►
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
Intro: NATS - Waldemar Quevedo, Synadia Communications, Inc
NATS is a high performance publish/subscribe messaging system that has as one of its main goals connecting services in the simplest, most secure and reliable way possible. The project has a long history of being part of production deployments as the core component for the internal communication of a distributed system, and in its latest release NATS v2.0, its feature set was enhanced so that it can also be used to create a global and decentralized communication network. In this talk, you will learn about the multiple communication patterns that NATS offers and how it can help you simplify and decouple the architecture of your cloud native application.
https://sched.co/MPiJ
A
Okay,
well,
thank
you.
Everyone
for
attending.
This
is
an
introduction
about
Nats.
First
I
just
want
a
quick
ask.
How
many
of
you
have
already
heard
about
Nats
before
was
plenty
of
okay
yeah.
Do
we
join
last
year
with
the
CNC
F,
so
there
has
definitely
been
an
uptick
in
the
number
of
contributors
and
maintainer
to
the
project
and
how
many
of
you
are
running
in
production
or
okay,
at
least
ringing?
Okay?
Oh,
that's!
Right!
A
Okay!
So
a
little
bit
about
myself,
my
name
is
Valdemar
Keveza.
You
can
find
me
in
twitter,
Wallach
us
I'm,
a
software
engineer
at
Sian
Adia
communications,
where
most
of
the
maintainer
for
Nats
are
currently
working
and
we
I'm
a
core
maintainer
for
Nats,
four
and
four
years
already,
but
I
have
been
using
Nats
based
systems
for
close
to
seven
years
already,
and
last
year,
I
managed
to
publish
practical
Nats,
which
is
one
of
the
first
books
about
Nats.
A
Anyone
has
a
book
we
gave
many
copies
at
the
cube
Conan
in
Seattle,
but
I
didn't
bring
many
this
time
so
agenda.
For
this
talk,
I'll
just
give
you
a
quick
overview
about
the
next
project.
So
the
basic
and
talk
about
this
messaging
paradigm
and
also
talk
about
whether
it's
coming
in
the
next
feature
from
in
the
next
release
from
Nats
Nats
version
2,
which
the
team
has
is
very
busy
at
in
shipping
right
now,
so
I'm.
A
Actually,
this
time,
I'm
the
only
one
from
the
nuts
team
that
is
attending
cube
con
and
I'll,
give
you
a
quick
demo
of
assembling
an
ad
super
cluster
on
kubernetes
using
the
Nats
operator,
also
not
operator
users,
okay,
one!
Okay!
Thank
you!
So
about
Nats
Nats,
it's
an
eight-year-old
battle-tested
production
prevent
cloud
native
messaging
system
that
is
made
for
developers
and
operators
that
want
to
spend
more
time
doing
just
communicating
within
their
microservices
that,
rather
than
architecting,
messing
solutions.
So
what
distinguishes
nets
as
a
project?
A
It's
it's
focused
on
simplicity
and
performance
and
with
the
v2
release,
the
team
has
really
doubled
down
on
making
it
a
very
secure
system.
It
has
built
from
the
ground
up
to
be
a
cloud
native
system.
Actually,
the
first
problem
then
add
salt
was
for
the
communication
within
cloud
foundry,
which
is
a
platform
as
a
service
project
that
was
based
in
Ruby.
Original
version
from
Nats
was
based
in
Ruby,
as
well,
even
for
Ruby
standards
at
those
times,
2012
11.
A
It
could
also
push
around
to
150,000
messages
per
second,
and
now
it's
written
in
gold
and
the
performance
is
way
way
higher.
It's
thanks
to
its
simplicity.
There
is
an
many
implementations
for
the
of
many
clients.
The
team
supports
around
10
official
clients
and
with
Nats.
Not
not
only
you
can
do
request
response,
but
also
you
can
do
multiple
communication
patterns
that
I
will
talk
about
in
a
bit.
