►
From YouTube: TAG Meeting - Kubecon EU 2023
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
B
D
D
Yes,
so
hi
I'm,
Addie,
based
in
London
I,
have
been
painting
around
the
tag,
particularly
the
platform
working
group
since
based
in
Detroit,
been
a
platform
engineer
for
five
or
six
years
and
now
working
on
a
open
source
framework.
That's
pretty
sandbox,
but
you
know
I
on
cncf,
so
I'm
interested
in
this
space.
D
So
with
the
white
paper,
we
did
a
lot
of
work
on
the
high
level
like
goals
and
intentions
of
platforms
and
platform
engineering.
But
if
someone
picks
that
up
it
doesn't
really
give
much
guidance
on
where
to
go
or
or
what
to
start
doing
from
there
and
I
think
there's
a
lot
of
levels
that
you
can
go
to
down
to
what
projects
might
choose
within
the
landscape
and
so
I
think,
there's
probably
a
lot
of
space
to
play
in
in
there.
D
So
some
higher
level,
but
more
tactical
than
than
the
white
paper
and
some
very,
very
low
level.
Tactical
work.
I've
got
a
proposal
I've
put
in
an
issue
yesterday
for
a
platform
maturity
model.
That
would
be
one
level
kind
of
more
detailed
and
more
tactical
than
the
white
paper,
but
not
down
to
a
level
of
take.
These
actions
hear
the
blueprint
for
how
to
do
a
platform.
D
Yeah
and
I
just
want
to
introduce
it
I'm,
hoping
that
we
can
go
through
a
similar
process
of
eating
it
up
figuring
out.
If
this
is
even
the
right
shaped
thing
to
release
from
the
working
group
and
then
getting
it,
yeah
wordsmithing
it
to
the
point
where
we're
happy
with
it.
It
is
the
right
shape
thing
so
really
quickly.
Why?
What
and
how
so
I
think
the
reason
why
we
started
writing
this
myself
and
some
colleagues
was
that
we're
in
a
hype
cycle.
There's
no
lot,
there's
no
denying
that
platform.
D
D
I
think
is,
I
can
look
at
like
a
hotel
where,
all
of
a
sudden,
all
the
vendors
came
together
and
used
the
same
language
and
were
being
evaluated
against
the
same
set
of
principles
and
intentions,
and
so
the
earlier
we
can
set
some
of
those
up
so
that
we're
vendor
agnostic
or
at
least
vendors,
shared
ownership,
I
think
that'd
be
really
good,
and
also
because
everyone
on
every
platform
engineering
talk,
puts
the
same
definition
of
what
a
platform
is
from
The
Martin
Fowler
hosted
edinbacher
quote,
but
when
you
actually
listen
to
what
they
described,
it
often
differs
so
getting
some
arms
around.
D
That
would
be
really
nice.
So
that
was
the
reasoning
for
trying
to
throw
out
a
straw
man
to
start
that
conversation.
D
So
what
we,
what
we
did
as
a
small
group
was,
we
wanted
to
have
a
way
to
compare
behaviors
that
we
see
as
benefiting
and
detracting
from
success
in
platform
engineering,
so
it
can
stay
away
from
hype
and
hopefully
get
towards
actual
outcomes,
and
we
want
to
take
one
more
step
from
the
white
paper
on
kind
of
tangible
practical
advice.
Without
going
to
the
point
of
you.
D
Don't
think
we're
mature
enough
yet
as
a
process
to
get
there,
so
we
identified
some
contributing
factors.
Associated
platform
engineering
tried
to
leverage
known
formats,
which
is
maturity,
model
we'll
speak
about
the
problems
with
that
and
get
some
of
the
ideas
challenged
really
early,
so
we
have
like
a
15
or
20
people
have
reviewed
it
already,
just
to
make
sure
that
we're
not
the
strawman's
at
least
worth
people's
time
here,
rather
than
wasting
that
yeah
and
then
so.
The
goal
is
to
get
this
offer
this
to
the
working
group.
D
If
it's
something
that
the
working
group
sees
value
and
it
wants
to
wants
to
move
to
publish
so
as
I
said,
maturity
models,
I'm
sure
there's
at
least
some
girl
that
I
appreciate
they
were
all
internal
when
I
said
maturity
model,
it's
a
thing
that
is,
can
be
quite
dangerous,
so
one
of
the
sentences
in
the
document
as
it
currently
stands
is
that
this
is
not
meant
to
be
a
ruler.
This
is
not
meant
to
be
something
that
you
can
compare.
D
Your
teams
decide
where
to
cut
headcount,
decide
who's,
doing
the
right
job
or
not.
Can't
stop
Consultants
from
being
Consultants.
I
was
one
for
seven
years:
I'm,
sorry,
I'm
still
atoning
my
sins,
but
like
we
can
at
least
try
and
ourselves
produce
something
that
is
high
value.
So
if
people
do
choose
to
use
it
in
the
wrong
way,
there's
at
least
like
we're
confident
with
the
information
the
background,
so
the
I
want
to
start
with
what
we
were
trying
to
answer
so
I'll
uncover
the
whole
grid
that
we've
come
up
with.
D
There's
gonna
be
a
whole
lot
of
improvements
we
made
to
it
just
to
be
clear,
but
I
want
to
start
by
saying
you're
taught
we
were
thinking
about
what
matters
when
it
comes
to
platforms
and
platform
engineering,
and
we
identified
these
kind
of
areas
that
you
may
have
different
behaviors
underneath.
So
how
does
your
company
choose
to
fund
your
platform
efforts?
How
do
you
get
users
to
engage
with
your
platform?
How
do
you,
what
is
their
actual
interaction
methods?
D
So
how
do
you
onboard
them
and
make
them
interested
in
using
it,
and
then
what
is
their
interaction
when
starting
to
use
it?
How
do
you
manage
like
incoming
requests
and
new
features?
How
do
you
manage
the
like,
non-differentiating
things
so
abilities
so
security
and
performance,
all
those
kinds
of
things
shared
across
the
whole
organization
and
yeah?
So
from
these
questions,
boil
everything
down
to
a
word:
people
love
people,
take
a
screenshot
of
words.
D
D
There
are
lots
of
questions
about.
Does
everything
need
four
categories
like
less?
In
some
cases,
are
they
the
right
categories?
All
those
things
that
I've
put
into
the
issue
as
big
open
questions
from
the
early
reviewers?
That
I
didn't
think
should
be
solved
by
a
small
group
of
authors.
I
thought
could
be
solved
by
a
larger
or
cross-functional
team
in
the
working
group,
but
yeah
so
and
one
thing
about
the
one
thing
about
this:
is
it's
not
meant
to
be
that
you're
either
level
one
two
three
or
four?
D
It's
like
you
may
be
level
one
in
some
areas,
four
and
others.
You
may
find
that
improving
in
one
unlocks
opportunities
in
another,
it's
sort
of
kind
of
figure.
Adventure
was
the
the
format
of
majority
model.
We
went
with
so
yeah,
so
that
was
what
we've
come
up
with
within
the
document
there's
also,
and
it
depends
which
allows
us
to
go
into
a
little
bit
more
detail
about
stuff.
So
this
is
an
example
of
the
detail
behind
funding.
So
it's
not
just
that.
D
The
one
word,
though,
it's
helpful
to
boil
it
down
to
try
to
give
some
examples
that
people
can
use
and
yeah
there's
an
issue
open
in
381
and
I'd
love
to
hear
as
I
say,
one
of
the
first
questions
you
can
see.
The
second
bullet
point
there
is:
is
there
a
better
shape
for
this
type
of
guidance,
like
even
the
term
maturity
model?
D
I
think
has
enough
issues
that,
depending
on
the
way,
this
group
approaches
it,
we
may
end
up
with
a
capability
Model,
A
fluency
model
or
something
completely
different,
not
a
model
at
all.
Yeah
and
I'd
love
to
just
start
the
conversation
about
that.
What
does
that
next
level
of
detail
below
the
white
paper
look
like
to
start
giving
people
a
chance
to
say,
okay,
I
understand
the
paper.
I
want
these
things.
Where
am
I?
How
mature
am
I
can
I
get
there?
