allowing
an
urgent
our
engineers
to
effectively
and
efficiently
pinpoint
the
problems.
So
in
all
of
these
dimensions,
what
we're
doing
is
reducing
the
cognitive
load
of
engineers
as
they
move
from
piece
to
piece
of
this
ecosystem.
B
C
Jose
we
open-source
envoy
in
September
of
2016,
and
it
has
been
an
absolutely
incredible
ride.
It
blows
my
mind
to
see
what
has
happened
in
the
past
two
and
almost
two
and
a
half
years
from
all
three
major
cloud.
Vendors
now
offering
envoy
managed
products
to
many
many
end-users
to
tens
of
startups
that
are
building
products
now
based
on
envoy.
It's
just
absolutely
incredible
and
we're
very
thankful
to
all
of
you
for
what
has
happened
in
such
a
short
period
of
time
and,
as
was
previously
announced,
we've
just
been
announced
a
CNC
F
graduated
project.
C
It's
a
great
milestone
for
the
project.
It's
an
indication
that
the
larger
community
realizes
that
envoy
is
not
going
to
go
anywhere.
It's
ready
for
wider
enterprise
adoption
and
that's
really
fantastic
I'm,
often
asked
why
has
Envoy
become
so
popular
in
such
a
short
period
of
time,
because
if
you
look
at
it,
the
growth
and
again
just
two
years
has
been
truly
incredible.
I
think
it's
for
a
couple
of
different
reasons.
The
first
is
performance.
C
When
we
originally
built
envoy,
we
recognized
that
if
it
was
going
to
be
deployed
in
different
edge
scenarios
and
services,
service
scenarios
and
other
other
types
of
proxy
deployments,
it
had
to
be
fast,
particularly
at
the
tail.
So
we
focused
on
performance
reliability
even
as
we've
scaled
the
project
too
many
developers,
many
maintenance
organizations
we've
been
able
to
keep
reliability
very
high,
focusing
on
integration
tasks,
fuzzing
tight
code
reviews
all
of
those
types
of
things,
modern
codebase.
Many
of
our
proxy
competitors
are
not
in
github.
C
They
don't
use
modern
best
practices
around
code
development
on
voice
written
in
modern
C++,
which
has
been
very
important
to
many
people
taking
it
up.
Extensibility
envoy
was
built
from
the
ground
up
to
have
numerous
extension
points.
We
allow
people
to
bring
in
the
code
and
build
all
types
of
things
on
top
and
that's
led
to
an
absolutely
incredible
number
of
use
cases
that
people
have
built
truly
amazing.
C
Observability,
you've
heard
everyone
talk
about
it
on
boy
spits
out
copious
stats,
logging
tracing,
allowing
people
to
build
higher
layer
systems,
which
is
amazing,
our
configuration
API,
which
we
call
our
XDS
api.
That's
really
the
cornerstone
of
Envoy
that
has
allowed
us
to
scale
to
you
know,
decouple
the
data
plane
from
the
control
plane.
We've
seen
many
control
planes
built
on
top
of
Envoy
and
that's
been
really
fantastic
and
finally,
last
it's
all
of
you.
It's
the
fact
that
envoy
does
not
have
any
paid
product
behind
it.
It's
a
community
driven
product.