►
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: CNCF Project Update - Bryan Liles, Senior Staff Engineer, VMware Bryan Liles
https://sched.co/MQVe
A
Before
I
get
started,
thank
you
dan
and
Cheryl
for
ticking
off
the
key
notes
today.
So
what
we
wanted
to
do
is
have
themes
for
the
keynotes
and,
as
we
organized
the
keynotes,
we
didn't
organize
them
with
beans,
but
we
were
able
to
come
up
with
something.
So
this
morning
session
is
about
the
pervasiveness,
the
far-reaching
this
of
kubernetes
as
Dan
informed
us
earlier.
You
know
there's
a
lot
of
different
ideas
out
there
from
a
lot
of
different
places
and
as
Cheryl
said
you
know
this
is
she
put
out
some
numbers?
A
A
A
A
It
says
all
right
so
be
sure
to
visit
with
all
of
our
sponsors
in
the
sponsor
showcase,
while
you
are
here
and
make
sure
that
you
thank
them
for
helping
us
put
on.
You
know
this
really
amazing
event
and
real
quick
I'm
gonna
go
off-script.
I
know
they
hate
this,
but
I
met
two
of
the
diversity
scholarship
sponsors
over
the
last
two
days.
You
know
what
you
thought
you
had
ideas.
These
folks
flew
halfway
around
the
world
and
they
have
crazy
ideas.
A
A
A
So
now,
I'm
gonna,
move
on
to
the
project
update
so
every
year
we
do
this.
What
we're
going
to
do
is
we're
going
to
give
an
update
for
a
few
of
the
CNCs
projects,
but
first
I
want
to
set
the
stage,
because
one
thing
that
I
didn't
have
was
context
about
how
CNCs
projects
work.
So
there's
this
quote,
and
it's
funny
that
Dan
actually
mentioned
this
in
his
talk,
but
I'm
gonna
read
a
part
of
it.
A
It
says
here
that
cloud
native
technologies
empower
organizations
to
build
and
run
scalable
applications
and
modern
environment
on
dynamic
varmints
such
as
the
public
and
private
clouds,
and
then
containers
and
service
messages
and
micro
service
is
immutable.
Infrastructure
and
declarative.
Api
is
exemplify
this
approach,
so
what
we
have
and
all
the
CNSC
F
projects
are
projects
that
just
helped
further.
This
mission
and
I
like
to
show
this
right
here.
This
is
the
crossing
the
chasm
chart
or
the
chasm
I'm.
A
Sorry
I,
not
good
at
this-
is
the
crossing
the
chasm
and
really
what
we're
doing
is
we're
trying
to
show
where
these
projects
fit
so
you'll
notice
on
starting
from
the
left.
We
have
the
innovators
and
then
we
have
the
early
adopters,
and
then
we
have
the
early
majority,
and
this
is
what
we're
focusing
on,
and
what
we're
looking
at
here
is:
how
do
we
help
set
up
projects
from
the
beginning
and
grow
them
into
mature
projects?
A
So,
right
now
there
is
16,
plus
16
plus
6,
so
that's
32,
plus
6,
there's
38
projects
and
the
CNC
F
is
sponsoring
right
now
and
what
we're
going
to
do
is
go
over
a
few
of
them.
Starting
with
staging
first
up
is
open,
EBS
and
what
open
EBS
is.
Is
it
enables
container
attached
storage
using
kubernetes
itself
as
a
substrate
for
storage
management?
Now
I
was
not
familiar
with
this
project,
but
as
I
dug
into
it
more,
it's
actually
pretty
interesting.
So
really
what
it
does
is
it
provides
multiple
functions.
A
Your
storage
can
live
in
a
container
or
open.
Bs
can
provide
a
common
layer
that
allows
for
for
developers
to
have
a
single
experience
for
exposing
storage,
and
so
here
are
the
updates.
So,
with
the
latest
update,
we
have
container
attached
storage
now
and
we
actually
have
a
lot
of
fixes
for
an
addition
to
the
data
engines.
So
there
is
the
Java
and
then
there's
C
store,
and
then
there
is
also
this
open,
EBS
extension
of
local
PBS.
A
So
moving
on,
let's
go
to
incubating
and
we
have
a
few
projects.
We
want
to
talk
about
nine
key
baiting.
The
first
one
is
linker
D
and
linker.
