►
From YouTube: The Past, Present and Future of the Knative Community
Description
The Past, Present and Future of the Knative Community - Moderated by Michael Maximilien IBM; Evan Anderson, VMware, Roland Huß, Red Hat; Whitney Lee, VMWare; Sebastien Goasguen, TriggerMesh
With the 1.0 release and becoming a CNCF incubating project, the Knative community has reached two significant milestones recently. Join us for a fireside chat about the beginnings when Knative was still called Elafros, how the community evolved over the last four years, and what big features are on the horizon. Also, learn about how you can become part of the Knative community and join the journey.
A
Yeah
hello,
my
name
is
roland
huss,
I'm
from
red
hat
working
there
on
openshift
serverless,
which
is
the
product
based
on
canadian
and
in
the
community.
I'm
part
of
the
toc,
and
also
the
canadian
client
working
group
lead
or
one
of
the
client
working
group
leads
yeah.
That's
me.
B
I'm
evan
anderson
I've
been
doing
k-native
for
the
last
four
years
or
so,
and
I'm
currently
security
working
group
lead
and
on
the
toc
currently
at
vmware.
C
D
And
I'm
sebastian
guess
again:
I'm
the
co-founder
of
trigger
mesh
and
I
was
involved
in
k
native
almost
since
the
beginning.
Okay,.
E
Excellent,
so
I'm
max
I
work
for
ibm
and
I'll
try
to
be
moderating
so
the
the
panel
is
about
the
past
present
and
future.
I
figured
I
will
start
with
three
questions
on
those
three
topics,
but
I
want
you
to
be
involved.
So
after
those
three
questions,
I
will
look
for
audience
members
with
questions
and
then
run
to
you.
So
you
can
ask
your
question
so
get
ready
to
be
prepared
and
then
I
will
ask
the
panels
to
limit
their
answers
so
that
we
can
have
time
to
make
sure
we
get
to
the
questions.
E
Okay,
all
right,
very
good,
all
right,
so
the
first
and
people
don't
have
to
also
answer
every
question,
like
everybody
has
to
answer
right
like
if
you
don't
have
an
answer,
then
that's
fine.
Let's
move
to
the
next
question.
Okay,
so
the
first
question
I
have
which
is
sort
of
talking
about
the
past-
that
just
make
it
fun.
B
So
this
is
a
little
bit
of
pre-history
villa
over
there
and
myself
and
brett
klauser
at
google
were
actually
working
on
a
next
generation
of
the
app
engine,
managed
vms
project
and
trying
to
see
if
we
could
move
things
to
kubernetes
and
shortly
after
that
there
was
a
blog
post
that
joe
beta
ended
up.
Highlighting
saying
you
know,
basically,
kubernetes
is
too
hard
for
developers.
B
You
know
hey.
If
you
want
to
start
an
application,
you
need
to
understand
pods
and
containers
and
deployments
and
replica
sets
and
services
and
horizontal
pilot,
auto
scalers,
and
you
know
all
of
a
sudden
everything
you
needed
to
know
about.
Your
application
fell
out
your
ear
because
you
packed
all
these
networking
and
distributed
systems.
Concepts
in
and
we
said
hey.
I
bet
we
could
do
something
about
that.
E
B
D
Sure
I
have
a
story,
so
I
had
created
cubeless.
I
think
that
was
december
2016..
You
know
there
was
a
lot
of
faz
solutions.
You
know
in
the
community
space
and
you
know
fast
forward
to
2018
I'm
contacted
by
dewitt
clinton
clinton
at
google
he's
like
hey
we're
doing
something
and
so
on
and
at
the
time.
Well,
you
know
I
was
transitioning
from
another
company
and
he
said
you
know:
do
you
want
to
be
part
of
the
native
announcement
that
at
google
next,
that
was
july
2018?
D
And
I
said
you
know
sure,
but
I
I
don't
have
a
company
yet
so
you
know
with
my
co-founder.
We
rushed
to
get
a
logo
400
bucks
on
99
designs.
We
you
know
we
called
each
other.
We
came
up
with
a
name
and
and
suddenly
it
was
funny.
If
you
go
back
to
youtube,
you
see
the
announcement
and
it's
like
you
know,
thank
you
to
ibm
and
t-mobile
and
pivotal
and
trigger
mesh,
which
was
a
you
know,
not
incorporated
yet.
So
that
was
quite
funny.
A
Yeah,
actually,
I
joined
canada
quite
late
in
the
game.
I
think
it
was
22
but
20,
something
like
that
two
years
and
yeah.
I
was
always
fascinated
by
the
by
the
really
the
the
ease
of
the
developer
experience
and
really
this
was
really
great
and
I
really
wanted
to
be
part
of
that.
I
have
to
say
I
was
really
accepted
by
the
community.
Very
there
was
a
very
rare
reception.
E
Okay,
so
the
next
question
is
very
specific:
it's
gonna
be
the
present.
We
are
here
first
k-native
khan.
Obviously
we're
part
of
the
cncf
tell
me
what
you're
looking
for
the
most
with
being
part
of
the
cncf.
So
maybe
we'll
start
with
you
with
you
whitney
like
what
what
do
you
expect
the
cncf
to
do
for
us
or
maybe
to
hurt
us
or
no?
I
don't
think
so
right,
but
you
tell
me
what
do
you
think
you
know
this?
This
means
to
you
being
part
of
cncf.