What
do
I
need
to
be
focusing
on
yeah.
B
F
D
It's
currently
in
a
Google
doc,
purely
based
on
my
experience
that
that
was
an
effective
way
to
collaborate
on
the
white
paper
early
days.
When
there's
like
a
lot
of
moving
funds.
What
we
did
with
the
white
paper
is:
it's
sorry,
Google
doc.
D
It
gave
space
for
lots
of
comments
and
people
to
like
interact
about
very
small
pieces
or
entire
paragraphs
at
the
same
time,
and
then
when,
when
it
started
to
settle
down
and
become
a
bit
more
stable,
we
went
into
GitHub
issues
or
like
into
a
commit
into
the
repository
and
use
pull
requests.
D
So
that
was
my
theory
as
to
what
might
be
useful,
so
the
Google
Doc,
it's
globally,
commentable
Slash,
suggestible
and
I
expect
I'm,
hoping
I've
asked
in
the
issue
for
people
who
want
to
help
kind
of
manage
this
process
and
and
those
people
get
edit
rights
and
and
be
able
to
manage.
However,
these
you
fit
the
like
contributions
to
it.
So
yeah.
G
What
happens?
Yeah
the
same
thing
happened
with
the
boss
working
grid
and
started
with
started
a
separate
repo,
but
then
ultimately
realized
that
there
was
a
benefit
of
having
a
place
for
long-lived,
not
just
documents,
but
things
that
would
then
be
version.
Then
additional
information
in
a
pro,
so
project
format
really
made
sense
there.
G
So
we
ended
up
just
splitting.
The
git
commits
like
just
bringing
in
all
the
gate
history
so
that
this
authors
don't
be
competitive
for
what
they
did
and
made
a
PR
against
the
Tag,
app
delivery.
So
the
work
all
the
working
group
stuff
was
there
and
then
all
the
other
things
that
made
more
sense
for
the
project.
We
did
the
same
thing
in
the
second
archive
yeah.
A
H
So
hey,
my
name
is
Colin
Griffin
I
run
a
software
development
group
that
focuses
on
cloud
native
enablement.
What
the
hell
does
that
mean?
We
we
essentially
we
build
applications
for
folks
from
the
ground
up,
so
that
they
can
leverage
that
as
their
new
standards
of
best
practice,
it's
almost
their
Center
of
Excellence,
a
practical
tool
that
we
can
actually
deliver
a
an
out
of
the
box
complete
solution
for
how
their
Cloud
native
applications
should
run
function
should
they
run
it
on
a
platform
that
kind
of
thing.
H
So
we
approach
platforms
and
we
approach
this
from
an
application
developer
point
of
view,
so
we're
trying
to
enable
app
Dev
and
teach
them
the
tools
that
they
need
for
itos,
rather
than
going
to
the
itops
guys
and
teaching
them
the
tools
that
the
developers
need
so
obligatory
chat,
GPT
quote
so
ours.
The
talk
that
that
or
what?
What
we're
talking
about
today,
really
is.
It's
just
kind
of
an
open
discussion
is
to
bring
visibility
to
a
challenge
that
open
source
and
even
like
internal
project.
H
Maintainers
have
today,
particularly
now
around
Cloud
native,
because
Cloud
native
is
so
complicated
and
if
we
choose
to
deliver
Solutions
complicated
system
that
has
new
skill
sets
that
many
Developers
for
the
most
part
developers
don't
have.
These
skill
sets
yeah.
What
are
the
kind
of
challenges
that
were
expected
to
take
on?
What
are
the
things
that
we're
expecting
to
provide
to
those
maintainers
and.
H
Really,
there's
a
lack
of
tooling
there's
a
lack
of
guidance
at
the
ecosystem,
for
how
should
I
support
my
end
users
and
my
adopters
for
this
process.
So
burden
of
responsibility
refers
to
the
weight
of
pressure.
A
person
may
feel
when
they're
entrusted
with
important
duties
or
obligations.
In
particular,
this
feeling
often
arises
from
an
awareness
of
the
potential
consequences
of
our
actions
or
decisions
and
the
expectations
and
Trust
placed
on
us
by
others.
H
So
as
we
publish
projects
and
put
things
out
there
in
the
wild,
what
have
I
just
done
to
myself
and
what
am
I
now
expected
to
to
do
for
these
people
so
when
I
release
an
open
source
project
or
when
I
bring
something
into
my
organization
or
I
bring
a
new
developer
on
board.
H
These
are
things
that
I
expect
to
happen.
Right
like
this
is
exciting.
I
can't
wait
to
add
more
features.
My
readme
is
going
to
have
everything
that
everybody
needs
for
whatever,
no
matter.
What
they're
going
to
do.
People
are
going
to
comment
on
my
app
and
say
this
was
amazing.
I
got
it
running
it's
incredible
right.
These
are
the
things
that
we
typically
expect
that
we're
going
to
have,
but
what
actually
happens
right?
H
How
do
I
run
this
thing
on
Windows
or
whatever
other
obscure
Arn,
based
thing
that
I'm
going
to
run
on
what
is
Kate's?
What
does
this
thing
even
mean
right,
I
can't
believe
you
chose
this
tool
right
when
you
get
this
like
pool
of
negativity
that
just
kind
of
comes
out
from
nowhere
and
it
sucks
the
life
out
of
our
project
and.
H
So
I
kind
of
want
to
do
this
around
a
little
bit
of
a
case
study.
So
we
got
involved
with
the
project
called
gen3.
It's
an
open
source
of
data,
Commons
platform,
so
they're
using
the
word
platform
right
and
Healthcare
today,
at
least
in
the
U.S
there's
a
mandate
with
most
of
the
publicly
funded
projects
so
from
like
the
National
Institute
of
Health
and
National
Cancer
Institute
they're
spending
billions
of
dollars
and
giving
them
research
institutions.
But
there
was
a
big
problem
where
this
research
wasn't
being
made
available
publicly.
H
So
people
would
get
all
this
money
and
then
the
research
would
go
poof
and
disappear
forever.
That
sounds
like
IEP
grants
and
things
as
well
right
and
that's
why
open
source
is
so
important,
because
now
at
least
somebody
got
a
good
from
that
stuff.
But
this
platform
Gen
3,
is
from
the
University
of
Chicago
and
it's
an
open
source
platform
for
this
concept
of
data
common
so
enabling
other
research
institutions
to
consume
and
download
data.
H
The
challenge
that
they
faced
is:
yes,
they
built
a
great
open
source
solution
and,
yes,
they
put
it
on
kubernetes.
And
yes,
it's
microservices,
some
everything
else,
that's
buzzwordy
today,
but
the
problem
was,
they
wrote
their
own
custom
operator.
They
load
their
bash
scripts
right.
They
they
built
it
for
eks
right,
it's.
They
have
terraform
scripts,
but
it's
for
eks,
and
so
so
how
did
this
manifest?
H
So
again?
A
research
research
grants
mandated
data
comments
and
this
is
kind
of
the
flow
a
doctors
are
seeking
in
Solutions
they're
seeking
projects
and
then
they're
working
backwards.
H
So
they
select
gen
3
to
say
I'm,
going
to
run
a
data
comment,
and
then
they
say:
oh,
how
do
I
run
this
that
comes
later,
then
they
suddenly
discover
the
kubernetes
thing
and
then
they
try
to
go
down
their
kubernetes
Journey
just
to
get
this
platform
to
run,
and
then
they
go
back
to
the
maintainers
and
say
how
do
I
run
kubernetes,
and
that
is
not
at
all
what
the
project
is
about
right.
H
I
H
Now
I
have
to
support
X
Y,
it's
terrible,
so
this
is
a
crappy
little
diagram
that
we
use,
but
we
expect
too
much
of
our
users
and
we
expect
that
they
go
down
a
different
maturity
pathway,
we're
thinking.
Oh
they've
got
mature
I.T
teams
they're
in
a
fake
organization,
they're
going
to
think
about
the
infrastructure.
First
then
they're
going
to
figure
out
a
way
to
manage
the
infrastructure.
Then
they're
going
to
select
an
application
that
fits
their
infrastructure
and
then
that's
going
to
meet
their
business
goals.