A
So
we
joined
last
year
as
part
of
the
club
as
an
incubation
project
for
the
cloud
native
computing
foundation,
we're
under
the
streaming
and
messaging
section
from
the
CNCs
landscape.
You
can
see
it
Nats
as
being
close
in
terms
of
functionality,
2g
RPC,
on
like
G
RPC
nuts.
It
can
also
do
request
response
and,
unlike
gr,
PC
Nats
doesn't
treat
its
payloads
are
opaque,
so
it
doesn't
couple
with
any
type
of
encoding.
A
So
you
can
use
message
back
protocol
buffers
or
just
plain
JSON,
and/or,
bytes
and
binary
data,
and
you
also
have
more
more
communication
patterns
and
load
balancing
all
in
this
part
part
as
part
of
the
same
binary,
which
is
really
helpful
since
last
year.
I
think
number
of
contributors
have
a
has
raised
by
one
third
model
and
third,
this
lack
they
have
a
various
active.
Its
lack
channel
definitely
join.
You
want
to
hear
more.
A
The
team
is
all
the
time,
ramping
up
users
to
the
project
and
we're
very
happy
to
see
new
members
are
picking
our
brain
and
devising
new
solutions
to
which
well,
it
is
very
exciting
time
to
form
part
of
the
nuts
community,
because
released
this
year
is
not
to
be
to
which
it's
about
to
ship
less
than
a
couple
of
weeks.
But
since
2014
the
number
of
releases
has
increased.
A
Nats
was
created
by
direct
coalition
and
their
cousin.
He
has
been
building
messaging
systems
for
for
a
long
time
for
over
25
years
and
many
of
the
commentators
from
come
from
comments
like
TIBCO
or
or
Google.
So
there
is
a
lot
of
accumulated
experience.
Messaging
experience
in
this
team,
so
I'm
very
happy
to
be
part
of
this
team
and
also
the
communities
it
it's
also.
Well,
you
can
see
from
the
slack
channel.
Just
reacted.
A
The
number
of
nuts
and
users
are
is
increasing
more
and
more.
We
are
recently
heard
about
tinder.
They
made
a
great
good
blog
post
of
how
reducing
that's
part
of
the
architecture
and
using
WebSockets
and
for
the
matching
logic.
That's
for
discovery.
Suddenly,
there's
no
end
user
talk.
This
conference
I
heard
the
worst
one
a
cloud
native
rejects,
so
we
looking
forward
to
hearing
more
about
user
voice
and
presenting
about
nets
and
how
are
you
adopting
it?
A
So
the
basic
use
cases
for
Nancy's
cloud
native
cognitive
messaging,
it's
very
reliable
optimized
for
doing
the
micro
service
communication,
but
you
can
also
use
it
for
eventing
publishing
events
or
data
streaming
for
observability
analytics
command-and-control.
Thanks
to
the
multiple
communication
patterns
that
you
can
have
with
Nats,
it's
it's
very
useful
to
be
able
to
orchestrate
multiple
nodes
on
their
single
request.
There
is
an
IOT
and
edge
use
cases
as
well.
A
Thanks
to
the
how
lightweight
the
protocol
is
so
telemetry
sensor
data
feel
really
well
and
also
we
haven't
seen
it
usage
where,
basically,
you
start
looking
for
something
that
is
a
little
more
simple
to
what
they
already
have.
This
is
like
complimenting
like
a
kafka
deployment,
sometimes
ok,
so
one
of
the
goals,
what
I
think
from
Nancy's
to
provide
and
always
available
dial
tone
you
want
to
connect
everything
and,
as
per
part
of
nuts
or
the
color
of
nets,
is
traceable
patterns.
Yes,
you
can
polish
subscribe,
especially
want
to
end
communication.
A
You
have
built-in
support
for
load
balancing,
so
you
can
enhance.