C
E
F
Well,
convincing,
I
guess
google
but
yeah
the
cncf
is
a
a
lot
of
people
actually
think
that
we
were
in
cncf
a
lot
of
people
talk
to
me
and
say
I
see
your
logo
in
the
cncf
landscape
and
and
they
think
that
we
are
in
the
cncf,
which
is
a
good
thing,
but
it's
not
was
not
real,
so
moving
to
the
cncf,
it
gives
us
an
umbrella
of
services
and
help
in
terms
of
legal
services,
marketing
services,
infrastructure,
we're
running
infrastructure,
thankfully,
for
for
google
that
started
the
project,
but
now
we're
moving
into
our
infra
and
people
like
hippie
kaleb.
F
I
think,
is
here
those
folks
from
ip
helps
now
now
so
we
grow
with
the
community.
I
think
it
get
us
closer
to
kubernetes.
So
one
of
the
things
that
I
didn't
encourage
is
to
join
and
meet
people
from
kubernetes.
I
know
a
lot
of
folks
already
have
like
relationships
with
kubernetes.
I
try
to
join
sig
release,
so
I'm
building
a
network
there
and
that
way
we
can
build
a
better
community
right
between
cncf
and
k
native.
E
C
What
I'm
hearing
is
it
moves
you
closer
to
kubernetes,
but
also
it
sounds
like
there's
a
lot
of
time,
money
and
effort
put
into
this
project
and
when
it's,
when
it's
owned
by
a
private
company,
that
is
a
little
bit
dangerous,
there's
a
lot
of
trust
being
put
in
google
and
now
the
trust
is
like
it's
there,
like
it's
owned
by
the
community
now
and
not
by
an
individual
entity.
D
So,
okay
I'll
do
a
quick
one
before
before.
Even
so,
you
know,
for
me,
cnc
moving
to
cncf
is
is
really
big
for
k
native.
I
really
thought
you
know
that
it
would
happen
much
sooner
right.
You
know
early
2018,
I
mean
you
know
that
would
have
been
great.
The
the
big
advantage
to
me
is
that
you
know
for
users.
D
They
want
to
know
that
the
project
is
sustainable,
that
it's
not
dominated
by
one
vendor,
so
that
you
know
people
want
to
know
that
they
I
mean
they
want
to
de-risk
the
investment
in
a
particular
software
right.
So
you
know
by
putting
in
cncf
it
doesn't
change
anything
in
terms
of
how
you
contribute
or
even
the
license-
or
you
know
things
like
this,
but
you
now
know
that
it's
you
know
under
an
umbrella.
D
E
B
Oh,
I
was
just
going
to
point
out
one
other
thing:
the
way
that
we're
all
here
we
talked
at
san
diego
in
2019
about
it
would
be
cool
to
do
a
k-native
con,
but
the
logistics
without
an
umbrella
organization,
where
everybody
you
know
who
wants
to
participate,
can
give
some
money
to
a
central
place
to
get
that
organization
to
happen.
It's
so
much
harder.
So
that's
one
of
the
things
I'm
excited
about
is
it's
going
to
be
easier
to
do
things
as
a
community
under
an
umbrella
like
cncf.
A
Yeah,
I
think
I
can
also
back
separations
opinion
about
that.
It's
really
much
easier
to
to
adopt
or
increase
the
adoption
of
canadian
when
you're
on
a
neutral
foundation.
This
is
really,
I
think,
the
big
big
win
for
me
and
also
for
for
users.
I
think,
because
they
have
the
kind
of
the
security
that
they're
are
not
go
away
anytime
soon.
Let's
say
like
this.
E
Yeah
and-
and
you
know
not
to
dwell
on
this
but
I'll,
give
you
a
specific
example.
Is
that
a
lot
of
time
when
somebody
wants
to
do
something
at
ibm
with
ibm,
they
go
to
me
as
if
I
represent
ibm,
but
you
know
I
may
not
be
there.
You
know
or
or
you
know,
maybe
I'm
busy
with
something
else
and
in
the
case
of
k-native
you
know,
izamal
was
the
person
that
represented
google
for
us,
but
now
we'll
have
an
organization
that
we
can
go
to
and
there's
a
formal
process
and
so
on
right.
E
So
you
don't
have
to
be
the
one
always
there,
even
though
we
were
great.
So
thank
you
all
right,
so
last
question
from
me
and
then
we'll
pass
it
to
you
so
think
about
your
questions.
Looking
to
the
future,
what
do
you
think
is
missing
nk
native
that
you'd
love
to
see?
Maybe
it's
a
community
thing.
Maybe
it's
a
specific
feature.
E
Maybe
it's
you
know
organizational
thing.
Think
about
like
what
do
you
think
you
would
like
to
see?
So,
let's
go
to
evan
first,
since
you
have
it
started:
yeah,
oh
geez,.
B
I
have
a
huge
list,
but
I
think
the
big
thing
that
I'm
really
excited
about
for
the
future
is
that
we've
been
improving
and
learning
on
how
to
make
this
software
in
the
open
for
the
last
couple
years-
and
you
know,
we've
got
a
big
feature
list
of
stuff
that
you
know.
Eventing
wants
to
work
on
some
task
flows.
Security
has
a
big
list
of
stuff
to
do.
There's
neat
stuff
going
on
in
the
networking
space
and
auto
scaling
and
functions.
B
But
we've
got
a
framework
for
doing
this
all
and
for
cooperating
and
coordinating
it,
and
I'm
looking
forward
to
that.
Getting
better
and
better.
D
Okay,
I'll
give
one,
so
you
know
one
thing
I
realized
fairly
recently
is
that
when
you
look
at
lambda
what
people
are
doing
on
lambda,
I
mean
it's
not
new,
but
they're
really
building
rest
apis
right,
and
I
think
I
for
some
time
I
forgot
about
this,
but
they
use
api
gateway
and
then
they
use
functions
to
really
build
a
rest
api.
D
So
we
can
do
this.
You
know
different
ways
with
with
knativ,
but
then
there
is
also
async
api.
