►
From YouTube: Service Mesh for Developers - WebAssembly
Description
The service mesh is often presented as a solution for network engineering and system operability, increasing security, reliability, and observability. However, the service mesh is also an incredibly useful tool for developers, and understanding how to leverage this technology can dramatically simplify your day to day workflow.
In the first of 4 live streams you will learn how to use Gloo API Gateway with Consul Connect.
By the end of this series you will understand:
How to use metrics and distributed tracing effectively
Reliability patterns like retries, timeouts, and circuit breaking
How to leverage Canary deployments
How you can effectively debug distributed systems
The cloud-native, open-source technology used in this tutorial include: Envoy Proxy, Prometheus, Gloo shot, Consul Service Mesh, Loop, Squash, Open Census and more...
Schedule:
1. Gloo API Gateway and Consul Connect
2. Distributed Tracing
3. Canary Deployments
4. Microservice Debugging
5. Web Assembly
Questions or Comments -- https://slack.solo.io or @soloio_inc
Links to more resources:
Slides https://hubs.ly/H0mxdG70
HashiCorp Consul https://www.consul.io/
Solo.io Gloo https://www.solo.io/products/gloo/
Workshop Links
https://instruqt.com/hashicorp/tracks/service-mesh-for-developer-workflow
https://kubecon.demo.gs/01_index.html
https://github.com/hashicorp/consul-service-mesh-for-developers
B
B
B
B
A
Well,
I
just
want
to
welcome
everybody
who's,
just
tuning
in
and
and
welcome
back
if
you've
kind
of
watch
your
show
before.
But
what
we're
gonna
kind
of
concentrate
on
and
take
a
look
at
today
is
something
kind
of
I
would
say
bleeding
edge,
but
something
very,
very,
very,
very
cool.
So
there
is
a
new
feature
in
envoy
which
allows
webassembly
modules
to
be
loaded
dynamically
into
the
the
hot
path
of
the
filters
right
and-
and
why
does
it?
Why
is
this
exciting?
Well,
I?
A
Think
it's
really
exciting
because
being
able
to
extend
the
functionality
of
envoy
is
well
I
mean
it's
pretty
limitless.
You
can
do
simple
stuff
like
just
being
able
to
manipulate
requests,
adding
headers
or
you
can
do
really
really
cool
stuff.
You
could
do
request
validation.
You
could
do
checking
for
sequel
injection,
potentially
in
in
a
request.
A
B
I
was
just
gonna
add,
is
it's
really
important
and
is
becoming
even
more
important
because,
as
we
see
people
adopting
envoy
based
technologies
like
solo,
we
have
glue,
which
is
our
API
gateway
built
on
envoy,
actually
court?
We
all
have
consul
serviceman,
show
a
service,
my
open
source
service,
much
built
on
Boyd.
What
we
end
up
seeing
is
unbelie--
is
amazing.
Envoy
has
a
lot
of
features
out
of
the
box,
but
this
new
technology
when
would
bring
it
into
existing
enterprise
organizations.
B
There's
a
lot
of
assumptions
and
constraints
in
those
organizations.
Russia
might
not
be
present
in
the
open
source
at
wild
or
Google
or
lyft
or
worse.
Some
of
these
technologies
may
have
it
have
originated
and
what
they
need
is
a
way
to
customize
and
be
backward
compatible
with
some
of
the
things
that
they're
already
doing,
and
just
like
you
were
you're
saying,
like
the
ability
to
extend
envoy
now
to
change
its
behavior
is
in
a
in
a
little
bit.
More
approachable
way
is
very
powerful,
so.
A
Some
of
the
naysayers,
I
think,
may
look
at
this
and
go
oh.
Is
the
service
match
becoming
the
Enterprise
Service
bus
and
I'm
gonna
put
my
cards
on
the
table?
I,
don't
actually
think
an
Enterprise
Service
bus
was
a
terrible
thing.
I
used
one
and
quite
I
enjoyed
using
it,
but
I
guess
is
that
too
much
information
or
too
much
logic
going
into
the
service
mesh?
What's
your
opinion
on
that
so.
B
It
was
more
the
the
processes
that
built
up
around
it
and
the
ownership
of
the
actual
service
implementations
which
started
to
bleed
into
the
service
process,
but
got
things
into
into
trouble,
I
think
what
we're
and
sold
it
was
highly
centralized,
I
think
even
the
right,
the
creator
of
WebSphere
he
at
IBM.
He
said
he
created
that
Webster,
creating
it.