H
H
So
the
challenges
for
these
gen
3,
guys
that
you
watch
that
are
going
through
is
they've
got
a
massive
amount
of
community
growth
out
of
nowhere.
But
people
are
asking
the
wrong
questions
so
they're
already
trying
to
solve
or
how
do
I
manage
an
open
source
Community.
We
all
know
that
that's
in
pain,
they
lack
skilled
resources,
so
the
University
of
Chicago
actually
has
one
or
two
people
that
understand
somewhat
kubernetes.
They
don't
have
mature
platform
things.
H
They
don't
have
the
guidance
they're
managing
expectations
for
opinionated
adopters,
so
everybody
has
their
own
way
of
running
things
and
they're
going
to
have
an
argue
on
this
stuff
and
what's
happening,
is
that's
pulling
attention
away
from
their
core
application?
They
should
be
thinking
about
how
do
I
help
somebody
data
comments,
not
thinking
about
how
should
I
run
and
I
hate
harp
on
kubernetes,
but
we
all
like
know
that
kubernetes
maturity
is
a
challenging
thing.
So
what.
H
Native
apps
are
hard
right.
People
are
still
figuring
this
stuff
out.
Demand
is
exceeding
Supply,
so
demand
for
information
and
help
is
exceeding
the
ability
to
actually
support
these
people.
Kate's
expertise,
isn't
there
yet
right.
It's
still
we're
a
small
group
in
a
room
but
I
think
I,
don't
know
how
many
nines
of
percentage
that
we
have
where
people
don't
have
any
idea.
H
What
kubernetes
is,
let
alone
understand
how
to
run
them
on
their
own
machine
or
the
concept
of
a
load
balancer,
and
we
have
a
tremendous
amount
of
tools
to
choose
from
and
it's
causing
Deadlocks
everybody's
afraid
to
commit
to
a
certain
tool.
There's
no
like
and
all
be-all,
there's
no
right
tool,
and
that's
the
beautiful
thing
about
this,
but
it's
also
creating
huge
challenges.
H
So,
what's
being
done,
I
kind
of
left
that
blank
intentionally,
because
we're
we're
kind
of
trying
to
discover
and
figure
out
like
how
to
do
more,
but
just
to
talk
about
what
this
means
for
app
delivery.
Is
that
we're
unintentionally
taking
on
this
burden
we
need
to?
We
do
still
need
to
find
ways
to
support
those
users.
We
can't
just
say
it's
not
my
job.
H
The
reality
is
users
are
finding
us
at
our
projects
and
then
need
to
be
routed
back
to
a
place
that
they
can
go,
get
the
information
and
serve
the
information
where
it's
kind
of
relevant,
as
we
were
talking
about
the
artifacts
right,
if,
if
I
had
a
way
to
guide
that
person
through
my
project
back
to
platforms
right
back
to
the
white
paper
back
to
app
delivery
and
or
a
way
to
say,
this
is
how
we
did
this
thing,
then
that
will
provide
a
lot
of
context
to
the
users
like
the
what
the.
H
Why
the
how
and
also
provide
some
strength
and
a
foundation
that
we're
referring
back
to
that
my
project
might
not
have.
So
we
must
find
creative
ways
to
help
users
or
find
Partners
or
somebody
that
we
can
hire
or
if
we
even
have
the
funding
available,
and
we
work
with
a
lot
of
small
and
medium-sized
business.
They
don't
have
the
money,
that's
just
the
reality
of
it.
H
We
do
need
to
provide
guidance,
or
at
least
redirect
the
doctors.
We
need
to
be
clear
and
honest
with
them,
even
at
the
risk
of
Abandonment.
So
if
I
tell
somebody
I,
don't
know
enough
about
kubernetes,
but
I
built
this
thing
on
kubernetes,
which
is
a
very
real
thing.
They
might
just
abandon
my
project
and
go
away,
and
so
that's
scary,
but
this
does
create
an
opportunity,
at
least
for
groups
like
these.
It
creates
a
funnel
for
consumers
and
a
contributors
that
we
can
leverage
to
redirect
people
back
to
these
groups.
H
So
a
couple
quick
thoughts
on
that,
and
then
we're
essentially
done.
How
do
we
protect
ourselves
so
project
people
that
are
Building
Solutions
on
kubernetes
and
are
with
a
cloud
native
approach?
They
need
to
embrace
the
concepts
that
maybe
kubernetes
effective
at
its
or
right.
We
need
to
embrace
the
platform.
We
need
to
provide
tools
for
interoperability.
We
need
to
make
sure
our
products
are
interoperable.
We
need
to
make
sure
that
they
run
on
the
core
platform
and
don't
require
a
bunch
of
customization.
H
One
thing
that
I'd
I'd
love
to
see
more
and
I.
Don't
know
how
to
make
this
happen,
but
be
a
derivative
now
if
we
can
find
ways
to
enable
other
communities
that
are
coming
up
with
these
Concepts
to
just
add
their
little
piece
of
clarification.
For
what
does
a?
What
does
a
platform
mean
in
fintech?
What
does
a
platform
mean
in
healthcare?
What
does
a
platform
mean
for
a
small
company?
What
does
it
mean
for
internal
Dev
teams?
Those
can
provide
really
really
valuable
clarification.
H
Help
people
discover
something
that's
validated
to
their
use
case,
but
then
also
can
direct
them
back
up
to
the
court
so
that
we
don't
have
to
do
as
much
work
back
up
at
the
Upstream
to
say.
Well,
your
definition
doesn't
really
fit
what
a
platform
is,
but
we
know
in
fintech
it
might
not
fit
our
core
broad
definition
right.
There
might
be
clarification,
so
be
a
derivative
identify
Community
initiatives
that
you
can
redirect
people
back
to.
H
If
you
can
identify
them,
clarify
referenceable
standards
provide
artifacts
and
references
to
use,
artifacts
and
references
other
people
have
produced
when
we
do
big
research
papers
right,
we
need
sources
and
we
direct
people
back
to
the
research.
Why
are
our
projects
not
doing
this
work
with
organizations
like
cncf,
small
plug
and
funnel
the
community
back
to
Upstream
initiative?
So
don't
try
to
do
everything
or
so
on
and
so
talking
about
this
today.
Some
of
the
needs
that
we
have
here
coming
from
a
project,
maintainer
standpoint
is
I.
H
We
need
an
easier
way
to
find
or
Identify
some
of
these
open
initiatives
like.
Where
do
we
actually
go
to
get
help
on
these
projects
to
get
guidance
to
redirect
users
back
to?
How
do
we
do
that?
How
do
we
report
how
we're
meeting
these
standards
I
don't
know
I'm,
not
sure
we
need
consumable
or
referenceable
artifacts,
so
the
white
papers
are
a
great
place
to
start.
I
would
love
to
have
a
way
to
even
going
back
to
the
maturity.
I
would
love
to
have
a
way
to
say
this.
H
Data
Commons
platform
meets
the
maturity
of
a
platform
in
these
ways,
and
maybe
it's
immature
in
these
other
ways,
and
that's
a
great
way
to
redirect
somebody
back
and
get
exposure
from
my
for
my
project.
Right
do
a
case
setting
and
then
you
know
we
we
need
guidance
on
platforms
or
whatever
other
initiative
for
those
that
are
in
communities
or
individual
or
have
individual
project
initiatives.
So
it's
not
enough
for
many
projects
to
say
you
could
just
donate
this
to
the
cncf.
H
C
H
Just
kind
of
wanted
to
highlight
some
of
those
things
from
like
an
adapter
point
of
view,
it's
great
that
we're
defining
all
these
standards,
but
there's
still
a
massive
amount
of
turmoil
and
shifting
and
changing
and
it's
all
of
our
communities
can
do
a
better
job
of
working
together
through
maybe
doing
some
of
the
derivatives
work.
But
even
if
maintainer
level,
it's
really
really
really
difficult
to
discover
where
to
go.
To
get
some
of
this
information,
like
the
the
things
that
we
bring
in
the
white
paper.
A
H
Absolutely
and
projects
like
those
they're
like
University
of
Chicago,
was
like
how
can
we
present
at
kubecon,
because
we're
in
Chicago
and
that's
a
I
mean
it's
a
great
use
case-
they're
willing
to
collaborate.