So
I
want
to
bounce
back
on
on
your
talk
and
it's
a
it's
a
big
big,
huge
use
case
async
api-
maybe
not
so
much
for
you
know
what
you
described
like
long-running
jobs,
but
just
you
know,
event-based
integration,
where
you
know
we
work
a
lot
on
this,
but
we
have
to
really
you
know,
think
deeper
about.
D
C
Technically,
when
I've,
given
a
couple
talks
about
k-native
at
conferences,
I
see
interest
in
duration
based
roll
out,
so
I'm
excited
to
hear
about
when
that
comes
to
fruition.
So
I
can
pass
that
along
and
then
community
wise.
I
have
some
experience,
volunteering
with
the
kubernetes
community
and
in
the
coop
in
the
k-native
side
I
maybe
haven't.
C
I
think
there
could
be
like
a
more
of
a
welcoming
committee
or
a
really
clear
place
to
where
to
go
when
you're
first
getting
started
to
be
able
to
navigate
all
the
different
working
groups
and-
and
so
like
maria's
presentation
earlier
I
didn't,
as
someone
knew
the
community,
I
had
no
idea
that
group
existed
and
so
now
I'm
really
excited
to
participate
in
that.
So
how
can
we
amplify
that
to
someone
who's
who's
approaching
the
community
and
a
little
intimidated
by
everyone
about
where
to
go?
First,
yeah
and.
E
A
Yeah,
actually,
I'm
also
very
excited
about
the
more
boring
things
like
security,
as
mentioned
already,
so
this
is
really
the
things
which
people
are
really
requesting
for
production
ready
workload.
So
this
is
something
I
hope
that
we
will
make
a
big
leap
forward
in
the
near
future
and
the
other
thing
which
I'm
also
very
kind
of
a
passion
for
is
the
developer
experience.
I
I
think
that
can
have
get
the
abstraction
quite
exactly
right
for
for
developer
application
to
combine
everything
into
this
single
resource,
and
we
can
do
do
better.
A
I
think
function
is
a
very,
very
impressive
direction
that
we
are
going
to,
and
I
I
really
hope
to
to
see
there
more
more
traction,
even
yeah.
E
C
Yeah
so
I
first
heard
about
k-native
when
some
co-workers
in
my
same
stand-up
would
talk
about
it,
and
I
was
so
new
to
tech
that
I
didn't
really
understand
what
it
was,
and
so
I
like
what?
What
is
that
decay
native?
They
keep
mentioning,
and
I
heard
that
it's
serverless
in
that
it's
attached
to
istio
and
it's
so
it's
heavy
and
I'm
wondering
what
y'all
are
doing
to
help
get
rid
of
those
stereotypes,
because
now
that
I'm
deep
in
the
project
I'm
like,
oh
it
doesn't
need
us
yo
and
like
serverless
is
cool.
C
B
So
actually
the
istio
thing
is
kind
of
funny,
because
that
was
one
of
the
first
pieces
of
feedback
when
we
launched-
and
I
feel
like,
we
reacted
to
it
pretty
quickly.
Matt
probably
remembers
the
exact
timeline,
but
I
feel
like
they.
Maybe
you
know
it
was
july
that
we
launched
and
without
having
thought
about
it,
maybe
february
or
so
the
following
year
we
had
ambassador
or
glue
or
one
of
the
other
gateway
implementations,
and
it
may
not.
B
I
don't
know
if
anyone's
done
a
like
here's,
what
it
looks
like
when
a
request
comes
through
k-native,
but
we
do
a
substantial
amount
of
processing
on
that
gateway,
and
I'm
I'm
excited
that
the
gateway
api
is
going
to
finally
standardize
some
of
those
capabilities,
but
your
standard
kubernetes
ingress
is
not
sufficient
for
what
k-native
needs,
which
is
why
we
have
all
these
different
adapter
layers,
and
I
think
that
plug-ability
is
really
good.
But
it
sounds
like
we
have
not.
B
Matt's
found
the
docs
it
looks
like,
but
that
plug
ability
is
not
advertised
as
much
as
we
need
yeah.
D
I'll
just
share
a
little
story,
because
you
know
we're
between
friends
here
so
yeah
remo
I
mean
removing
the
dependency
on
istio
was
a
very
early
feedback
and
I
agree
with
you.
It
was
addressed
very
quickly,
but
you
know
the
the
perception
that
it's
still
a
dependency
persists,
which
is
you
know
interesting
from
almost
social
perspective.
D
The
the
funny
thing
is,
I
remember,
having
a
discussion
with,
I
think
brian
brian
greg
ryan
greg
the
pm
at
brian
greg
yeah,
and
you
know
I
was
telling
him
hey.
We
are
running
we're
running
k
native
and
we
have
you
know
we
have
issue
and
the
full
service
mesh,
and
you
know
that
thing
is
hard
to
debug
and
monitor
and
then
there's
a
huge.
You
know
startup
time
and
so
on,
and
he
said
oh
yeah,
but
sebastian.
You
know
in
cloud
run
we
disabled
the
mesh.
D
H
Hey
it's
me
again,
I'm
michael
garcia!
I
work
at
vmware.
Can
you
hear
me
is
that
okay?
Okay?
So
my
question
is
not
so
much
to
evan,
because
I've
talked
to
him
about
that.
So
it's
more
to
the
three
of
you
in
so
in
k-nativ
we
talk
a
lot
about
developer
experience
and
making
developer
experience
better
and
my
work.
H
My
daily
work
with
k-native
involves
working
with
administrators
operator
kind
of
type
of
personas,
and
I
realized
that
a
lot
of
these
persons
don't
consider
themselves
to
be
a
developer,
so
the
world
itself
is
kind
of
overloaded
a
little
bit
right,
because
I
know
that
k
native
doesn't
want
to
be
just
for
developers
or
business
developers
right
it's
for
everyone,
but
conceptually
and
perceptually
a
lot
of
people.