The
way
that
he
did
was
a
big
mistake,
because
it
just
forced
everything
through
this
it
was
like
building
a
castle
right
and-
and
you
had
to
go
through
that
castle,
to
get
what
you
wanted.
B
I
think
what
we're
doing
here
in
general
and
we're
service
mesh
solves
these
communication
challenges
in
a
highly
decentralized
way.
Yeah
and
we're
workflows.
Can
be
built
in
a
decentralized
way,
I
think
we
allow
teams
to
own
their
config
their.
You
know,
parts
of
the
service
that
they
need
without
having
to
try
to
centralize
all
of
us.
So.
B
A
You
know
that's
they're,
still
kind
of
deployed
globally,
and
it's
not
that
you
have
to
embed
this
sort
of
same
code
into
a
number
of
common
services.
You
can
deploy
it
into
the
service
mesh,
but
the
opportunity
is
there
for
the
individual
teams
to
own
those
filters
to
own
that
code
and
and
to
be
able
to
deal
with
that
so
I'm
pretty
excited
about
this
I
think
this
is
going
to
be
incredibly
incredibly
great
piece
of
technology
very
excited
to
see
this
coming
down
the
line.
I
love
the
fact
that
it's
also
very
portable.
B
B
A
It's
webassembly:
it's
anything
that
can
compile
webassembly
modules,
which
is
which
is
exciting
and,
and
today
we're
going
to
kind
of
take
a
quick
look
at
this,
but
but
I
think
we
if,
if
there's,
if
folks
want
to
and
if
folks
want
to
see
kind
of
some
more
stuff,
we
can
kind
of
continue
working
through
these
examples
and
and
kind
of
iterating
hitter
in
over
this.
So
give
me
give
me
the
101
Christian,
explain
to
me
what
you're
doing
with
webassembly,
hope
and
kind
of
the
this
process
flow.
B
Okay,
that's
a
great
that's
a
great
question:
let's,
let's
also
start
a
little
bit
context
right,
so
webassembly
was
built
originally
for
speeding
up
applications
in
the
web
browser
and
doing
that
in
a
way
that
extended
to
be
able
to
write
in
to
write
these.
These
applications
in
languages
other
than
let's
say,
JavaScript,
and
to
run
it
dynamically
in
the
browser
in
a
safe,
sandbox
way.
A
B
Right
and
to
be
able
to
dynamically
bring
that
down
to
the
browser,
just
like
a
web
page
and
run
it,
and
so
what
we're
doing
is
we're
taking
that
same
model
and
bringing
that
to
something
like
envoy
where,
in
the
browser,
you
would
typically
write
in
CSS
and
JavaScript
in
envoy.
If
you
wanted
to
extend
envoy,
you
would
have
to
write
in
C++
and
compile
it
and
statically
link
that
into
the
Envoy
binary.
Well,.
A
You've
you've
got
to
get
that
upstream
and
you've
either
got
a
four
convoy
or
you've
got
to
try
and
persuade
the
the
Envoy
Governance
Committee
that
your
change
is
beneficial
to
the
rest
of
the
world
and
actually
I.
Think
one
of
the
things
I
love
about
the
web
azzam
filters
is,
you
can
write
functionality
which
is
very
bespoke
to
your
organization
and
not
have
to
kind
of
deal
with
any
of
that
you
can.
You
can
create
that
very,
very
unique
functionality
that
that's
very
nice
yeah.
B
Exactly
and
and
even
if
you
did
fork
it
on
boy
and
run
your
own
custom
Envoy.
So
first
of
all,
now
you're
on
a
14
I
have
to
deal
with
that
right.
But
the
second
thing
is
for
all
of
your
extensions
that
you
create
they.
Whenever
you
deploy
envoy
all
of
those
extensions
that
you
create,
which
might
be
applicable
to
team
a
you
know,
some
of
them
for
team,
a
some
of
them
for
Team,
B
and
so
forth.
B
A
A
And
I
do
want
to
put
a
big
big
disclaimer
on
some
of
the
things
that
certainly
some
of
the
integrations
around
console.
It
is
very,
very
experimental,
we're
kind
of
just
looking
around
this,
but
but
I
want
to
get
people
to
get
a
feel
for
things.
Some
of
the
stuff
you
doing.
Obviously
the
web
assembly
hub
is
getting
quite
mature
and
things
like
that.
But
when,
when
are
we
gonna
see
this
an
envoy
are
we
are
we
gonna
see
1.4,
oh
sorry,
114
or
115,
maybe
or
maybe
later
on
than
that
yeah.