You
can
create
subscriptions
that
are
not
only
to
receive
for
one
to
n,
but
one
to
one
randomly
load
balancing
the
request
and,
of
course
you
have
a
request.
Reply,
recurse
reply
as
well,
which
is
itself
built
on
top
of
pure
poly
subscribed.
So
it's
using
a
female
subscriptions
with
limited
interest
that
it
announced
the
number
of
subscribers,
and
then
you
expect
the
response
with
the
lowest
latency.
For
example,
you
publish
and
subscribe
to
subjects
simple
subjects
for
sampler
foo.
A
You
can
use
the
dot
character
to
create
hierarchies
of
subjects,
for
example
food,
dot
bar
and
based
on
those
delimited
subjects
by
the
dots
you
can
create
wildcards.
So,
for
example,
food
dot
asterisk
would
match
anything
that
is
published
to
food
bar
food,
odd
bass,
food
dot
asterisk.
That
bar
will
match
anything
that
is
gets
published
with
food
odd,
a
bar
or
food
OB,
the
bar
and
also
there's
a
stronger
variation
from
the
wild
card,
which
is
the
fool
wild
card
that
can
match
anything
from
a
certain
point
from
the
subject
after
a
dot.
A
A
So
you
can
do
request
response,
which
is
end
to
end
going
to
be
2,
n,
+,
2,
+
1
to
n
communication,
or
you
can
also
do
1
to
n.
So
you
can
polish
on
through,
and
this
gets
announced
to
any
number
of
clients
that
were
announced
interest
on
the
food
subject,
and
you
can
enhance
their
subscriptions
with
by
appending
the
Q
group
name,
for
example,
soup
food
workers.
In
this
case,
only
one
of
them
will
be
receiving
the
request
randomly
there's
only
one
random
load-balancing.
A
A
So
we
summarized
this
as
a
tip
to
users
that
never
assume
the
audience
of
a
message
so
never
assume
that,
for
example,
this
message
is
going
to
only
be
consumed
by,
let's
say,
like
service
a
the
day
will
come
when
you
actually
want
to
find
out
these
messages,
for
let's
say
for
auditing
purposes
or
many
other
use
case
and
requirements
come
up.
So
it
is
really
helpful
to
be
able
to
take
advantage
of
the
publish/subscribe
paradigm
to
accommodate
those
and
new
requirements.
A
So
this
time
they
saw
the
main
trains
for
Nats
performance
scalability
and
its
resilience,
the
single
binary,
the
small
binary
form
the
nut
server
with
lingo
can
push
around
it.
18
million
messages
per
second,
and
if
you
create
multiple
connections,
you
can
even
in
a
very
beefy
machine,
you
can
get
up
to
80
million
messages
per
second
and
the
team
is
very,
very
careful
with
the
performance.
A
Availability
is
another
thing
that
great
thing
about
the
design
from
Nats.
The
nuts
servers
are
very
famous
for
having
very
high
up
times,
and
this
is
because
the
nut
server.
We
say
that
it's
a
has
a
selfish,
optimization.
Let's
say
if
client
is
not
being
able
to
process
messages
fast
enough
by
default
kind
of
read
data
from
two
seconds
from
a
socket,
then
the
stat
server
considers
that
client
to
be
unhealthy
and
it
will
disconnect
that
client.
A
This
is
in
order
to
protect
the
service
for
and
all
of
the
other
clients
that
want
to
keep
on
communicating.
So
this
is
very
different
from
other
types
of
systems.
Where
really
focuses
on
trying
to
ensure
that
a
certain
message
is
processed
by
a
certain
client,
but
then
it
affects
the
service
from
from
others,
and
this
becomes
like
the
in
terms
of
errors.
This
becomes
like
a
slow
consumer
error
and
they
we
have
it
as
part
of
the
client's
logic.
You
can
eagerly
detect
whether
your
client
is
becoming
a
slow
consumer.