Think!
Oh,
that's
not!
H
For
me,
this
is
purely
for
javascript,
whatever
java
developers,
so
my
question
is,
and
especially,
if
you
look
at
aws
lambda
and
some
of
their
eventing
services,
which
are
heavily
used
for
automation,
notification,
like
not
necessarily
business
logic.
The
way
we
understand
business
logic
right,
but
it's
super
successful
there.
So
my
question
is:
how
can
we
grow?
The
canadian
community,
especially
these
kind
of
personas
in
the
future,
to
get
our
numbers
up
on
one
hand,
but
also
appeal
more
to
these
persons.
A
A
I
also
think
that,
for
example,
for
lambda,
the
big
big
asset
is
really
that
you
have
so
many
servers
that
you
can
connect
to
it's
not
so
much
the
programming
model,
how
you
could
connect
them,
but
the
sheer
availability
of
everything
like
that,
and
we
need
to
to
level
up
on
that
on
on
that
capability,
to
help
an
easier
way,
how
you
can
create
sources
and
bring
things
together
so
yeah.
I
don't
think
I
have
anything.
D
To
add
on
this
one,
so
I
mean
user
experience
you
know
to
me
is
huge.
These
days
and
and
honestly,
I
would
say
that
canadian
is
super
hard
to
use
right
and
I've
been
involved
since
the
beginning.
D
When
you
look
at
all
the
apis,
all
the
construct,
even
things
that
we've
done
at
trigger
mesh,
if
you
take
somebody
who
is
new
and
and
they
try
to
use
it,
it's
challenging
right
lots
of
things
that
you
need
to
understand
right,
ingress,
cloud
events-
you
know
things
like
this,
so
I
think
we
really
need
to
do
a
much
better
job
in
terms
of
developer
experience
right,
it's
not
on
boarding
people.
D
D
D
It's
not
a
feeling,
it's
just
even
us,
trying
to
use
things
if
you
try
to
build
a
real
application
with
like
events
and
so
on.
You
know,
you're
going
to
struggle
right
right,
riding
all
the
yammo.
I.
E
Mean
it's:
okay.
Did
you
use
kn?
That's
what
I'm
talking
about
because
there
are
there.
We
have
a
solution,
I'm
not
saying
it's
it.
It
may
not
be
as
good
as
you
you
you
need,
but
we
do
have
a
solution
and
at
some
point
they're
going
to
give
a
talk
about
that.
I'm
not
trying
to
defend
it.
I'm
just
saying
that
you
know
I'd
love
to
know
the
level
if
it's
yaml.
Yes,
that's
a
problem,
but.
D
Even
even
you
know
it's
interesting,
but
even
kn.
I
totally
respect
it.
I
understand
I
understand
it,
but
when
you,
when
you
ask
somebody
who
is
totally
new
to
start
using
kn,
they
need
they
need
kubernetes,
they
need
candidates,
they
need
tecton,
they
need
to
connect
to
a
docker
registry
and
so
on.
There
are
lots
of
things
happening
that
are
very
hard
for
you
know
an
operator
you
know
to
to
get
on
board.
E
Okay,
that's
good,
okay!
So
that's
feedback!
Anybody
wants
to
add
to
this.
Otherwise
do
we
have
a
question
from
the
audience?
I.
B
C
So
I
I
do
a
streaming
light
board
show
and
what
I
do
is
from
behind
the
light
board.
I
have
a
guest,
come
on
and
teach
me
a
concept,
and
then
I
draw
it
as
I
understand
it,
it's
a
pretty
long
format
show
so
I
had
carlos
come
on
and
teach
me
about
k-native
eventing
and
I
sorry
serving
and
I
drew
it
out
and
then
I
had
mauricio
come
on
and
teach
me
about
eventing
and
I
drew
it
out.
C
So
those
are
some
beginner
resources
to
at
least
get
the
high
level
concepts
down
and
then
for
me
personally
once
I
have
that
mental
model
in
place,
the
rest
of
it
making
making
use
of
it
comes
together
much
more
easily.
I
Hello
guys
drop
the
site
from
working
with
jpmorgan
and
chase,
so
I
have
a
question.
Basically,
the
current
function
model
is
very
http,
request,
centric
and
very
driven
by
the
http
request
right
so
like.
I
just
wanted
to
understand
that
if,
in
the
future,
is
there
a
scope
for
a
job
like
support
where
you
can
have
batch
data
transformations
and
administrative
tasks,
support
as
well.
B
The
benthos
one
yeah,
that's
definitely
one
thing
that
you
can
do,
particularly
if
you're
looking
to
fan
out
your
work,
a
lot
you
can.
Basically,
if
I
apologize,
if
I
misre
represented
beth
those
here,
but
it
looked
like
benthos
was
basically
you
know
queuing
up
one
request
for
each
piece
of
work
and
then
firing
it
off.
You
know
at
a
k-native
service.
The
other
thing
is,
you
still
got
all
of
kubernetes
out
there.
So
there's
nothing
says
you
have
to
only
use
knativ,
it's
totally
fine
to
reach
for
batchview
one
job
or
cron
job.
B
If
that's
the
right
tool,
you
know
k
native
the
k
there
is
really
is
for
kubernetes
and
all
that
kubernetes
is
there.
So
I
don't
think
we
have
specific
plans
for
like
a
job
type
interface.
At
the
moment.
I
A
Yeah
actually
yeah.
I
I
kind
of
share
the
same
feeling
like
evans.
Actually
I
like
really
the
the
the
focus
that
kinetic
has
on
this
particular
workload
and
actually
nothing
prevents
you
from
combining
this
with
other
workload,
types
and
other
workflow
frameworks.