H
Stay
exposure,
so
we
just
need
to
create
a
channel
for
that,
but
I
think
it's
an
important
problem
that
maybe
we
look
past
because
the
majority
of
us
work
in
large
organizations,
but
there's
a
lot
happening
at
the
smaller
bigger
this
out
level.
I
I
think
also
like
connecting
the
dots
like,
for
example,
if
suddenly
you're
seeing
kubernetes
it's
like.
Okay,
like
how
do
you
connect
that
back
to
a
general
platform
story
and
kind
of
you
know,
work
backwards
to
you
know,
solve
kind
of
the
business
outcomes
and
things
maybe
providing
and
I
don't
know
if
we
can
do
this
as
part
of
the
maturity
thing
or
whatever
but
kind
of
providing
a
framework
on
how
to
think
about
these
things
and
I
think
the
industry
use
phase.
One
is
really
good.
I
Is
the
new
kind
of
talk
about
specific
business
problems
as
well
right,
okay,
yeah.
H
G
H
L
L
C
I
think
I
think
that
I
think
there's
one
of
the
things
when
we
started
working
with
was
we
think
where
we
wanted
to
go.
We
just
started
at
the
point
that
more
or
less
because
you
need
a
definition
of
weapon
is
for
me
to
start
building
all
that
and
I.
B
C
F
D
Is
everyone
has
a
platform
today,
it's
like
when
people
say
like:
oh,
we
should
start
testing
and
productive
like
you
already
are
new
results
and
you
may
not
be
doing
it
effectively,
like
everyone
has
something
that
does
platforms
for
them,
whether
it
be
a
SAS
product,
they're
using
or
something
bash
scripts
internally,
so
making
sure
that
whatever
we
talk
about
for
building
a
platform
team
takes
up
into
account
that
it's
going
to
be
an
evolution
within
the
organization
and
like
how
do
you
absorb
what
is
currently
happening
and
make
that
assessment?
C
Questions
I
think
I
mentioned
several
times
to
be
at
the
the
sort
of
like
the
I
support
us
on
the
definition
of
of
the
infrastructure.
That's
called
like
that.
Wasn't
material
to
that
and
start
off
by
writing
on
YouTube
in
Minnesota,
probably
having
the
same
files
forward
on
the
Android
Department
I
received
from
it
was
interested
I
think
the
same
could
be
done
and
when
it
comes
back
platform
to
do
is
not
the
structure
and
complicated.
A
G
And
these
items
that
I
mean
one
I,
keep
thinking
well,
what
is
actually
missing
but
like
don't,
we
have
Excellence,
even
though
but
I
think.
If
what
I'm
hearing
is
oh,
it's
just
not
clear
where
those
things
are
or
can't
find
them,
I'm
wondering
if
those
already
are
or
could
be
as
intended
work,
products
or
outcomes
from
the
platforms
we're
at
the
group
like.
G
You
know
I
mean
like
that.
We
have
these
structures
for
a
reason.
You
know
it's
like
like
hearing
people
talk
about
like
having
cncf
projects
that
we're
on
working
groups
under
them,
and
you
know
just
like
Anonymous.
That's
like
that's
what
I
understand
about
the
app
Liberty
tag
is
for
is
to
try
to
kind
of
like
build
on
some
of
the
learnings
from
the
kubernetes
six.
G
You
know
and
like
say,
okay,
we'll
have
something:
that's
not
just
about
that,
but
a
little
bit
more
abstract
and
right
and
then
underneath
that
it
makes
sense
that
platforms,
work,
improve
operator,
get
off
workhouse.
Others,
you
know,
would
have
these
areas.
Maybe
there's
something.
Excuse
me
some
alignment,
but
to
me
it
makes
I.
Don't
know
like
what
I
kept
thinking
is.
Oh
well.
Maybe
you
could
give
some
feedback
on
like
some
open
some
issues,
maybe
in
the
and
I
the
accessibility
of
things
and
yeah.
G
H
Far
as
that
so
like
when
you,
when
you
guys
like
when
you
start
a
fresh
project
right,
if
you
don't
have
something
to
reference
for
a
pattern
or
something
you're
going
to
operate
this
kind
of
and
avoid
right
with
no
direction
or
anything
there
and
part
of
the
value
that
we
talked
about
with
having
internal
platform,
it's
been
internal
platform.
Teams
is
a
place
to
go
to
have
that
reference
and
how
things
should
be
structured.
What
tools
should
I
use?
How
do
I
provide
artifacts?
H
What
are
the
requirements
of
being
around
and
that's
a
great
way
to
start
is
just
like.
So
how
should
we
structure
these
communication
Pathways?
How
to
treat?
Let's
read
our
own
projects
in
the
cncf
the
same
way,
let's
build
a
platform
framework,
even
if
the
platform
is
just
starting
with
how
to
move
and
what
are
primary
part
of
that.
H
F
E
G
E
H
H
B
H
A
H
I
know
my
project
I
know
my
domain
I
know
my
expertise
and
Specialty,
and
so
I
can
meet
you
guys
at
the
I'm.
You
know
role
playing
right
but
like
I,
can
meet
the
group
at
the
middle
and
be
able
to
provide
my
perspective
from
my
expertise
and
then
the
the
working
group
can
provide
the
platform
level
expertise
back
right
and
use
that
as
attraction
I
like
that.
A
All
right:
well,
let's,
let's
go
on
to
the
next-
is
a
good
segue
into
Mauricio's,
yeah
yeah.
K
A
K
Interesting
conversations
I
think
I'm
coming
from
a
completely
different
angle,
here,
I'm
more
like
a
and
I,
don't
want
to
offend
anyone
I'm
more
like
a
graphical
person.
I
want
to
build
stuff
and
for
me
the
only
way
to
keep
the
discussions
going
in
a
practical
sense
and
to
actually
get
stuff
done,
concrete
stuff
done.
It's
like
you
need
to
have
some
prototypes
and
in
order
to
get
some
prototypes,
the
question
is
like
okay.
K
If
this
goal
is
this,
like
you
know
some
configuration
fast
or
whatever
I,
don't
think
that
that's
not
relevant,
but
what
is
relevant
is
like,
like
the
audience
for
this
protocols
like
who
who
are
this
products?
Who
are
the
consumers
of
these
platform
prototypes
and
how
are
they
going
to
interact
with
these
things?
And
then,
as
a
result
of
that,
we
can
actually
evaluate
okay,
which
projects
are
being
used,
which
products
are
useful,
which
products
are
having
some
issues.
K
K
This
is
not
a
sales
page,
but
I
wanted
to
share
this
because
I'm
writing
this
book,
just
because
I
collaborated
with
nine
different,
open
source
projects,
around
platforms
and
basically
they
needed
to
change
stuff
in
their
projects
in
actually
in
order
to
get
this
working
together
by
doing
that,
I
actually
I'm
trying
to
talk
less
about
the
projects
now
in
the
book
and
I'm
trying
to
talk
more
about
like
patterns
and
stuff
and
behaviors.
K
So
the
title
of
this
presentation
is
the
building
platform
prototypes,
but
I
would
say
that
it's
like
also
identifying
platform
capabilities
and
and
then
linking
capabilities
to
projects
and
then
going
into
those
kind
of
discussions.
So
we
know
we
have
that
stuff.
We
need
to
make
some
some
high
decisions,
if
you,
if
you
are
not
doing
you,
know,
platform
engineering
nowadays,
you
are
kind
of
like
doing
this.
It's
like
you
have
clusters
with
stuff
installed
in
it
and
you
actually
don't
know
what
to
do.
K
K
Those
are
all
my
clusters.
It
looks
like
that,
so
we
need
we
need.
Definitely
we
need
some
some
platform
examples
and
what
I
did
for
rejects.
Presentation
like
this
yesterday
is
that
I
I
had
a
very
good
friend
who
that's
a
lot
of
mlops,
like
machine
learning,
stuff
and
I'm
coming
more
from
a
developer
perspective.
Right,
like
I,
want
clusters
with
some
tools
in
there.
So
I
can
develop
Cloud
native
applications
right.