B
Probably
pretty
soon
I
would
say
that,
for
example,
so
glue
our
eight
that
gateway.
We,
we
ship
a
you,
can
install
glue
with
webassembly
enabled
right
sto
just
announced
1.5
their
service
mesh
have
assembly
enabled
so
as
as
you
know,
things
mature,
of
course,
it'll
be
an
upstream
ongoing
right
and-
and
I
don't
be
with
it'll-
be
pretty
soon
and
then
we'll
be
able
to
take
full
advantage
of
this
across
across
the
board
across
any
frameworks
using
using
online
and.
A
A
The
kind
of
the
the
envoy
was
and
stuff
which
is
going
into
glue
and
and
some
of
the
experimental
proxies
and
I've
been
playing
around
with
as
well
is
based
off
envoy
one
bought
one
dot
13.0.
So
it's
it's.
A
pretty
recent
release,
like
113
113
one
I
think
is,
is
that
the
latest
sort
of
patch
release
on
on
yeah,
so
yeah.
B
B
B
I
think
I
can,
let's,
let's
come
over
here
so
well,
we've
been
working
on
is
the
web
assembly
hub,
but
that's
kind
of
the
registry
for
discovering
webassembly
modules
and,
alongside
the
web
assembly
hub,
we
have
some
tooling
to
help
developers
get
started
with
with
with
building
Assembly
modules.
So
if
you
click
on
on
dots
here
and
webassembly
Hub,
we
can
will
we've
taken
to
the
the
docks
where
you
click
on
installation.
B
We
can
get
the
actual
CLI
to
to
begin.
Our
efforts
here
with
lettuce.
Only
one
thing
I
want
to
point
out
is
that
this
tool
is
intended
to
simplify
what
all
the
pieces
that
you
need
to
get
into
place
to
get
this
working
all
the
tool
chains,
setting
up
the
correct
versions
of
dependencies
and
and
so
forth.
So
let
me
actually
let
me
actually
show
you
that
right
now,
I've.
B
A
B
A
B
A
B
A
B
A
repo
for
some
of
the
existing
efforts
around
writing
the
SDKs
for
for
specifically
for
Envoy
for
these
different
languages
and
that's
gonna,
that's
gonna
continue
EXO!
So
right
now
we
see
we
have
a
couple
different
options:
we're
gonna
choose
assembly
script,
C++
I
mean
if
you
can
extend
envoy
today,
that's
how
you
would
do
it
with
C++
we're
gonna,
we're
gonna
think
it's
only
scope,
we're
gonna
target
a
particular
version
of
envoy
that
that
supports
web
assembly
modules
and
that's
what
we
would
find
in
these
platforms.
A
A
B
B
You
should
be
able
to
see
that
thank
you
for
the
welcoming
and
what
we've
created
is
the
the
package.json
which
brings
in
the
correct
runtimes
right
able
to
bind
to
the
Envoy,
api's
or
abs.
We
have
a
runtime
config,
which
also
binds
in
the
correct
versions
or
targets
to
correct
versions,
and
then
under
assembly.
We
see
the
source
code.
B
Okay,
so
we
come
in
here
we
can
see
the
source
code
for
the
for
the
extension
or
the
module
that
we
can
write
that
will
fit
in
nicely
with
with
on
the
way
on
void,
loads
and
expects
the
call
in
to
the
various
functions
here
or
methods
here,
and
we
can
see
all
this
one
is
fairly
simple.
What
we
can
do
with
some
of
the
out-of-the-box
example
here
is
adding,
for
example,
headers
to
the
requests
or
to
the
response
right
now,
there's
not
body
support,
but
that
that
should
be
coming,
and
you
know
this.
B
B
As
long
as
everything
that
gets
compiled
down
into
the
web
assembly,
you
have,
you
basically
have
all
the
implementation
or
all
the
functions
and
everything
that
you're
eight
that
you
need
to
do
within
the
web
assembly
module,
because,
when
you
deploy
into
web
assembly,
you
can't
reach
out
of
that
sandbox.
It.
B
So
everything
needs
to
be
compiled
into
that
into
that
level.
Assembly
module.
Okay.
So
in
this
case
we
just
see
we'll
just
go
with
the
out-of-the-box
sample
here,
we're
adding
branding
headers,
but
if
there's
a
configuration
of
all
add
a
header
with
a
dynamic
configuration.
So
let's,
let's
take
a
look
at
that.
B
First
thing
we
need
to
do
is
get
the
build
chain
in
order
and
build
our
web
assembly
module.