So
actually
I
I'm
I'm
more
in
the
camp
like
stay
focused,
they
stay
make
that
what
you
are
doing,
make
it
good
and
make
it
well
yeah.
E
J
Hi
I'm
anthony
from
gamba
in
the
uk.
There's
a
lot
of
talk
around
aws
and
obviously
lambda
is
a
a
huge
service
used
a
lot
you're
on
about
kind
of
de-risking.
Do
we
or
do
people
have?
Obviously
your
huge
companies
talks
with
aws
and
speaking
about
lambda
and
how
we
can
try
and
make
sure
the
the
project
continues.
E
So
we
do
have
aws
representative
here,
but
let's
get
the
answer
here.
I
don't
want
to
embarrass
him.
He
doesn't.
A
Yeah
yeah,
so
actually
I
I
would
say
so
for
for
aws,
for
the
up
runner
is
more
the
equivalent
to
k
native
and
because
it's
container
based
so
actually
everything
in
kinetic,
is
container
based
and
we're
trying
to
provide
a
lambda-like
experience
for
containers
as
well
from
the
with
native
native
containers.
So
actually
there's
no,
I
think
for
aws,
there's
no
real.
So
this
is
my
feeling
is:
there's
no
real
need
to
go
to
knife
because
they
have
already
their
attack,
and
the
same
is
probably
true
for
microsoft
as
well
and
and
yeah.
A
So
this
is
kind
of
my
impression
that
that
yeah
that's
the
reason
why
aws
is
not
so
engaged
in
the
canadian
community.
D
D
So
you
know
that
that
you
can
do
no
problem
and
now
the
fact
that
lambda
supports
containers
straight
up,
you
know
it's.
I
would
say
it's,
even
it's
even
easier
right,
but
you
know.
Overall
I
mean,
when
you
know
to
me,
when
you
look
at
you
know,
serverless
what
it
is.
It's
really
a
fully
managed
service
right.
So
it's
the
you
know:
people
not
wanting
to
care
about
the
scalability,
the
operation
of
the
of
the
system.
It's
fully
managed.
You
know
cloud
right.
D
You
know
the
the
interface
per
se
like
the
function
thing
in
the
function,
interface
and
the
you
know
the
the
way
users
consume
it.
I
think
it's
actually.
You
know
a
little
bit
secondary.
E
K
And
no
promises
director
for
solution
architecture
for
serverless
at
aws,
I'm
here,
I'm
interested
for
sure
before
aws
I
was
at
ibm
with
carlos
and
max
and
so
have
a
lot
of
experience.
K
All
I
could
say
is
we
have
a
lot
of
customers
on
aws
that
run
kubernetes
and
use
k
native
on
eks
or
even
rosa
as
well
right,
so
rosa's
another
managed
openshift
offering
and
they
run
k
native
as
well
right,
obviously,
lambda
and
the
ecosystem
is
huge.
You
mentioned
the
integration,
but
I
think
there's
always
room
for
conversation
for
sure.
I'm
here
all
week.
Let's
talk.
B
I
I
saw
that
adverse
lambda
also
recently
offered
http
endpoints
directly
to
lambdas
without
having
to
go
through
api
gateway.
So
yeah.
E
F
So
I
didn't
introduce
myself,
I'm
carlos
santana,
I'm
on
steering
for
k
native.
I
work
for
ibm.
One
of
the
things
I'm
doing
lately
is
doing
user
interviews
for
admins
and
operators,
so
not
a
real.
Where
is
it
a
round
of
developer
experience,
but
one
of
the
things
that
the
working
group
is
doing
is
finding
out
what
are
the
admins
doing
with
k-native?
What
are
the
struggles
they
have
like
roland
said.
F
One
piece
of
feedback:
a
lot
of
folks
are
using
eks
and
trying
to
figure
out
what
is
the
best
solution
for
integration
with
observability,
like
things
they're
asking
like
how
do
I
integrate
with
datadog,
or
how
do
I
integrate
with
cloudwatch,
or
how
do
I
integrate?
How
do
I
do
monitoring
observability
tracing?
D
So
I
don't
know
for
kennedy
specifically,
but
at
trigger
match.
We've
done
a
huge
effort
to
to
look
at
all
our
components
and
expose
metrics
that
can
be
scraped
with
promoters
and-
and
we
do
a
lot
of
configuration
on
amazon
to
make
sure
that
you
know
all
the
all
the
logs
get
properly
to
cloudwatch
and
that
they're
ordered
you
know
properly,
so
that
you
can,
you
know,
look
at
your
logs
and
and
so
on.
So
yes,
there
is
a
lot
of
effort
in
observability
and
logging
yeah.
B
Oh,
I've
got
too
many
scars
here,
we're
all
we're
doing
it
all
badly.
So
again,
I'm
I'm
looking
at
matt
over
there
because
matt's
familiar
with
a
lot
of
this
stuff
too.
B
So
the
early
work
on
k-native
attempted
to
sort
of
split
the
world
using
open
census
and
the
stackdriver
libraries,
and
so
you
could
either
export
to
prometheus
through
opencensus
or
you
could
export
to
stackdriver,
and
that
code
is
still
a
bit
of
a
mess
and
if
someone
would
like
to
tilt
at
that,
I'm
happy
to
talk
with
them
afterwards,
but
yeah
observability
is
a
big
problem.
B
One
thing
that
I
feel
like
I
don't
do
well,
I
suspect
a
lot
of
us
who
are
deep
in
the
project
don't
do
well
is
trying
to
use
it
as
an
end
user
with
only
end
user
permissions.