He
has
a
completely
different
kind
of
like
use
cases.
K
He
has
like
machine
like
a
data
scientists
trying
to
basically
run
some
learning
algorithms
and
they
actually
need
glasses
because
they
want
to
parallelize
work
but
they're
using
Python,
and
they
are
used
to
run
stuff
in
their
computers
because
in
a
different
use
case,
but
at
the
end
of
the
day
they
are,
they
want
kubernetes
to
scale
and
at
the
same
time
they
don't
want
to
learn
kubernetes.
So
that
made
me
think
a
lot
about
like
okay.
What
are
the
common
traits
and
capabilities
that
we
need
to
share
between
these
things?
K
Right,
like
we
have
developers,
building
web
applications
and
back-end
services
and
data
scientists,
training
models
and
then
deploying
models,
and
in
fact
there
are
like
tons
of
similarities,
I
work
for
a
project,
that's
called
K
native,
it's
part
of
the
cncf,
and
it's
one
of
the
projects
that
data
scientists,
like
only
a
machine
learning
space
on
kubernetes.
Are
you
seeing
a
lot
for
serving
their
models?
So
you
know
there
is.
There
is
definitely
Common
Ground
there
on
how
it
will
work.
K
So
again,
we
need
to
we
build
kind
of
like
this
example
of
a
small
platform
that
it's
targeting
both
like
machine
machine
learning,
people,
data
scientists
and
also
developers,
try
to
reuse
some
of
the
tools
and
try
to
try
to
find
common
ground
which
tools
are
doing
what
things
and
how
do
we
configure
all
these
things
to
work
together?
K
What
kind
of
problems
do
we
find
and
then
just
going
back
to
projects
and
report
about
hey,
you
know
you
are
like
I
can
clearly
see
that
you
are
coming
from
a
place
where
kubernetes
is
not
your
main
thing.
So
when
we
are
deploying
your
projects
into
our
clusters,
they
are
doing
kind
of
like
great
systems,
and
then
you
identify
that,
like
as
a
kubernetes
person,
then
you
discover
report
and
improve
improve
these
products.
So
we
created
a
tutorial
where
we
installed
all
these
things
into
a
single
kubernetes.
K
Cluster
I
was
installing
that
in
my
kind
cluster
and
I
needed
like
four
minutes
of
RAM
and
like
1060
users,
no,
no,
this
is
not
going
to
run
anywhere
and
again.
We
we
try
to
recognize
different
capabilities
and
different
things
that
we
can
be
doing
in
order
to
show
people
how
to
use
these
tools
but
for
very
concrete
use.
Cases
which
are
more
like
that,
like
self-service
things,
so
I
need
a
new
environment
for
data
scientists
or
a
new
environment
for
developers
and
the
mechanisms
for
provisioning.
K
These
environments
and
the
tools
that
we
were
using
are
in
some
way
opinionated,
because
we
are
just
using
concrete
tools,
but
at
the
end
of
the
day,
if
we
are
doing
githubs,
it
doesn't
really
matter
what
group
we
use
or
if
you
are
doing
Auto
scaling,
you
can
use
different
things
as
well.
It
doesn't
really
matter
so.
The
first
thing
that
I
showed
yesterday.
It
was
something
pretty
simple
right
like
just
to
get
this
demo
working.
K
There
are
like
a
bunch
of
tools
that
I've
been
I've
been
showing
like,
for
example,
b-cluster
and
cross
playing,
and
how
do
you
make
things
a
little
bit
more
to
work
together
when
you
are
using
these
tools
for
being
a
little
bit
more
cost
efficient?
Because
that's
the
other
thing
it's
like
kubernetes
is
expensive.
So
when
you
are
adopting
these
tools,
you
need
to
try
to
make
things
to
work
a
little
bit
better
and
a
little
bit
more
like
an
efficient
way.
K
So
from
that
experience
of
like
just
installing
all
these
tools
together
and
just
getting
them
working
in
the
same
cluster,
you
know
we.
We
have
learned
a
lot.
My
friend
also
learned
a
lot
from
the
from
the
data
scientist
side
of
things,
and
we
try
to
come
up
with
again
like
the
tutorial
just
for
you
to
follow
the
steps
and
get
this
up
and
running.
That
can
be
automated.
That
can
be
shared
with
some
people
for
review
and
saying,
okay,
you
know
I
want
to
use
the
different
tool.
K
What
kind
of
changes
do
you
want
to
invade
in
there
and
also
to
identify
the
capabilities
here?
Do
you
want
to
have
do
you
want
to
use?
It
also
is
gift
of
the
capability
of
a
platform
or
not?
Do
you
want
to
install
developer
oriented
tool
into
your
kubernetes
cluster,
so
developers
can
be
a
little
bit
more
efficient
and
focus
on
things
that
they
should
do
instead
of
learning
kubernetes?
K
Do
you
want
to
expose
kubernetes,
or
do
you
want
to
abstract
it
I
think
that
these
are
all
the
questions
that
we
need
to
try
to
start
answering
in
a
way
that
we
are
not
talking
about
concrete
projects,
but
we
can
actually
say
well
look.
This
is
sound
accessible
and
it's
you
know
a
project
that
is
solving
this
problem.
K
This
is
one
or
multiple
projects
solving
this
problem,
so
we
wanted
to
go
into
this
kind
of
discussion
so
enabling
teams
to
different
things
like
data
scientists
or
developers
through
different
stuff
and
one
of
the
things
that
for
developers
coming
from
a
developer
point
of
view,
I
want
to
enable
developers
to
experience
right
if
anything,
kubernetes
I
want
to
really
leverage
you
know
being
able
to
release
multiple
versions
of
my
services.
K
At
the
same
time,
doing
Advanced
profit
splitting
coming
feature
flags
all
these
things,
because
if
I
do
not
install
this
in
my
Google
newscasters,
then
I
do
not
have
those
capabilities
to
try
like
I,
do
not
have
feature
flying
and
that's
something
that
if
I
don't
have
as
a
developer,
then
I
cannot
understand.
I
need
to
create
a
release
and
then
my
release
will
go
out
and
it
needs
to
work,
because
if
it
doesn't,
then
I
mean
in
trolls.
So
we
need
to
kind
of
like
start
defining.
Okay,
we
want
to
enable
experimentation
or
not.
K
K
And
then
the
thing
that
I
was
mentioning
before
with
Leanne's
projects
from
from
the
cluster
is:
how
do
we
enable
like
projects
like
big
cluster
that
are
basically
aims
to
be
to
help
us,
with
more
detainment
scenarios
and
efficiency,
to
interrupt
with
other
projects
like
KDP
and
Dapper?
In
this
case,
for
each
for,
for
making
this
example,
work
I
actually
created
around
like
10
issues
in
all
these
projects
that
they
needed
to
fix.
K
So
now
you
can
have
like
this
virtual
cluster
that
are
very
cost
efficient,
reusing
installations
of
other
projects
that
are
shared
across
you
know
across
different
payments,
and
these
kind
of
collaborations
I
think
it
takes
us
down
like
a
step
closer
to
actually
build
real
life
stuff
that
people
can
actually
start
using
and
to
have
very,
very
concrete
examples
on
how
this
actually
work.
Yeah
and
I
will
not
show
that,
but
the
the
idea
here
is
like
we've
had
this.
K
This
issue
created
there
like
in
the
group
where
we
said
okay,
we
need
to
actually
start
doing
this.
This
platform
prototypes
and
I
totally
agree
with
you
guys,
like
it's
very
difficult
to
maintain
these
examples,
but
I
think
that
the
the
exercise
like
the
the
experience
that
we
gain
by
building
these
things
is
something
that
we
can
use
to
report
back
into
different
projects
and
also
into
the
community
I'm.
Really
like
a
really.
You
know
practical
person
again
so
when
I
think
about
platforms.
K
I
think
about
like
is
this
group
trying
to
solve
the
entire
platform
space,
or
are
we
just
more
focused
on
kubernetes
and
the
projects
that
are
there
and
then,
if
you
are
focused
on
kubernetes,
there
are
tons
of
products.
So
I
will
try
always
just
copy
down
and
make
it
practical.