The
tooling
here
was
me:
build
allows
us
to
kind
of
have
the
same
experience
that
we
would
have
if
we're
dealing
with
docker
images
yeah
right,
so
these
are
just
these
was
just
my
basically
tar
balls
right
and,
and
here
we're
building
the
assembly
script.
We
give
it
a
name,
give
it
a
tag,
filter,
demo,
b01
and
should
take
a
few
moments
not
too
long.
B
A
B
This
is
where
you
guys
are
gonna
build
but
yeah,
but
you
can
see.
I
didn't
have
to
go,
pull
down,
libraries
and
mismatch
on
versions
and
so
I
just
buried
in
it
add
my
code,
all
the
stuff
that
I
need
to
worry
about
and
then
build-
and
we
can
see
here
now
that
it's
done.
It's
built
it
into
this.
This
image
and.
B
B
Underlies
me
store.
We
can
see
where
all
of
the
images
live.
If
we
want
to
look
at
a
particular
1:38,
let's
say
we
can
see
in
here.
We
have
the
hawazin
module
itself
yep.
We
have
a
descriptor
about
that
module
and
we
have
a
runtime
config,
which
is
part
of
a
web
assembly
for
envoy
spec,
that
we've
been
working
on
just
metadata.
That
describes
all
of
the
versions
and
all
of
the
the
things
that
are
necessary
to
be
able
to
run
this
particular
module.
B
B
B
B
B
A
B
A
The
way
the
way
eyes
look
at
this
as
well
is
that
what
I'm
doing
with
a
lot
of
my,
what
I
want
to
do
with
a
lot
of
my
webassembly
filters,
is
I
want
to
move
code
from
my
application
services
and
I
want
to
put
it
out
into
the
proxy.
So
I
want
to
kind
of
platform
wise
a
lot
of
my
code,
so
yeah
I
might
not
get
the
same
performance
as
native
a
C++,
but
I'm,
probably
gonna
get
comparable
performance
if
that
code
was
living
in
my
application,
absolutely.
B
B
A
I
think
the
nice
thing
to
mention
as
well
is
that
this
is
entirely
optional.
You
do
not
have
to
put
the
Web
Filter
in
every
single
sidecar
inside
of
your
estate.
You
can
pick
and
choose
if
you
have
very,
very
performant
performant
requirements
for
for
service
communication,
then
maybe
you
forego
web
assembly,
but
if
you
find
it
it's
okay.
B
B
B
A
A
A
B
B
A
B
A
Let
me
let
me
have
a
try.
I,
don't
know
is
that
there's
plenty
folks,
he's
gonna
be
going
like
Christian,
it's
all
about
smoke
and
mirrors.
There
is
no
smoke,
sometimes
yeah,
sometimes
but
I.
Think
one
of
the
things
that
I
really
like
this
work,
it's
really
kind
of
simple
straightforward.
It
gives
you
the
ability
to
kind
of
do
all
of
the
things
extend
extension.
You
may
not
even
know
about
one
of
the
things
that
I
want
to
show
folks
which
I've
just
got
on
my
screen.
A
Here
is
it
this
is
actually
the
envoy
configuration
or
or
something
very
similar
to
what
you
were
running.
So
if
you
look
at
it
in
terms
of
context
of
envoy,
if
you're
familiar
with
it,
you
can
see
that
envoy
has
a
concept
of
filter
chains,
various
different
filters,
such
as
TCP
proxy,
various
different
sort
of
database
protocols,
and
things
like
that.
A
Path,
which
I
think
is
is
pretty
pretty
cool,
so
I've
got
I've
got
a
very
similar
example
to
to
what
you
have
have
here,
and
let
me
just
let
me
just
get
this
up
and
running
in
in
console
service
master.
Again
massive
caveat
experimental
here
be
dragons,
but
it's
really
interesting
I
think
to
see
just
this
stuff
working.
A
So
let
me
just
run
my
environment
and
I
will
share
the
the
repo,
so
you
can
play
around
with
this
on
your
own,
but
what
I'm
doing
is
I'm
basically
just
running
console
Service
mash
inside
of
docker
and
I'm,
going
to
start
a
couple
of
application
services.
It'll.
Take
about
10
seconds
to
spin
up
there
we
go
and
we're
good.
So
of
course,
I'd
opened
up
in
the
wrong
browser
window.
A
B
A
A
The
sort
of
the
process
will
be,
as
you
would
expect
it.