It's
super
easy
to
debug
things
when
you
have
cluster
admin
compared
with
you
only
have
permissions
in
this
namespace
and
maybe
some
monitoring
dashboards,
and
I
think
we
could
do
better
there,
and
I
would
I
would
love
to
see
someone
tackle
that.
E
Any
question
from
the
audience
there
we
go.
A
Yeah
yeah,
actually
I
I
I
typically
distinguish
cada
and
kinetic
that
can
have
really
it's
all
about
consumption
based
scaling
based
on
http
traffic.
The
circada
is
not
so
good
at
that.
Cada
is
really
good
for
specific
scalars,
where
you
have,
as
you
mentioned,
scaling
on
kafka
messages
within
the
topic
or
other
systems,
and
it's
also
pulled
where
it
is
pushed.
Actually,
I
think
they
are
not
really
competing
against
each
other.
They
are
really
complementary.
So
actually
you
can
combine
both
of
them
very
nicely
to
provide
a
super
duper.
A
B
Oh,
I've
got
a
couple
other
things:
if,
if
cada
is
working
for
you,
you
should
keep
using
it
like
I'm
not
going
to
tell
you
go
switch
to
something
else,
because
it's
the
new
hotness
use
the
stuff
that
works
for
you.
There
is
an
interesting
sort
of
philosophical
difference
as
well
between
cada
and
k-native.
So,
where
k-native
sort
of
came
from
is
let's
improve
the
developer
experience,
oh
and
by
the
way,
auto
scaling
should
be
part
of
that
have
a
url
that
you
can
reach
to.
B
You
know,
access
things
and
so
forth
should
be
part
of
that,
but
it's
sort
of
all
those
decisions
build
on
each
other
to
kind
of
reach
a
point
and
when
you
start
trying
to
take
a
leg
away,
you
end
up
with
something
that
you
lose
more
than
one.
You
lose
more
than
one
support
when
you
try
to
pull
one
piece
out.
B
D
Yeah
relate
to
kafka,
I
don't
think
yeah.
I
agree
with
roland
and
even
on
the
under
the
scaling
and
it's
there
is
no
real
competition,
but
re.
You
know
with
respect
to
kafka,
if
you,
if
you're
thinking
about
k
native
it's
more
about
whether
you
want
to
re,
move
away
from
kafka
connect,
for
example,
which
is
all
about
defining
your
sources
and
things
into
kafka,
whether
you
want
to
do
data
transformation
differently
than
what
you're
doing
right
now
with
kafka
and
there,
with
k
native
you're
going
to
be
much
more
declarative
right.
D
So
you
know
whether
that's
a
big
benefit
for
you
or
not.
You
know
that
depends
on
what
you're
doing,
but
I
think
you
know
you.
You
have
to
compare
more
on
the
kafka
connect
side
of
things
rather
than
you
know,
scaling
or
you
know
your
use
of
cada
right.
B
That's
a
good,
that's
a
good
point
with
cada.
You
say:
hey,
I'm
looking
I'm
inter
scale
this
based
on
the
depth
of
this
queue,
but
there's
nothing
that
forces
that
to
connect
back
to
the
application
code
that
is
necessarily
reading
for
the
same
q,
so
you
could
actually
have
a
different
q
name
and
probably
get
some
really
wild
scaling
results.
M
Yeah,
I
just
wanted
to
add
a
few
comments:
the
observability
conversation
earlier
so
late
last
year.
I
think
we
redid
some
of
our
most
of
our
observable
observability
stack
to
kind
of
make
it
easier
for
any
users
to
get
started,
which
is
kind
of
my
perspective
as
a
systems
engineer
devops
person
as
opposed
to
a
software
developer.
M
M
Our
monitoring
stack,
so
we
saw
I
and
so
few
other
people
went
ahead
and,
like
created,
some
grafana
dashboards
joined
all
the
dots
together
and
clarified
the
documentation.
How,
when
you
deploy
canadian
to
a
cluster,
how
I
go
ahead
and
be
in
a
place
where
you
can
see
all
the
metrics
and
everything
I
remember.
A
Yeah,
I
I
think
we
have
still
a
gap
in
general
communication
about
new
features
to
the
community,
so
we
had,
I
think,
the
community
meetings
where
we
had
some
some
section
where
we're
talking
about
news
in
the
in
in
the
product
itself,
and
I
think-
and
I
totally
agree
that
we
need
to
to
to
get
better
as
a
community
to
to
announce
the
stuff
that
we
are
doing
and
how
we
can
use
it
more
and
I
think,
having
this
kind
of
regular
meetings
or
scenarios
where
people
can
join
and
can
get
information
about
them.
A
Then,
of
course
the
website
could
list
it,
maybe
more
prominently
yeah.
So
I
agree
that
it's
probably
there's
still
something
to
do.
B
E
And
there's
also
a
toc
right
like
meeting
where
you
can
come
and
kind
of
present
to
people.
It's
very
open.
Evan
runs
it.
You
know
usually
there's
a
space
for
like
a
new
new
topic,
yeah
any
other
questions
from
the
audience.
G
O
I'm
yours
from
arctic
finance,
since
vmware
and
and
openshift
of
tundra
and
openshift
is,
is
represented.
I'm
still
struggling
with
it
with
the
question
that,
like
keynative
works
really
well
in
the
in
the
google
cloud
environment
because
they
have
it
glued
well,
very
well
onto
their
infrastructure,
but
you
both
like
provide
like
managed
kubernetes
services,
so
you
don't
always
have
to
control
of
the
the
the
scalability
of
the
underlying
infrastructure.
I
can
perfectly
scale
up
to
a
thousand
parts,
but
I
have
three
worker
notes.
It
doesn't
work
out
so
yeah.