D
is
the
lightweight
service
special.
It
enhances
your
applications,
observability
reliability
and
security.
Without
code
changes
now
I
could
do
and
say
a
whole
bunch
of
things
about
linker
D,
but
I
want
one
better.
So
what
I
like
to
do
is
invite
all
our
gold
up
on
this
stage
from
buoyant,
so
he
can
actually
tell
you
the
updates
for
this
project.
B
Well,
I
could
tell
hello:
what's
up
how's
it
going,
my
name
is
Oliver
Gould
I
am
the
CTO
at
point
and
creator
of
Linkwood
II,
the
last
nine
months
or
so.
Since
we
launched
link
32
I.
Think
September
of
last
year
has
been
an
incredible
ride.
The
project
has
never
been
at
a
faster
pace:
we've
had
33
edge
releases
since
then
five
stable
releases
and
we'll
have
a
sixth
stable
release.
Once
we
all
recover
from
coop
con,
there's
been
a
whole
bunch
going
on,
because
we're
focused
on
being
lightweight.
B
B
We
also
have
some
exciting
things
coming
up
in
the
next
release.
I'm,
not
gonna,
really
talk
about
that
here.
Today,
I
heard
there
may
be
some
crazy
demos
at
a
keynote
this
afternoon.
No
so
check
that
out
and
of
course,
I
have
to
tell
you.
We
have
a
boost
in
the
vendor
haul.
If
you
see
anyone
wearing
a
link,
ranee
logo-
and
you
have
questions,
that's
what
we're
here
for
come.
B
Ask
us,
but
before
I
get
off
the
stage,
I
want
to
try
to
do
a
little
demo
of
my
favorite
new
limb,
pretty
feature
automatic
MPLS
for
service
identity,
and
so
when
you
deploy
micro
services,
especially
in
Cooper
Nettie's,
but
anywhere,
they
tend
to
share
lots
of
data
over
the
network.
That
means
anyone
sitting
on
the
network
sitting
on
the
node.
Anyone
with
net
admin
and
kubernetes
can
actually
just
sniff
this
traffic,
and
so
here
we
see
just
sniffing
some
TCP
dump
traffic.
We
get
user
names,
ages
dates
social
security
numbers
whatever.
B
That's
just
really
not
an
acceptable
default
in
our
opinion,
and
so
we
set
out
in
liquid
e23
to
make
this
happen
by
default,
and
so
now
proxies
provision
their
own
identity
without
configuration
when
they
start
up
an
proxies,
communicate
with
each
other.
They
validate
identity.
Do
encryption
and
everything's
good
to
go
so
now
when
we
go
try
to
tap
into
our
TCP
stream
here.
What
we're
gonna
find
is
that
we
see
nothing.
We
see
only
health
checks
and,
of
course,
a
bunch
of
opaque
TCP
traffic.
Again.
B
This
is
all
part
of
lincolnÃs
philosophy
of
being
lightweight,
providing
zero
config
solutions
so
that
you
don't
have
to
change
your
code
to
get
this
benefit.
If
that's
interesting,
you
come
talk
to
us.
Come
find
me
come
find
anyone
flynn,
currie
logo.
I
hope
you
have
a
great
conference
thanks.
Everybody.
A
You
know
they
sat
right
up
front
just
for
that,
so
I'll
miss
a
kubernetes
package
manager
who
here
has
not
heard
of
helm
all
right.
Well,
we
do
have
some
exciting
updates
straight
from
the
helm
team
here.
So
what
is
helm
because
they
provided
these
slides
and
I
want
to
do
them.
Justice,
helm
is
a
coin,
is
client-side
only
what
it
does
it
pushes
the
chart,
so
it
listens
with
helm.
Three
now
pushes
the
charts
and
those
CI
registries
it
validates
chart.
Reddish
tar
configuration
with
JSON
schema
release.
Names
are
scoped
to
namespace.
A
This
is
a
big
deal
because
now
in
production,
you
won't
have
elephant
dog
as
a
release,
name
because
no
no,
no,
you
don't
want
that.
You
call
it
what
it
is.
My
next
startup
idea,
but
also
you
can
declare
dependencies
in
chart.
Yeah
Millan
set
a
requirements,
camel
one
less
file
and
there's
a
lot
more
in
the
release,
notes
I
swear
the
update
I
got
from
this
was
like
50,
slides
and
I'm,
like
I.