So
we
actually
have
a
next
step,
because
if
not,
we
like
the
conversations
are
too
abstract
for
me
and
I
can
definitely
go
and
execute
something.
So
it
will
be
great
if
we
can
just
maybe
creating
some
groups
of
explaining.
K
To
find
that
yes,
yeah,
exactly
and
I
think,
like
the
other
thing
that
usually
shows
like
when
I'm
showing
this
platform,
is
that,
like
I
I've,
actually
defined
something
like
an
environment
like
if
I
do
that
right,
an
environment
right
like
in
this
case
what
I
have
is
like
two
types
of
environments,
one
for
developers
and
one
for
machine
from
data
scientists,
and
you
can
actually
provision
these
environments
by
by
having
a
common
interface
between
them
right
like
so.
If
I
have
that
interface.
F
K
K
Yeah
yeah,
so
it
is
so
so
yeah
yeah.
So
that's
like
that's
like
a
project
that
it's
dealing
with
that,
but
I
hadn't
need
to
create
my
own.
You
know
environment
controller
for
doing
this,
which
I
think
it's
also
a
thing
that
I
would
love
to
avoid
like
promoting
people
just
to
create
controllers,
because
they
are
just
extending
kubernetes
for
their
own
risks.
There
are
tons
of
projects
that
are
doing
some
of
those
things.
They
should
do
that
I
think
that
they
shouldn't
should
they
did.
K
They
should
try
to
avoid
it
as
as
much
as
they
can.
If
they
are,
there
are
tools
that
are
already
solving
their
problems
and
if
these
tools,
for
example,
are
not
solving
their
problems,
that's
where
we
go
back
and
report
and
said
you
know
this
is
a
common
use
case
like
you
should
also
support
this
kind
of
stuff,
but
yeah.
That's
my
practical
view
on
the
subject.
A
I
feel
like
that,
fits
again
going
back
to
the
use
cases
and
stuff
like
if
we
Define
use
cases
in
certain
high
level,
and
then
we
did
I
saw
the
phrase
delivery
driven
development
like
a
week
ago
and
I
like
that,
and
that
reminded
me
when
you
were
talking.
What
is
the
end
outcome
we
want
and
then
like
one
of
the
things
we
could
do
is
build
a
prototype
towards
the
use
case.
Yeah.
B
K
K
G
Exactly
so,
that's
part
of
the
vision
of
cncf,
initially
anyway,
projects
first
but
I
mean
like
as
it
expanded.
The
tradition
was
to
be
able
to
have
proposability
between
these
different
projects
and
there's
much
interoperability
as
possible,
and
I
really
want
to
manage
interoperability,
but
you're
talking
about
composability.
You
almost
want
to
take
these
these
projects
as
they'll
move
off
and
say:
oh
here's
sort
of
a
profile
in
a
sense
yeah.
That's.
K
K
It's
composability,
it's
it's
it's
gluing,
stuff
together
and
recognizing
that
you
can
just
not
pick
up
pick
the
stuff
that
it's
out
there
and
just
put
them
together,
and
you
just
need
to
write
some
blue
on
top
and
that
blue
is
really
important.
So
we
need
to
show
people
how
that
looked
like,
instead
of
pushing
them
to
create
everything
from
scratch,
which.
C
Is
what
some
companies
are
a
lot
of
these
things
or
things
that
you
might
not
have
sold
by
by
creating
an
application
platforms?
I
recall
them
basically
a
lot
of
times
so
yeah,
a
certain
type
of
argument
would
kind
of
get
a
starting
point
and
then
people
will
be
adding
whatever
device
basic
examples
of
deploying
to.
Let
us
create
squats
across
certain
tools
would
be
an
archetypals,
and
then
you
build
on
top
of
that.
The.
K
Problem
that
I
see
with
that
is
that
yes,
but
what
I've
seen
companies
doing
is
that
they
will
say:
okay,
we
put
flux,
we
put,
they
need,
we
put
something
else,
and
then
we
build
an
experience
on
top
of
it.
I
think
that
we
are
at
the
point
where
we
are
saying
we're
just
putting
products
together
yeah,
but
we
need
to
build
an
experienced
layer
on
top
of
it,
and
then
we
need
to
start
thinking
about,
like
from
the
consumer
point
of
view,
how
that
does
that?
K
C
That's
how
we
delivered
that
type
of
experience
is
completely
now
we
deliver
the
platform,
so
platform
is
built
on
certain
goals
and
it's
a
starting
point
and
all
right
at
some
point.
It's
going
to
be
very
flexible
and
exchange
parts
and
stuff
like
that.
At
some
point
we
just
have
to
start
with
somebody
yeah
sure,
and
so
we
that
type
of
idea
we
just
call
an
architect
so.
G
Question
since
we're
like
a
bulk
of
this
meeting
is
about
platform
working
group
I'm
just
is
there
a
goal
to
have
user
profiles
to
develop
I.
G
Oh
yeah
yeah,
for
example,
in
the
health
project
that
there's
there's
a
pretty
like
and
I
know
we
have
a
few
home
containers
in
the
room
but,
like
you
know,
like
there's
a
good
there's,
a
really
good
list
like
right
inside
the
community
repository
within
the
helmet
org,
just
called
user
profiles,
and
it's
really
it's
not
this.
It
wasn't
really
a
build
it
and
they
will
come
kind
of
situation
like
we're
like.
Oh
who
are
they?
Who
would
we
like
them
to
be
it's
like
over
time?
G
H
Oh
sorry,
that
it's
really
funny
you
mentioned
that,
like
so
on
that
the
project
that
I
mentioned
this
gen
3
project
people,
so
many
people
were
trying
to
figure
out
kubernetes.
They
put
together
a
Helm
working
group
and
we're
providing
Guidance.
The
first
thing
that
we
did
was
standardize
their
home
charts
to
make
sure
it
could
be
Deployable
anywhere
within
the
company
no
within
the
open
source
project.
So
we
helped
them
redo
their
home
charts,
but
we
took
them
to
to
the
documentation.
H
And
so
they
went
through
this
process
of
like
they're,
going
through
this
process
of
freedom
in
their
stuff,
but
they
had
60
people
show
up
to
their
Helm
charts
working
group
meeting
that
were
from
all
kinds
of
different
organizations
because
it
created
that
connection
just
wanted
to
shout
out
to
the
home
docs
like
where
we
find
our
like.
Our
learning
tools
is
really
interesting.
Yeah.
G
And
to
be
fair,
I
mean,
like
homes
are
really
interesting
project.
It
was
the
second
well
I
guess
third
graduated
project
right
after
kubernetes
anyway.
In
any
case,
it's
been
around
for
a
long
time
and
it
was
probably
very
nice
Sports.
So
like
a
lot
of
those
learnings,
it's
easier
to
get
because
there's
been
a
lot
of
use
of
it,
usership
levels
very
high.
So
we
get
a
lot
of
issues.
A
lot
of
cars
and
stuff
like
that.
G
So
I
know
it's
it's
better
to
count
that
data
and
then
to
kind
of
work
with
it.
But
I
was
thinking
about
your
presentation
too,
and
when
I
was
just
saying
that
about
these
rules
you
know
it
might
check
up
delivery
might
not
or
sorry
the
platforms
working
group
within
Apple
Liberty
might
not
have
you
know,
as
many
users
within
this
community
already
giving
a
lot
of
feedback
just
like
eating
down
your
doors
but
with
their
experience
but
I
feel
like
we
have
some
ideas
of
who
who
the
users
are
right.
It's.
D
Interesting
you,
you
mentioned
it's
a
lot
of
platform
chat,
I'm,
actually
super
curious
on
Tag
app
delivery.
Like
are
we
equally
split
between
the
people?
Who
are
you
who
are
creating
these
apps
that
need
to
be
delivered
and
the
people
who
are
managing
the
tooling
that
delivers
those
apps
or
we
are?
We
do
we
need
one
way
or
another
as
a
tag,
and
then
some
working
groups
could
mean
heavier
one
way
or
another
in
order
to
make
up
that
higher
level
goal
of
either
50,
50
or
or
otherwise,
because.