A
For
example,
there
is
full
mesh
clustering
as
well
full
mesh
one-hop
for
high
availability,
and
so
any
client
as
part
of
your
cluster
only
needs
to
be
connected
to
any
one
of
the
nodes
from
part
of
the
cloud
in
the
nuts
cluster,
and
it
will.
The
message
will
be
properly
routed
to
the
to
the
endpoint.
A
A
The
it
is
a
plain
text
based
protocol
has
a
very
few
number
of
burbs.
This
is
essentially
the
same
protocol
as
when
Nats
was
created,
so
it
has
not
changed
for
since
it
for
more
than
seven
years,
and
also
for
that
version
to
the
next
client
protocol
has
not
changed.
On
the
other
hand,
the
clustering
protocol
has
changed
and
I'll
take
a
talk
about
it
later
or
maybe
tomorrow,
tomorrow
also
have
a
deep
dive.
A
There
is
built-in
gossiping
of
the
network
topology,
so
your
client
can
just
connect
to
any
node.
That
is
part
of
the
cluster,
and
it
will
be
aware
of.
Where
are
the
other
end
points
that
it
kind
of
fell
over
to,
for
example,
and
if
I
knew
no
joints,
then
it
will
be
gossiping
that
new
network
end
points
that
you
can
only
as
long
as
you
can
connect
to
at
least
one
of
them,
then
the
failover.
A
Will
you
can't
rely
on
the
failover
to
eventually
be
available,
as
well
as
a
client,
so
there
are
two
main
that
they
will
delivery
modes
in.
We
have
separated
them
into
different
projects.
I've
been
talking
about
and
Nats,
which
we
call
cornets
is
again
almost
one's
delivery,
meaning
that
you
have
to
be
connected
in
order
to
receive
the
message
we
have
another
project
named
NAT
streaming,
which
gives
you
at
least
once
delivery
guarantees.
This
is
was
more
in
line
to
the
classic
notion
of
what
we
think
a
queue
is.
A
Where
do
you
publish
something
and
expect
someone
to
pick
something
from
that
queue,
and
so
there's
at
least
one
delivery
with
message
we
playing
and
all
those
kind
of
features
and
the
team
is
very
open
in
the
area
in
the
exactly
ones
requirements
and
so
they're
constantly
implemented
exactly
once
in
previous
messaging
systems
that
he
architected
and
the
lesson
is
that
there
is
not
worth
the
effort
so
Nats
streaming.
It's
a
layer
on
top
of
Nats
is
using
protocol
buffers
on
using
the
request
response
protocol
from
that
you
can
replay
from
a
certain
sequence
sequence.
A
The
number
of
messages
has
durable
subscribers,
so
you
can
connect
and
resume
consuming
the
messages
that
have
been
published.
Then
there
has
a
notion
of
an
ACK,
unlike
unlike
Nats,
so
you
can
have
a
number
of
set.
What
is
the
maximum
number
of
in-flight
messages
that
can
connect
this
for
a
certain
client
and
for
the
high
availability?
We
have
a
raft
clustering
for
the
replication
of
messages
so
back
to
Nats
Nats.
This
is
written
in
go
this
streaming
and
high
performance
has
TLS
support.
A
Authorization
authentication
are
done
in
the
Nats
layer
and
when
announcement
is
that
we
have
changed
all
of
the
repos
from
the
Nats,
how
your
organization,
so
for
a
long
time,
that's
the
NAT
server,
what's
called
a
Jeanette's
D
so
because,
because
of
the
rig
right
from
from
go,
and
originally
the
name
from
net
was
not
server
and
we
went
back
to
the
original
name
from
that.
So
it's
now
not
servers
and
we
have
redirected
from
Jeanette's
d2
to
this
one
and
a
binary
name
as
well
will
change
to
NAT
server
for
the
video
release.
A
We
have
also
renamed
all
of
the
Nats
clients,
libraries,
the
official
ones.