Can't
do
all
this
so
I'm
going
to
distill
it
down,
but
here's.
A
A
And
more
and
more
importantly,
now
you
can
validate
your
chart
values,
that's
a
very
big
deal,
and
the
neat
thing
is
now
that
we
don't
have
to
concoct
all
these
mechanisms
to
actually
share
our
charts
home.
Three
actually
will
support
library,
charts
so
more
information
about
how
three
there's
the
github
release,
notes
on
the
release
so
go
to
v3,
LSH
and
then
there's
also
a
bloom,
a
a
blog
post
at
Elm
bug,
IRA,
slash
blog,
so
another
announcement
for
helm.
A
The
helm
summit
is
happening
this
year
on
September
11th
to
12th
in
Amsterdam,
and
the
CFPs
are
open
deadline
is
in
two.
This
month,
I
went
to
the
first
helm
summit.
It's
a
good.
It's
a
good
event.
I
swear,
there's
a
lot
of
smart
people
in
the
helm
community
that
get
together
to
share
all
these
information.
You'd
be
very
surprised
at
the
quality
of
talks
and
the
quality
of
information
to
share
it.
There
so
helm
talks.
Well
today
we
have
a
helm.
A
No,
yet
today
we
have
a
helm,
101
and
then
there's
also
on
Thursday,
there's
a
helm,
deep
dive
if
you're
really
interested
in
helm.
This
is
where
you
should
be
next
up
is
Harbor.
Now
harbour
is
an
open
source
cloud
native
registry
that
enables
organizations
to
enforce
policy
and
compliance
for
container
images
ma'am
today,
I
believe
VMware
is
announcing
that
harbor
1.8
has
been
released
and
I
swear.
A
The
there
is
a
lot
of
changes
in
helm,
one
that
eight
I
was
not
going
to
put
the
whole
blog
post
in
slides,
so
I
will
do
you
a
favor
and
summarize
them
for
you
really
quick.
So
really.
What
we
have
now
is
support
for
open
ID
connect.
This
is
with
more
ways
to
authenticate
people
to
connect
to
your
helm,
there's
also
support
for
robot
accounts.
A
Now,
there's
also
more
features
the
helm
team
has
been
really
are.
Our
team
has
been
really
really
busy
and
so
there's
the
health
check,
API
and
then
there's
also
an
update
to
the
docker
registry
so
to
make
sure
that
they
are
compliant
with
the
latest
version
from
docker
and
then
there's
also
support
for
scheduled
tasks.
A
Next
up
is
rook,
rook
is
production
at
ready
block
and
file
storage,
so
Brooke
recently
has
released
rook
top-10
and
once
again,
I
will
say
that
there
was
a
very
long
blog
post
about
this,
but
I
read
it
for
you
and
took
out
the
highlights.
The
big
features
are
there
in
the
Ceph,
the
edge
F
ace
and
the
mini
operators.
A
There
is
up
there's
big
updates
for
all
of
them,
especially
for
Steph
Steph
I,
think
it
moved
up
to
the
current
version,
the
Ceph
and
then
minion
op
and
edge
FS
operators
operated
as
well,
but
take
a
look
at
the
edge
FS
there's
actually
support
for
backups
now,
so
you
can
actually
back
up
these
things.
It's
actually
made
for
that
and
then
there's
CSI
support
and
then
also
this
blog
posts
here
at
the
end.
A
So
next
up
is
cryo
from
Red
Hat
cryo
is
an
implementation
of
the
kubernetes
CRI
to
enable
the
use
of
OC
I
compatible
runtimes
and
before
I,
bring
up
Arashi
I
want
to
say
that
that
is
a
mouthful.
So
really
what
it
is
is
saying
that,
because
kubernetes
is
a
platform
for
platforms,
all
the
things
for
all
the
people
we're
not
stuck
with
just
having
container
deep.
We
can
actually
replace
the
runtime
and
cryo
is
a
good
example
of
that.
C
Thank
You
Bryan
and
hello
Barcelona
wow.
This
is
amazing.