B
D
Think
a
lot
of
what
we're
talking
about
when
we're
talking
about
composability
of
the
different
apps
is
like
actually
coming
down
to
the
more
platform,
engineering
teams
or
the
more
like
opsy-oriented
teams
that
are
doing
that.
Some
some
organizations
truly
are
end-to-end.
Devops,
like
the
team,
is
putting
up
their
own
cluster
and
their
own
apps
and
delivering
the
web
apps
on
top
of
it,
but
at
scale
you're,
just
not
seeing
that
very
much
right,
you're
seeing
more
18
handles
the
creation
of
the
cluster.
D
Another
team
handles
providing
Services
inside
the
cluster
and
other
team
handles
creating
end
user
value
inside
of
those
apps
inside
of
those
clusters.
So
yeah,
that's
a
that's
been
brewing.
This
entire
day
of,
like
I
joined
I've,
been
very
heavily
involved
in
the
working
the
platform
working
group,
because
that's
the
side
I
have
experienced
it
and
I
I
feel
like
I
can
contribute
to
you
first,
but
I
have
interest
in
all
around
I'm,
just
not
sure
what
the
tag
yeah
focuses
on.
A
Yeah
that
kind
of
team
up
in
that
first
line
when
we
mentioned
development
and
that's
kind
of
a
side
line,
we
seem
to
focus
more
on
delivery.
Getting
it
out.
We
are
I,
think
most
of
us
are
platform
type
Engineers
right
and
that's
one
of
the
solutions
we
found
to
enable
this
app
delivery
for
our
customers.
I.
The
one
thing
you
said:
that's
user
profiles,
I,
do
think
that
there's
two
Dimensions
actions,
the
use
cases
which
is
more
like
the
industry
or
the
vertical
and
then
you're
talking
about
user
profiles.
G
M
Yeah
I,
guess
that's
why
it
comes
the
the
the
actual
apps
and
development.
They
are
the
end
users
they
are
like
they
are
going
to
be.
We're
gonna
recognize
them
by
like
drawing
these
use
cases.
These,
like
Journeys
like
and
then
looking
at
the
personas
in
this
is
the
journeys
that's
the
side,
but
it's
still
because
they
are
the
customers.
So
we
are
interested.
That's
why
we
are
the
majority
of
the
people.
It's
not
someone,
maybe
writing
an
app
unless
they
are
like
they.
M
They
know
they
can
get
some
help
with,
like
okay
I'm,
a
data
scientist.
How
do
I
deploy
my
app?
What
what
products
should
I
pick
like?
They
should
be
able
to
come
to
a
meeting,
maybe
and
then
or
look
at
the
tag
and
then
the
path
and
then
look
and
find
something
there
like.
Okay.
This
is
the
path
for
me.
This
is
like
an
opinionated
set
of
tools
that
I
can
use,
and
this
is
a
project
that
has
them
together
and
I.
Can
I
can
go
on
a
tutorial
so.
A
M
D
G
Exactly
and
this
document
by
the
way
hasn't
changed
in
the
last
five
years
so
but
like
I
can
just
like
read
out
if
you
want
just
like
the
headlines,
it'll
take
like
that
10
seconds.
So
it's
really
just
I'm
gonna,
go
there
five
profiles
and
then
there's
a
list
of
not
let's
go
so
application
operator
number
one:
application,
distributor,
application,
developer,
supporting
tool
developer
and
in
this
case,
because
it's
for
the
hell
project,
specifically
Helm
developer,
for
people
that
are
working
on
the
Home
Tool
itself
and
then
profile
is
not
in
scope.
G
Our
cluster
operator,
because
we
gotta
love
a
lot
of
questions
about
that
and
to
your
point,
but
to
your
point
earlier:
it's
like
hey
we're,
building
an
open
source
tool.
Do
we
also
need
to
teach
you
everything
about
kubernetes
and
how
to
do
your
clusters?
Probably
not
like
you
know,
let's
not
forget
about
the
wheel
here,
I
think
they're.
Probably
the
same
goes
for
tag
after
app
delivery.
H
G
And
in
my
understanding,
an
attack,
app
delivery
is
a
little
bit
a
little
bit
more
focused
because
it's
focusing
primarily
on
delivery
and
not
just
not
everything
about
applications,
but
there's
still
a
lot
of
the
learning
that
came
from
that
and
some
of
it's
distilled
in
some
of
these
projects.
So
it
might
be
nice
to
Bubble
some
of
that
up
to
the
tag.
Now
that
we
have
the
structure
in
place.
H
So
do
you
guys
consider,
as
a
part
of
this
is
a
question
that
I
was
going
to
ask
to
a
few
people
today
as
a
part
of
an
app
delivery
if
I'm
building
a
project-
and
somebody
is
going
to
self-run
this
app
like
they're,
going
to
run
this
on
their
own,
like
I,
typically,
would
think
of
app
delivery
as
like
how
my
app
is
going
to
get
to
production.
But
what
about
how
my
consumers
are
going
to
make
biscuits
production?
That's
that's
part
of
app
delivery.
Also
right!
So
is
that
also.
B
H
C
C
You
learn
how
to
deliver
us
and
it
kind
of
depends
on
like
I.
Don't
think
we
have
any
work
groups
or
projects
that
focuses
on
that
platform
might
be
the
one.
That's
gross
this
to
doing
that.
At
the
same
time,
I
think
we
could
have
a
working
group,
for
you
know,
youth
experience
or
some
some
lymphatic
in
effect
somewhere,
where
you
kind
of
gathered
those
types
of
like
how
should
we
do
how
the
charts
and
foreign.
G
That's
what
I
was
thinking.
It's
like
the
work.
That's
been
done
in
some
of
these
projects
already
from
this
evolving
Community.
Now
it's
expanded.
It's
not
just
kubernetes
anymore
right.
It's
a
it's
a
it's
a
lot
wider
community,
so
we
can
kind
of
take
some
of
those
learning,
learnings
and
maybe
mine,
some
of
those
and
bring
bring
them
up
and
when
I
said,
like
I
think
we
have
that
not
to
say
that
it
can't
improve
or
be
more
clear
but
I.
Think
in
general.
That's
what
at
least
structurally
you
have
that
right!
G
We've
got,
we've
got
this
specific
tag
in
its
focus
and
then
we've
got
working
groups
and
where
there's
a
gap
where
we
need
another
working
group.
Well,
we
when
there's
enough
people
that
come
together
and
it
got
aims
to
fit.
Then
we
create
another
working
group,
however
short
and
overload,
and
then,
when
you
need
expertise
from
let's
say
how
do
you
most?
G
How
do
you?
How
do
you
advance
the
deliver?
An
application?
Okay?
Well,
there's
there's
like
different
category
categorical
places
to
look
like.
Are
we
talking
about
like
oh
we're?
Looking
in
the
area
of
operators?
Are
we
looking
in
the
area
so
we're
looking
at
an
area?
You
know
you
know:
application
delivery
like
packaging
and
and
artifacts
Etc,
right
and
I
think
like
those
are
where
we
fill
the
gaps
with
with
with
work
orders
and
then
hopefully,
I'm
thinking
that
those
working
groups
have
more
interaction
with
the
projects.
G
Basically
like
what
what
you
were
saying,
you
know
like
we
need
more
people
playing
with
moving
these
projects
together
and
bringing
these
into
the
working
group
discussions
and
then
hopefully,
those
working
groups
then
bubble
up
too
because,
like
this
is
an
awesome
conversation
but
like
how
many
of
these
do
we
really
get
like?
We've
got
the
tag,
meetings
and
they're
they're,
relatively
short,
and
key
between
I
think
they're
supposed
to
be
in
the
working
groups.
That's
how
you
scale
this
right,
yeah.
A
G
F
A
Want
to
go
around
a
chance,
the
last
one
of
the
of
the
first
session,
so
you
go
ahead
and
do
those
and
oh
we're
gonna
pass
these
also
out
after
our
boots,
yeah.
J
It
being
the
person
between
the
break
and
everybody
else,
but
I
just
wanted
to
spend
like
a
few
minutes
talking
about
the
cloud
Foundry
story
that
has
picked
up
in
the
past
few
years
that
I've
interacted
with
the
community
largely
because
I
find
that
there's
a
nice
overlap
between
a
lot
of
the
issues
that
we
were
talking
about
today
and
some
of
the
things
that
the
community
has
come
through.