So
what
used
to
be
Nats
on
their
Nats
on
your
gonads?
Now
it's
a
Nats
go.
So
that
way
you
can.
If
you're
looking
for
the
Python
client,
for
example,
it
would
be
Nats
dot
pipe,
and
these
are
all
day
so
the
official
clients
that
the
team
is
supporting
so
next
dot
seed,
Nats
net
nos
da
GS
and
I've
tried
to
follow
that
convention.
This
is
a
sample
API
from
the
Nats
go
client
basically
subscribing
to
the
greeting
subject.
A
This
is
an
asynchronous
subscript
subscription,
so
also
I
had
so.
If
you're
adopting
some
it's
sometimes
new
users
expect
that
the
synchron
subscriptions
are
going
to
be
processing
the
messages
in
parallel,
but
actually
the
clients
that
is
up
to
the
end
user
of
the
determining
what
is
the
level
of
concurrency
that
they
want
for
that
for
our
clients.
So,
for
example,
if
there
is
an
asleep
inside
of
this
greetings
callback,
then
the
next
message
won't
be
processed
until
this,
the
processing
from
this
message
is
done.
A
So
you
need
to
train
like
an
employee
group
of
guru,
things
for
example,
or
or
threads
in
case.
You
want
to
increase
the
level
of
concurrency
and
since
Nats
and
most
wants
system,
then
you
have
to.
You
will
only
be
receiving
a
message,
a
client
from
the
moment
that
you
expressed
interest
into
the
subject.
This
is
an
example.
Next
cluster
topology,
where
you
can
publish
a
message
and
then
it
will
be
routed
properly
to
the
other
client
that
rest
announced
interesting
to
a
certain
subject,
and
there
is
in
case
one
of
these
nodes
fails.
A
A
We
also
have
renamed
the
clients
from
what
used
to
be,
for
example,
go
Nats
streaming
into
standard,
go
story
going
back
a
little
bit
to
the
original
name
to
Stan
and
then
similar
to
nuts,
the
nuts
clients.
You
can
find
a
standard,
go
standard,
PI
standard
net
and
stab
dot
RB,
and
these
are
all
the
implementations
for
the
requirements
on
protocol
form
the
knot
streaming
or
stamp.
A
The
API
is
very
similar
to
the
Nats
client
API.
You
in
this
case
I
flip
the
order
so
just
for
purpose
text,
for
example,
purposes
by
default.
It
will
connect
to
the
locally
available
NAT
server.
So
it
has
a
embedded
Nats
connection,
but
you
can
also
pass
your
own
customized
and
nuts
connection,
and
in
this
case
this
client
will
receive
the
messages
that
have
ever
been
published
from
the
time
well
to
add
to
a
channel
from
from
the
beginning.
So
you
can
pass
a
function.
A
Stand
deliver
all
available
and
it
replay
all
the
messages
that
have
been
published
so
far
and
again,
that's
streaming
is
a
layer
on
top
of
naps,
so
you
have
to
prepare.
There
is
an
embedded
mode
for
Nats
to
run
instead
of
the
net
streaming
server
as
well,
and
but
you
can
also
it
is
flexible
to
that.
A
This
is
the
biggest
release
from
the
project
since
its
beginnings,
as
you
can
see
from
the
spice
from
commit
the
goal
from
that
is
this
release
in
particular.
Is
that
and
Nats
for
a
long
time
was
a
technology
that
would
be,
for
example,
instead
of
a
data
center
optimized
for
the
service
discovery
and
communication
within
the
same
data
center
and
with
Nats
version
2,
we
are
enhancing
Nats
Nats
becoming
hands
so
that
you
can
create
clusters
of
clusters
and
also
communicate
a
very
secure
way
using
all
of
these
well
new
technologies.