I'm
super
excited
to
be
here
to
talk
to
you
all
of
a
crier
today
that
this
is
my
first
keynote
by
the
way
so
yeah
so
honest,
cryo
cryo
is
a
lightweight
container
engine
designed
for
making
kubernetes
container
deployments
a
seamless
and
secure
as
possible,
as
Brian
mentioned,
cryo
implements
the
kubernetes
container
runtime
interface.
Hence
the
CRI
and
cryo.
C
The
O
stands
for
open
container
initiative,
which
means
that
cryo
supports
all
those
two
compatible
images
and
all
those
two
compatible
run
types
such
as
run
C,
cada,
G,
visor
yeah.
We
were
really
creative
with
the
name
weren't
we
so
cryo
was
built
to
embody
UNIX
philosophy,
which
states
that
you
should
design
programs
to
do.
One
thing
to
do
it
well
and
to
work
well
together
to
that
end
prior
uses
a
number
of
libraries
under
the
hood
that
focus
on
different
areas
of
the
container
space,
such
as
storage,
image
management
and
networking.
C
So
now
cryo
is
not
built
to
solve
all
the
problems
out
in
the
container
space.
It's
a
huge
area
right,
crys
main
focus
is
to
run
containers
securely
and
seamlessly
in
production
environments.
We
have
other
products,
such
as
builder
apartment
and
Scorpio,
which
use
the
same
libraries
to
solve
adjacent
problems
in
the
container
space
cry
of
love's
kubernetes
kubernetes
as
Pryor's
one
true
love.
What
this
means
is
that
cria
does
not
cater
to
any
other
orchestration
container
orchestration
tools
out
there
cryo
tailors
itself
to
meets
the
needs
of
kubernetes,
only
nothing
more,
nothing
less.
C
Now,
let's
take
a
look
at
Criers
journey
so
far,
so
in
early
2016
there
was
a
proposal
to
create
a
simpler
and
more
secure,
OC
a
compatible
container
engine
just
for
kubernetes
and
voila.
A
few
months
later,
in
September
2016
trial
was
born
about
a
year
later
in
October
2017
prior
1.0
was
released
and
since
then
we
have
got
multiple
versions
of
trial,
with
version
wonderful
team
being
our
most
recent
stable
version
and
finally,
the
best
news.
We
joined
CNCs
last
month
as
an
incubation
project
and
we're
all
super
pumped
to
finally
be
here.
C
So,
as
of
now
cries
running
in
production,
at
Red,
Hat,
Seussian
lust
cry,
oh,
is
going
to
be
the
only
supported
container
engine
and
OpenShift
4.1
and
above
and
prior
has
integration
with
tools
such
as
cystic
aqua.
Second,
video,
just
to
name
a
few.
We
have
over
a
hundred
contributors
and
Red
Hat
loves
Suzy
until
and
IBM
are
main
contributors
and
users.
So
far
now
I'm
sure
has
somewhat
intrigued.
C
You
guys
to
dig
more
into
cryo
and
if
you
all
can't
wait
to
get
started
here,
the
links,
tour,
github
page
and
our
website
we're
available
on
IRC
and
slack
and
don't
forget
to
stop
by
the
Red
Hat
booth
for
some
technical
demos
and
crier
and
friends
as
well
as
a
chat
with
us
and
get
some
cool
swag.
So
I
would
ask
sorry.
I
would
like
to
end
by
giving
a
quick
shout
out
to
our
prior
contributors
and
weekly
news
out
there
for
their
amazing
effort.
Thank
you
and
have
a
great
coupon.
Everyone.
A
So
next
up
is
open
census
and
open
tracing
I've
actually
used
both
of
these
and,
as
I
was
sitting
down
to
write.
The
updates
for
these
I
realized
that
you
know
what
I
can
do
better
than
that.
So,
instead
of
me
actually
talking
about
their
updates,
I
brought
someone
from
open
census
and
someone
from
open
tracing
here
to
talk
about
what's
new
in
their
world.
So
let's
welcome
to
the
stage
been
single
man,
flight
step
and
Morgan
MacLean
from
Google.
They
have
some
pretty
cool
stuff
to
share
today.
D
Hi
everyone
for,
for
the
record,
I'm
Ben
mr.
Morgan,
thanks
a
lot
for
coming
to
this
wonderful
conference.
So
let's
talk
about
your
application
and
how
different
people
see
it.
So
your
end-users
see
something
very
small
and
elegant
this,
this
little
white
iceberg
that
they
use
and
derive
tremendous
business
value
from
I'm
sure.