O
B
If,
if
you're
using
google
cooperatives
engine-
and
you
don't
check
the
cluster
auto
scaler
box,
you
also
don't
get
scaling
but
yeah,
that's
a
that's
a
limit.
Most
people
seem
to
kind
of
understand
that
we're
not
giving
you
magical
extra
capacity,
although
maybe,
if
you're
at
a
large
scale,
you
get
a
little
bit
of
over
subscription
extra
capacity.
That
you
know
was
hard
to
extract.
Otherwise,
but
I
don't
know,
most
people
seem
reasonably
comfortable
with.
Oh
I've
got.
You
know
three
machines,
they've
each
got
12
cores.
B
A
Yeah,
I
I
think,
also
that
that
for
for
products
like
like
openshift,
it's
a
there's,
an
additional
persona
that
is
not
needed
for
google
cloud
one.
This
is
the
operator
who's
really
driving
the
the
cluster
itself
and
and
of
course,
the
operator
needs
to
to
think
about
how
we
can
set
the
limits
for
canadian
like
it
has
to
do
like
she
or
he
has
to
do
for
for
any
other
workload
running
on
openshift,
for
example.
So
there's
not
there's
no
magic
silver
bullet
that
helps
you
with
the
kinetic
there
an
option,
serverless
and
yeah.
B
E
D
I
was,
I
was
just
going
to
respond
a
little
bit
differently,
which
is
that,
when
k
native
started,
there
was
one
aspect
of
it
which
was:
can
we
have
a
better
api
abstraction
for
an
application
in
kubernetes
right?
Because
you
needed
before
you
would
need
to
create
a
service,
a
deployment,
an
ingress
and
so
on.
So
we
said,
I
mean
they
said
cognitive
service
right,
one
object
which
then
you
know
is
going
to
create
everything.
D
So
that
meant
that
you
know
the
the
the
mental
shift
is
a
little
bit
different
you're,
not
talking
that
much
about
serverless
as
much
as
you
know,
you're
looking
for
a
pass
right,
you're,
looking
for
a
platform
as
a
service
right
and
a
lot
of
people
that
turn
to
canadian,
that's
what
they
wanted.
They
wanted
a
pass.
They
wanted
a
new
heroku
to
be
able
to
deploy
applications.
You
know
better
on
and
and
faster
and
easier
on
on,
kubernetes
right,
but
then
you
know
we
said
hey,
you
know
it's.
D
It's
serverless,
we're
gonna,
give
you
to
scaling
and,
and
things
like
this
and
and
you
know,
with
the
aws
person
in
the
room,
I'm
gonna
say:
serverless
on-prem,
it's
totally
nuts
right.
So
to
your
point,
you
cannot
do
service
on-prem,
because
you're
not
going
to
be
able
to
scale.
You
know
as
much
as
in
the
cloud
right.
D
E
Okay,
anybody
else
want
to
add
or
okay
any
other
questions
from
the
audience.
Don't
be
shy,
train
guard
people,
creators
of
they
were
there
for
most
of
it.
I
see
so
they
don't
have
any
questions.
So,
let's
try
to
close
with
this
question
that
I've
been
trying
to
ask
before
is
well
first,
let
me
ask
this:
how
many
people
here
are
brand
new
to
k
native
so
you're
just
here,
because
you
just
heard
about
canadia,
okay,
so
just
a
couple,
how
many
people
have
committed
to
any
of
the
projects
in
canada?
E
E
Oh
look
at
that
gene
god
love
it
and
then
how
many
are
essentially
helping
build
it
because
you're
gonna
sell
it
right,
like
you,
have
either
some
kind
of
a
product
that
you're
based
on
it:
okay,
not
too
much
okay!
So
there's
a
bit
in
the
middle
all
right!
So
my
question
now
that
you
have
an
understanding
of
who
the
audience
is.
E
You
have
a
lot
of
experience,
not
just
on
k-native,
but
also
a
lot
of
you
in
previous.
You
know
communities
so
looking
to
the
future,
how
can
you
like
what?
What
are
the
things
that
you
want
to
bring
from
your
previous
experience
to
k-native,
to
make
it
better
and
if
it's
something
that
you
find
that
you
know
on
the
other
end,
a
canadian
does
better
than
the
rest.
E
What
would
you
bring
from
k
native
to
the
rest
to
the
other
communities,
because
I
asked
this
because
I've
been
in
multiple
communities
and
I've
seen
this
particular
community
as
being
fantastic?
I
think
a
lot
of
it
has
to
do
with
the
people
that
started
it.
It's
not
because
they're
there
I
like
them,
but
I
think
they
set
the
the
march
right
on
how
things
should
be
done.
B
I
actually
took
a
little.
It
took
a
little
while
for
me
to
think
of
something,
but
there
have
been
a
couple
of
changes
recently.
You
know
feature
tracks
where
there's
a
change
proposed
to
roll
out
and
there's
some
extra
flags
to
enable
it
and
everyone's
been
really
accommodating.
When
I
pushed
back
and
said
you
know
hey,
we
should
do
this
behavior
by
default.
Everyone
should
get
this,
it
shouldn't
be.
B
B
Well,
I
and
I
think,
when
we
miss,
we
try
to
learn
from
it
and
we
fix
it,
and
I
really
like
that
in
comparison
with
you
know,
I'm
working
on
a
container
d
change
and
it
looks
like
maybe
it's
gonna
miss
two
releases,
because
it's
hard
to
move
these
files
on
disk
forward
and
yeah.
I
think
we
do
better
than
that.
E
C
Along
with
the
five
people
in
the
room
who
are
new
to
the
project,
I
think
when
I
was
first
trying
to
navigate
the
docs,
I
was
reaching
for
like
how.
How
does
this
relate
back
to
kubernetes,
and
I
was
especially
confused
by
a
k-native
of
a
service
being
called
a
service.