A
That
I
will
explain
and
we
have
built
one
example
solution
that
is
using
all
of
these
pieces
named
NGS
and
here's
the
linkage
you
want
to
get
more
info,
but
it's
essentially
a
global
communication
network
single
URL
that
you
can
connect
to
and
in
half
that's
like
seven
regions
across
the
world,
and
it
is
G
aware
gr.
Where
from
where
do
you
want
to
cut
for
anyone?
You
are
connecting
and
will
be
routing
the
messages
properly.
A
So
just
you
only
need
to
know
one
single
URL
and
then
you
can
communicate
over
the
world
with
all
the
clients
so
list.
So
the
features
that
are
coming
with
that's
version,
two
and
I'll
explain
more.
If
you
want
to
drop
by
a
deep
dive
tomorrow,
have
gateway,
superclusters
and
leaf
nodes.
Supercluster
are
the
cluster
of
clusters,
so
that
span
more
than
one
data
center
gateways
are
the
what
connects
those
clusters
and
we
also
have
leaf
nodes.
A
What
are
called
leaf
nodes,
which
is
essentially
a
new
type
of
nut
server
that
connects
to
a
super
cluster
so
that
you
can
have
a
number
clients
that
do
not
implement
all
of
the
features
from
my
nuts
client
and,
for
example,
like
devices
that
require
a
lesson
less
with
more
constrained
environments,
the
leaf
nodes
can
be
helpful
and
the
clustering
protocol
was
completely
rewritten
so
that
we
can
have
a
very
cool
feature
named
accounts
accounts.
It's
say
we
we
say
that
is,
they
are
like
containers,
but
for
messaging.
So
that's
that's
version.
One.
A
Everything
is
done
over
a
single
global
account
and
you
can
also
use
a
ninth
person
too
in
that
way,
but
that
is
brave
fight.
Well,
it
is
a
new
dollar
G
global
account
over
all
of
that
is
happening,
but
well
that
is
well
that
communication
is
happening
and
you
can
create
more
accounts
to
be
able
to
separate
the
subject
namespaces
for
different
purposes
under
the
same
server.
So
it's
well
to
keep
multi-tenant.
A
There's
a
new
technology
named
end
keys,
which
is
which
are
8255
1-9,
based
keys
that
you
can
basically
a
PKI
kind
of
system
where
you
can
create
a
private
key
and
then
use
used,
your
own
private
key
stable
to
sign
the
challenges
that
the
server
send
you.
So
that
way,
the
nut
servers
do
not
have
any
any
private
keys
or
any
account
information
whatsoever.
A
Now
we
say
that
anything
that
is
any
type
of
a
subscription
that
where
a
requester
is
expecting
a
response,
we
call
that
a
service
and
we
are
when,
whenever
it
is
just
a
data
flow,
just
publishing
data
and
sending
events,
we
call
those
strings
and
using
the
accounts
and
export
and
import
they
can
mechanisms.
You
can't
mount,
for
example,
a
namespace
from
another
account
so
that
you
can
share
data
between
multiple
accounts
and
in
case
of
the
NGS,
you
can
do
that
in
a
global
scale,
using
the
permissions
dictated
by
the
JWT.
A
This
is
an
an
example
of
an
export,
for
example,
on
the
assign
a
dia
account.
It
is
exporting
for
anyone
that
wants
to
import
this
data.
From
this
account,
you
can
import
the
cloud
medical
status
or
private
dot,
Deb
stats.
In
these
cases,
the
account
the
CN
CF
account
is
important.
That
data
know
n
Keys,
Angelo
duties.
This
was
designed
so
that
the
nut
server
does
not
hold
in
any
private
keys,
or
you
this.
A
Essentially,
this
usage
essentially
deprecated
the
basic
authentication
that
we
used
to
have
so
or
bcrypt
based
credentials
and
there's
a
number
of
a
metadata.
The
digit
abilities,
half
I,
just
go
through
and
covered
deep
dive
tomorrow,
so
example,
NAT
supercluster
important
to
say
about
the
Nats
clusters
are
dead.