Let's
talk
about
how
someone
else
sees
it.
This
is
how
your
CFO
sees
your
application.
D
It's
probably
more
accurate!
It's
big!
It's
got
a
lot
of
stuff,
it's
real
expensive
things
are
breaking.
They
probably
don't
understand
what,
but
they
know
that
there's
a
bunch
of
shipwrecks
at
the
bottom
and
that
they're
getting
charged
for
those
too.
So
your
CFOs
not
very
happy
about
it,
but
so
be
it.
And
then
you
have
the
developers
the
developers
understand,
though
this
is
not
a
monolithic
iceberg.
This
is
actually
a
microservice
iceberg.
There's
lots
of
little
pieces
they're
all
talking
to
each
other.
I,
don't
really
get
it,
but
like
there
are
lots
of
services.
D
I
would
argue
that
none
of
these
people,
the
end-user,
your
CFO
or
at
least
you're
devs.
If
this
is
their
view,
actually
have
any
idea.
What's
going
on,
that's
my
contention
what's
actually
happening,
that's
the
big
question
what's
actually
happening,
the
answer
has
to
do
with
observability.
This
is
a
very
trendy
word.
It's
not
a
new
word.
It
was
coined
in
the
1960s
in
the
control
theory.
Discipline.
It's
basically
observability
is
a
measure
of
how
well
you
can
understand
the
system,
given
only
the
telemetry
data
that
escapes.
So
this
is
really
important
to
understand.
D
Your
ability
to
observe
your
system
cannot
be
any
better
than
the
data
that's
coming
out.
If
you
have
to
redeploy
your
software
in
order
to
see
it
you're
already
losing
so
the
telemetry
is
it's
the
essence
of
any
observability
strategy,
not
the
entirety,
but
the
essence
of
it.
So
if
telemetry
is
so
important,
it
must
become
a
built-in
feature
for
cloud
native
software.
D
It
must-
and
this
is
what
typically
in
our
world
that
we
inhabit
is
distributed:
traces
metrics
logs,
that
sort
of
thing
and
it's
necessary
for
us
to
find
some
way
to
amortize
the
cost
of
getting
high-quality
telemetry.
It
is
not
reasonable
for
every
developer
to
hand
instrument
every
service.
We
need
to
somehow
manage
to
move
that
effort
into
the
open
source
and
do
it
once
for
everyone
who
depends
on
cloud
native
software.
E
And
so
this
is
a
good
pitch
right,
so
anyone's
worked
in
this
space.
It's
probably
pretty
aware.
The
telemetry
is
important
and
they're
also
really
aware
that
getting
it
out
of
your
application
is
incredibly
difficult.
The
problem
with
this
pitch
is
you've,
heard
it
before
both
from
open
tracing
and
from
open
census.
Now
intuitively.
You
might
not
think
that
this
is
a
problem.
Hey
there's
two
successful
projects
here,
they're
both
growing
quickly.
They
both
achieves
some
amount
of
escape
velocity.
E
This
is
this
is
a
challenge,
though,
for
anyone
who
maintains
things
like
shared
code,
so
shared
libraries
web
frameworks
that
you
might
use.
They
now
have
two
different
interfaces
that
they
need
to
bind
against
if
they
want
to
export
traces,
metrics
or
other
types
of
telemetry.
This
is
also
problem
for
service
developers,
who
often
want
a
single
implementation
to
work
against.
E
So
we've
heard
all
the
challenges
that
people
are
having
in
the
space
and
we're
really
excited
to
just
talk
about
open,
telemetry,
open
telemetry
is
what
we're
merging
open,
tracing
and
open
census
into
both
the
projects
and
the
communities
themselves.
It's
effectively
the
next
major
version
of
both
open
tracing
and
open
census
is
what
people
will
migrate
to
open.
D
Yes,
so
we
only
have
a
few
minutes
up
here
today,
because
it's
a
new
project,
it's
very
important
for
us
to
actually
talk
with
end-users,
get
your
feedback
and
share
more
details,
so
we're
hosting
two
separate
workshops.