It
will
be
easy
in
the
sense
that
you
can
set
it
in
terms
of
configuration
either
through
some
sort
of
centralized
configuration
or
probably
through
a
service
registration
again
experimental.
This
is
not
not
destined
for
production
just
yet,
but
I
have
I
have
that
web
service
and
the
web
service
has
an
upstream
upstream
on
API
API
is
going
to
use
wasm
filter
in
the
hot
path.
So,
in
the
request
path,
the
proxy
will
receive
a
quite
the
request
from
web.
A
A
Well,
it
seems
to
be
working,
so
what
I
can
see
here
is
there
we
go
so
I've
got
the
the
HTTP
headers
which
have
come
from
the
the
upstream
api
service
there
and
you
can
see
there
that
I've
added
a
header
called
hellos
and
it
has
a
value
of
world.
So
just
a
kind
of
just
to
show
you
a
little
bit
around
that
further
I'm,
just
gonna
log
into
the
the
API
directly,
and
we
can
just
take
a
look
that
if
I
manually
hit
that
that
end
point.
A
A
Oops
and
again
you
can
see
there
there
we
go
so
we've
got
that
that
hellos
and
that
world.
So
you
should
see
the
difference
there
there
it's
the
original
request,
which
hasn't
been.
It
hasn't
mutated.
It
has
passed
through
envoy,
so
the
there's
no
header
on
there
and
then
this
one,
which
does
does
have
that
header,
so
I'm,
all
about
kind
of
just
proving
that
this
is
deployable
and
how
this
works.
So,
let's
just
quickly
make
some
changes
and
just
kind
of
test
out
that
workflow.
A
Yeah,
you
can't
see
my
screen
yeah
I'm,
just
gonna,
say
hello,
YouTube,
so
I'm
gonna
save
that
file
and
I'm
gonna
bounce
over
and
what
I
have
here
is
just
a
terminal
into
the
the
build
folder.
So
I'm
kind
of
I'm
not
gonna,
use
the
Y's
me
docker
build
but
I'm
using
your
your
build
container
here
and
I'm
just
going
to
do
an
NPM
install.
B
B
A
Then
I
can
do
that
build
so
that
build
is
exactly
the
same
as
as
what
we
saw
earlier
on,
which
which
I
think
is
a
it's
incredibly
quick
in
it.
It's
sort
of
incredibly
sort
of
easy.
So
let
me
again
I'm
just
going
to
roll
that
API
container
and
start
a
new
one
so
that
it
it
will
pick
up
the
the
new
filter
and
it
will
get
loaded
into
Envoy.
We
can
see
that
that
has
come
up
and
then
I
can
curl
that
endpoint
there.
A
So
now
we
can
see
there
if
you
look
there,
hello,
YouTube,
so
that
change
I've,
just
literally
changed
that
wasn't
filter
I've
just
gone
through
a
simple
deployment
process.
I've
played
the
role
of
the
the
CI
system,
but
yeah
I'm
impressed
that
that
was
that's
relatively
easy.
I
mean
it.
Well,
it's
not
relatively
easy.
It's
very,
very
easy
to
yeah.
B
A
Work
and
so
what
a
folks
want
to
see
I
mean,
like
I,
think
this
is
interesting,
something
to
play
around
with
I'm
I'm
interested
to
look
at
the
types
of
things
that
we
can
do
with
this,
and
and
some
of
this
sort
of
the
more
sort
of
procedural
elements
like
building
this
out
into.
How
do
you
manage
this
in
production?
How
do
you
kind
of
what
is
the
workflow
for
for
developing
this
stuff?
B
B
A
A
A
Validation
and
I
can
share
this
in
webassembly
hub
somebody
can
use
it.
They
can
pull
my
my
code,
maybe
I
write
a
sequel
injection
sniffer
in
the
for
the
request.
Validation
really
release
four
things
there,
but
I
hope
this
has
been
interesting
and
I
hope.
I,
hope
you
like
this
stuff
I
will
share
all
of
the
links
in
the
video
description
below
just
as
soon
as
we
drop
off.
A
A
B
B
B
A
If
you
could
build
a
global
waz
me
filter,
Wiseman
filter
which
would
just
handle
that
JW
tease,
taking
all
that
stress
and
pressure
off
the
engineering
teams
and
just
kind
of
putting
that
as
a
centralized
platform
component
I'm
also
interested
in
seeing
is
whether
we
can
kind
of
chain
those
things
right.
The
way
through
from
may
be
looking
at
glue,
API
gateway
and
open
ID
connected
things.
It's
not
possible.