A
A
It's
an
example
of
super
clusters:
I
can
give
you
leaf.
Nodes
are
essentially
an
extra
node
that
connects
to
a
super
cluster,
and
then
you
can
communicate
with
the
leaf
node
instead
and
the
same
thing
would
happen
so
I
would
Nats
on
kubernetes
just
to
briefly
touch
on
this.
Have
a
visitor
who
released
the
new
burst
of
the
Nats
operator,
be
0.5
that
is
Nats
b2
ready
has
all
day,
basically
Nats
operator
composing
many
of
the
what
we
have
learned
out:
good
best
practices
of
deploying
Nats.
A
When
deploying
Nats
u4
created
clusters,
it
is
kind
of
a
kind
of
tricky,
sometimes
like
setting
the
air
records
and
assembling
the
full
mesh.
So
the
nuts
operator
can
do
this.
For
you,
it
works
best
on
kubernetes,
be
1.12
or
I.
Think
the
users
are
getting
ahead
of
us
and
I
think
digital
ocean,
which
is
what
I,
based
in
this
demo,
already
supports.
Abi
114,
gke
I
think
still
be
1.12,
but
basically
all
of
the
cloud
providers
already
support
B
1.12,
at
least
if
NASA
preto
is
very
helpful
for
crane.
A
The
clusters,
which
was
very
simple
as
a
small
ya,
know
you
have
a
cluster,
is
code
mode
so
that
he
can
govern
and
create
clusters
with
throughout
all
of
the
namespaces.
There
is
a
very
this
advanced
feature.
You
can
use
bound
tokens
from
kubernetes
and
if
you
have
kind
of
niche
feature
where
you
can
not
only
use
the
service
accounts,
but
have
the
service
accounts
signed
targets
with
a
different
audience
only
for
the
purpose
of
Nats,
and
we
also
have
a
recent
contribution
using
the
cert
manager.
A
You
can
define
a
TLS
support
all
done
by
the
inside
of
kubernetes.
You
don't
have
to
use
an
out-of-band
certificates,
there's
a
primitive
side
card
support
and
again,
as
I
mentioned,
that's
b2
ready,
so
yeah
you
can
create
gateways,
leaf,
nodes
and
trust
operator,
not
super
Matthias
export.
There
are
two
supports
and
that's
about
net
net
streaming
and
they'll.
Just
give
you
a
quick
demo
of
deploying
that's
B
to
cluster
on
kubernetes
and
see
them
running
short
of
time.
Just
move
to
questions
as
well.
Any
questions.
B
A
B
B
B
A
A
And
just
a
key
quick
example
of
how
to
create,
and
that's
with
to
cluster
with
gateways.
This
is
just
in
the
digital
ocean
kubernetes,
and
that
I
just
got
help
with
that.
I
was
blocked
on
an
issue
with
the
host
port,
so
just
well
become
more
familiar
with
psyllium
psyllium
is
pretty
cool.
This
is
an
example
manifest
where
we
are
stitching
together.
Two
different
data
centers.
A
A
Aware
of
whether
they're
already
connected
or
not
to
the
other
data
center,
so
so
that
they
have
a
single
outbound
connection,
sending
requests
have
a
couple
of
cue
subscribers
connected
to
you
and
SF,
so
the
queue
subscriptions
when
you
are
in
gateway
mode.
They
are
GOI
where
so
in
this
case,
the
it
is
sending
requests
from
Europe
to
another
queue.
A
A
couple
of
cue
subscribers
are
also
in
Europe,
so
all
the
requests
are
going
to
the
Europe
queue
subscribers
and
it
is
only
when
those
queue
subscribers
in
Europe
fail
when
the
requests
are
going
to
start
being
received
by
the
consumers.
There
are
in
san
francisco
digital
ocean
data
center,
so
it
can
help
with
disaster
recovery
and
use
cases.