Today,
one
is
just
a
panel
discussion
which
is
basically
to
say
that
a
bunch
of
folks
from
open
telemetry,
which
is
open,
census
and
open
tracing,
will
be
there
to
answer
questions
about
the
project,
how
we're
maintaining
backwards,
compatibility,
etc
backwards
compatibility
is
such
an
important
thing
that
we
have
a
special
session
just
about
that.
E
A
So
we're
gonna
conclude
this
session
with
something
that's
pretty
pretty
important.
It's
the
fact
that
there's
actually
a
process
that
C
and
C
F
projects
run
through.
So
we
heard
we
talked
about
a
staging
project.
We
talked
about
a
couple
of
products
in
our
inner
incubator,
but
now
what
we're
really
trying
to
do
is
move
projects
to
this
graduated
state
and
really
what
that
is.
Is
that
is
the
project
in
early
positions
that
have
an
early
fast
mover?
Is
it?
Is
it
really
the
early
majority?
Does
it
has
it
captured
the
mind
chair?
A
Does
it
have
all
the
things
that
have
proper
governance?
Is
it
just
a
proper
project?
So
right
now
we're
up
to
six
graduated
projects.
The
first
graduated
project
was
kubernetes.
It
was
but
now
we're
up
to
six,
so
the
six
project
is
fluid
D
and
they
just
graduated
and
because
I,
like
this
theme
of
me,
not
talking
about
other
people's
software,
what
I'm
going
to
do
is
I'm
going
to
invite
Eduardo
Silva
up
from
arm
up
here
to
talk
about
flow
in
D,
so
Eduardo.
F
Well,
yes,
how
are
you
doing
fine
good,
that's
good
first
day,
so
my
name
is
Eduardo
Silva,
a
principal
engineer
at
arm
and
today
a
before
to
talk
about
flu.
Indeed,
self
we
are
going
to
talk
about
logging
and
log
in
itself
exists
because
we
want
to
solve
all
the
data
analysis
problem.
You
cannot
do
their
analysis
if
you
don't
have
the
data
right.
We
have
databases,
but
there
is
a
fundamental
piece
for
that.
F
F
Flu
indeed
has
its
origins
on
2011,
a
treasured
data
company,
which
is
now
part
of
arm,
and
at
that
moment
a
if
use
Hyper
companies
is
going
to
sell
a
service
to
proceed
era.
Do
you
need
to
collect
the
data,
but
the
huge
win
here
is
that
it's
not
about
what
we
did
as
a
company
to
solve
our
problem,
because
also
when
we
create
a
fluent
e
flu
and
it
was
created
in
open
source
and
happened
that
at
the
same
time,
many
people
has
the
same
problem.
F
We
have
a
bunch
of
the
area
for
different
formats,
different
streams
and
they
didn't
know
how
to
collect
this
data.
After
that,
we
went
through
a
huge
adoption
of
this
because
of
open
source
and
it's
not
after
well
a
few
years
of
about
2015
one
of
our
people
from
one
person
from
our
team
implemented
the
native
daughter
logging
drivers,
which
means
on
that
moment
the
ability
to
take
your
docker
container
and
ship,
the
logs
straightaway
over
the
network
to
flew
in
the
escaping
the
file
system
or
any
kind
of
storage
in
persistency.
F
On
that
moment,
one
year
after
that,
we
join
it
as
CN
CF
in
2016
as
an
incubation
project,
and
then
one
year
after
that,
we
reach
one
that
oh,
but
it
was
not
until
now
until
April
of
this
year
that
flew
indie
has
officially
graduated.
It
took
us
two
years,
but
well.
I
would
like
to
get
some
applause
for
for
the
team
that
really
working
really
hard
on
this.
F
And
how
about
one
pretty
graduated
is
not
about
to
get
an
attack
or
a
label
a.
We
are
graduated,
it's
more
than
that.
It's
a
responsibility
because
we
have
a
huge
adoption.
Most
of
cloud
providers
are
using
fluently
most
of
companies
that
had
logging
problems.
They
use
fluent
E
and
I'm
talking
about
Microsoft
Google,
Red
Hat
being
world
by
bottle.
So
an
in
generally
statistic.
You
can
see
that
now
we
have
in
general,
almost
two
hundred
releases.
The
community
has
built
a
thousand
of
plugins.
F
That's
one
was
I
would
say
that
the
second
key
of
success
of
Lundie.
It
was
not
just
to
ship
data
to
just
one
database.