I
thought
it
was
a
kubernetes
service
and
I'm
like
where's
the
pod
running.
I
don't
unders
like.
C
I
got,
really
hung
up
in
that
for
a
while,
and
so
I
think
it
would
be
nice
when
you
explain
everything
relating
it
back
to
how
the
traffic
moves
through
and
where
the
application
actually
is
running.
E
D
I
I
think
you
know
the
biggest
thing
is
to
get
as
much
feedback
from
from
users.
You
know,
I
think,
we've
been
very
good
from
the
the
the
supply
side.
You
know
vendors
and
then
developer
of
the
of
the
technology,
but
we
really
need
to
bridge
the
gap
with
the
users
so
that
you
know
we
can
all
improve
the
the
user
experience
the
developer
experience
the
operational
experience
so.
A
Actually,
this
is
the
same
direction.
I
really
try
to
bring
in
my
experiences
really
to
provide
a
excellent
user
experience
with
the
cli
in
that
case
and
also
try
to
convince
people
that
the
client
is
a
good
thing.
It's
because
it's
not
so
easy
to
convince
people
that
I
really
stick
to
yaml
files,
and
you
really
love
that
and
I
really
try
to
have
some.
A
E
P
P
You
might
choose
k
native,
because
you
want
to
defer
the
choice
of
maybe
even
kubernetes,
because
there's
a
bunch
of
these
managed
services
that
run
k
native.
So
what
is
the
next
choice?
Deferral
that
we
could
think
about
trying
to
help
like
what?
What
are
people
having
to
lock
themselves
into
today?
That
that
is
in
the
purview
of
k-native.
B
I
I
I
think
scott
was
also
asking
what's
the
next
thing
like
eventing
and
I
think
the
function
stuff
is
pretty
promising.
As
a
you
know:
hey
yes,
underneath
there's
http,
but
you
know
underneath
there's
cloud
events,
but
you
just
see.
I
get
a
thing
and
it's
been
unwrapped
for
me.
I
do
my
stuff
and
then
I
hand
it
back
to
you
and
you
wrap
it
back
up
and
send
it
off,
and
I
don't.
A
What
I
can
see
as
an
extension
or
the
next
step
to
eventing
is
really
to
to
settle
on
this
idea
from
inventing
mesh
that
you
have
really
comprehensive
overview
of
everything
of
the
whole
topology
that
you
have
in
preventing.
So
like,
like
a
single
crd
that
really
lists
all
the
triggers
and
everything
else,
so
that
you
really
have
a
have
a
good
overview
without
trying
to
pick
up
everything
on
your
own
and
have
to
lift
it.
So
maybe
really
a
read-only
custom
resource
that
just
represents
the
topology
of
your
eventing
setup.
D
If
I
understood
your
question
properly,
you
know,
I
would
say,
the
api
gateway.
You
know
it's
a
it's
a
big
thing.
You
you
face
it
all
the
time
and
right
now,
there's
lots
of
work
in
in
ingress.
You
know
alpha
objects
and
so
on
that
are
very
relevant.
But
you
know
it's
a
little
bit
spread
out
and
I
think
we
need
to
clarify
this
related
to
k
native.
It's
the
fact
that
there
is
no
eventing
ingress.
E
And
I'll
give
you
a
free
answer.
So
when
I
was
in
research
2000
the
biggest
concern
that
all
the
students
were
going
after
is
this
idea
of
dynamic
composition,
so
you'd
have
a
bunch
of
services
to
do
say,
for
instance,
setting
up
the
trip
to
come
here,
reserve,
hotel
and
so
on,
and
then
how
could
you
dynamically
compose
it?
E
So
you
have
to
have
a
way
to
compose
the
services
and
then
dynamically
instantiate
it
to
specific
ones,
and
I
don't
think
we're
anybody's
doing
this
right
now
right
now
you
have
to
kind
of
code
it
you
can
use
lenses
stuff
where
you
can
build
the
functions,
but
you
still
have
to
put
it
all
together
the
glue.
Can
you
make
it
a
little
bit
more
dynamic?
B
Okay,
actually,
I
just
remembered
related
to
functions
or
something
that
I've
I've
been
pointing
a
couple
of
people
at.
If
you've
ever
used,
firebase
functions,
they
have
a
very
neat
mechanism
for
specifying
what
event
triggers
should
trigger
your
function,
all
in
the
same
source
file.
B
Okay,
it's
an
acquisition
that
google
made,
but
it's
it's
its
own
product
name.
So
if
you
just
search
for
firebase
functions,
you
can
see
some
examples
of
that.
A
One
final
thing,
which
I
nearly
forgot
is
really
that
I
also
agree
that
this
composition,
that
you
mentioned,
is
really
one
of
the
next
things
like
with
also
with
integration
of
serverless
workflow.
This
could
be
really
the
combination
with
these
kind
of
high-level
abstractions
and
bring
those
together.
I
think
this
will
make
a
big
thing.
E
Together,
the
time
might
be
right,
okay,
so
we
only
have
a
few
seconds
left
really
so
I'll
just
give
everybody
a
chance
to
say
final
word,
whether
or
not
it's
a
tip
that
you
want
to
give
people
to
kind
of
get
started,
or
you
want
to
plug
something
from
your
company
or
from
your
work.
That's
fine!
We
appreciate
your
time
so
hold
on
go
first,.
A
C
But
my
show
that
I
keep
talking
about
is
called
enlightening.
It's
on
tonzu.tv
I
have
some
stickers
so
come,
say:
hi
and
I'll,
give
you
some
stickers
and
hang
out
excellent.
D
F
And
we
return,
let
me
get
the
time
1
30
or
30
yeah
it's
here
outside.
So
we
have
lunch.