Let
the
community
decide
what
to
build,
to
which
integrate,
which
kind
of
filters,
and
nowadays
is
a
thousand.
We
have
a
thousand
github
stars
and
we
hope
to
reach
200
contributors
this
year
and
one
of
the
channels
to
distribute
fluent
D
is
docker
hub.
F
If
you
are
deploying
on
kubernetes
or
any
cloud
native
solution,
you're
using
docker
hub
and
now
we
reach
a
hundred
million
downloads
or
docker
hub
pools
and
talking
about
a
to
the
fluent
in
my
internet,
who's
master.
Here
on
the
Kagawa,
he
is
here
with
you
this
time.
I
asked
him
a
what
what
do
you
think
about
this?
F
F
Everything
can
be
more
performant
and
everything
can
be
more
secure
and
that's
what
we,
what
is
flowing
the
1.5
in
the
forward
protocol
to
send
data
from
one
through
indeed
to
another,
we
implemented
a
full
keep
Ally
filter
that
was
not
there
before,
but
it
implements
and
allows
you
to
ship
data
faster
and
optimize.
The
bandwidth
also,
we
implemented
more
security
in
some
output
plugins
and
we
improve
it.
F
A
support
to
have
more
scalability
on
environments
where
we
have
multiple
CPU
cores
and
in
community
added
to
a
fluent
is
now
part
of
docker
hub
library,
so
you
can
do,
can
run
fluency
and
you
will
get
the
latest
version
officially,
and
this
is
not
maybe
too
fancy.
But
what
is
fancy
is
that
we
have
support
for
ARM
architecture,
I'm,
64
and
32.
F
So
in
a
community
one
thing,
the
most
most
important
thing
is
the
community
and
I
will
say
that
the
second
one
is
documentation.
So
the
team
put
a
lot
of
effort
on
write,
a
new
documentation
website
for
you,
because
if
you
will
look
at
the
stats
of
a
project,
actually
it's
not
about
how
many
people
hits
flew
in
the
web
site.
It's
about
how
many
hits
do
we
really
get
in
the
documentation
side
and
thinking
about
how
to
improve
testing
holy
yci,
for
example,
and
who
has
created
a
pull
request
here.
F
Raise
your
hand,
half
of
you.
Okay,
pull
request
is
a
contribution
when
that
contribution
hits
an
open-source
project.
That
project
needs
to
run
a
couple
of
tests
over
your
contribution
because
we
don't
want
to
have
backs,
we
want,
we
don't
want
to
have
regressions.
So
how
do
we
improve
that?
So
fluent
II
team
has
officially,
we
did
laughs.
F
So
what's
new,
also
from
a
community
perspective,
if
you
want
to
learn
more
about
fluently,
we
have
a
new
training
course.
We
has
been
created
by
Linux
Foundation
and
our
exam.
You
can
learn
how
fluent
it
works,
how
to
configure
it,
how
to
take
the
best
practices
for
cloud
native
environments
and
it's
already
available,
and
it's
not
just
about
fluently
fluency
I,
always
say
that
it's
a
huge
ecosystem.
We
have
SDKs
languages
and
so
on,
full
and
bit,
which
is
one
a
sub
price
of
flu
in
D
it's
getting
into
H
computing.
F
It's
solve
login
problems,
but
now
we
are
getting
with
stream
processing
capabilities.
If
you
know
what
stream
processing
in
is
it's
basically,
the
ability
to
perform
data
processing
when
the
data
is
in
memory
is
in
motion
and
we
are
investing
a
lot
of
effort
in
general
effort
on
that
area.
So,
if
you
want
to
know
more
about
flu
and
the
flu
embed,
the
whole
team
is
here
in
the
conference.
So
today
we
have
a
talk
about
2:00
p.m.
about
flu.
F
A
Thank
you
all
right,
so
that
brings
us
to
the
end
of
our
CSS
project.
Updates.
I
just
want
to
leave
that
with
there's
a
lot
of
people
out
there
doing
a
lot
of
hard
work.
Writing
this
software
and
giving
it
to
us.
You
know
it's
a
lot
of
blood,
sweat
and
tears.
I
didn't
mention
notice.
I
said
there
was
16,
plus
16,
plus
6
projects,
I
believe
that's
38.
We
didn't
mention
that
many
projects,
but
there's
definitely
work
going
on
out
there.