In
terms
of
this
experience
being
like
a
geriatric
platform.
J
I,
don't
want
to
use
the
word
legacy
because
I
know
my
team
will
beat
me
up
for
it.
So
I'm
going
with
like
geriatric
Branch
this
time,
so
the
one
of
the
one
of
the
first
projects
to
come
out
of
the
cloud
Foundry
stable
after,
like
the
whole
world,
had
decided
to
embrace
kubernetes
was
cubes
here
and
basically
in
terms
of
architecture.
Cloud
Foundry
had
already
evolved
into
like
a
massive
set
of
tools
that.
F
J
Know
did
a
lot
of
stuff
in
terms
of
taking
the
application
source
code,
applying
it
onto
something
like
a
container
and
returning
like
a
URL
for
configuring,
like
load,
balancing
and
interest
and
and
allowing
inputs
for
like
services
to
connect
to,
and
things
like
that,
and
so
obviously
the
community.
That
is
a
great
idea
which
is
replace
our
application
runtime
with
kubernetes
and
everything
would
work
right
so
sort
of
like
what
Mauricio
was
referring
to
in
terms
of
take
a
project.
J
They
take
a
project
and
just
put
it
together,
and
you
know
it
should
work,
but
obviously
in
theory
it
would
be
great,
but
in
practice
what
the
limitations
were
that
cloud.
Foundry
was
never
a
great
tool
that
a
developer
could
like
come
back
home
to
in
the
evening,
because
I
mean
we're
not
working
from
home
all
the
time
now,
but
essentially
use
it
for
a
project
like
keep
the
tires.
Try
it
out
on
like
a
laptop
things
like
that.
J
We
have
to
take
it
back
to
work
and
say:
hey:
let's
try
this
for
a
new
party,
so
it
was
very
inefficient
at
being
a
light
tool
and
no
matter
how
many
times
a
community
tried
to
do
that,
it
never
happened,
and
so
that
and
a
lot
of
performance
issues
were
sort
of
the
downside
of
this
first
approach,
where
they
said:
let's
just
replace
our
container
runtime
with
kubernetes
and
see
how
that
worked.
J
So
from
there
the
community
went
to
an
approach
where
they
said:
hey,
there's,
there's
a
whole
bunch
of
these
cncf
projects
and,
let's
start
making
use
of
those
for
logging
and
for
like
Ingress
and
all
of
these
things,
and
so,
let's,
let's
really
put
like
a
bunch
of
these
projects
together
and
see
where
we
can
go
with
it.
And
so
the
approach
again
made
a
lot
of
sense
in
terms
of
what
it
was
trying
to
achieve
like
I,
get
different
projects
together
to
the
interoperability
see
if
they
were
so.
J
It
basically
used
like
if
I
recall
correctly,
like
fluently
for
logging
PTO
for
Ingress.
J
It
used
a
bunch
of
cargo
tools
for
the
actual
deployment
of
the
tool
itself,
but
then
it
still
was
like
a
massive
Beast
didn't
work
on
local
machines.
It
required
like
16,
gigs
of
RAM
and
5V
cqs,
and
something
really
big
in
order
for
somebody
to
even
drive
right.
So
that
was
like
I.
Remember,
trying
it
out
just
to
do
some
demos
and
it
was
like
a
hundred
dollars
worth
of
infrastructure
for
a
week
before
I
could
actually
get
something
going,
and
so
it
was
not
easy.
J
And
now
the
community
has
coalesced
around
this
approach,
where
they
they're
making
use
of
kubernetes
API
and
the
cloud
Foundry
API
and
they've
created
like
a
very
lightweight
layer
over
it
they're
using
help
to
deploy
and
right
now.
This
approach
has
yielded
a
sort
of
a
a
paradigm
where
you
can
get
it
to
work
on
kind
as
well,
so
it
can
it
can
it's
very
composable
and
I
think
that
was
one
of
the
things
that
we
were
talking
about.
J
So
earlier,
approaches
were
like
you've,
got
to
get
this
massive
sort
of
set
of
tools,
and
it
was
very
opinionated,
and
it
was
just.
Oh,
if
you're
using
Hardware
container
register-
and
you
don't
do
that
so
right
now-
it's
it's
a
lot
more
composable,
so
you
can
use
like
a
very
basic
local
Docker
registry.
If
you
want
Harbor,
you
can
do
Hardware.
If
you
want
a
Google
container
registry,
you
can
make
users
easier.
So
it's
it's
transformed
over
time
into
being
something
more
composable,
something
you
can
use
locally.
J
Any
sort
of
change
you
want
to
make
in
terms
of
oh
I,
don't
want
to
use
this
kpac
image.
Printing
service
I
want
to
replace
it
with
my
own,
that's
possible.
So
it's
come
a
long
way
in
terms
of
being
a
very
massive
sort
of
Mammoth
sort
of
Technology
on
kubernetes
into
something
much
more
lightweight,
much
more
composable
and
I
thought.
You
know
this
group
might
be
interested
in
the
story
of
the
evolution
and
where
we
are
right
right
now.
So
quick
question
like
when
you
thank.
K
J
It's
a
bit
of
both
to
be
honest,
so
I
think
you
cannot
do
one
without
the
other.
Again,
that's
my
opinion.
So
the
tools
exist
because
you
want
certain
behavior
and
I
think
behaviors
just
below
certain
tools
and
that's
why
we
have
the
massive
cncf
landscape
right,
so
I
think
it's
a
bit
of
a
quote.
So
there
are
ways
in
which
you
can
get
the
system
to
behaved
in
the
way
you
want.
Okay,
and
there
are
also
ways
to
just
swap
out
here.
Okay,.
K
So,
but
when
you
choose
things
to
add
right,
because
when
you
talk
about
swapping
tools
that
that's
perfect,
but
when
I
when
I
want
to
add
something,
what
I
want
to
always
behave
you
and
I,
it
would
be
great
that
if
I
don't
even
need
to
be
exposed
to
it,
I'm
just
kidding
you
everything
and
usually
behaviors
are
matched
one-to-one
with
those
that's
when
it
becomes
a
little
bit.
Yeah
fussy,
it's
not
an
experience.
It's
just
a
problem,
yes
and
I
need
to
learn
to
use
the
tool.
K
J
I
want
to
add
these
two
I
mean
it
really
depends.
So
each
of
the
components
that
we
are
actually
consuming
are
all
open
source,
and
you
know
you
you
can,
if,
if
you
have
a
new
use
case,
whether
you're
welcome
to
like
open
a
PR
and
contribute
and
stuff
like.
F
J
And
so
I
think
you
know
you,
you
could
modify
technically
the
behavior
of
any
of
these
components
to
you.
K
Think
that
they
are
like
just
too
low
levels
right,
I
I
would
love
to
see
they
are
like.
You
know,
rollouts
or
release
strategies.
What
do
I
need
to
have
in
mind,
set
up
for
being
able
to
do
Canada
reviews?
What
do
I
need
to
add
or
do
feature
flagging
in
there,
even
if
that's
a
project
or
a
tool
or
a
bunch
of
tools
integrated
together
to
give
me
that
functionality
or
that
platform
like.
K
M
A
J
Can
do
a
bunch
of
stuff
with
that
right,
so
I
think
that
again,
in
both
levels,
we've
seen
again.
This
is
in
the
past.
I,
don't
know
if
the
support
for
it
currently.
But
if
you
wanted
to
enable
like
blueprint
deployments,
let's
say
technically
that's
a
capability
that
cloud
Foundry
picked
up
over
the
years
and
allowed
people
to
say
this
is
a
rolling
deployment
enable
them.
J
So
it
became
a
part
of
the
tool
itself
that
we
are
slowly
pointing
to
this
version
of
it
as
well,
but,
like
typically
Cloud
Foundry
comes
with
this
capabilities.
So
it's
a
very
sort
of
Cathedral
versus
Bazaar
kind
of
project
and
traditionally
Cloud
Foundry
has
been
known
to
have
all
of
these
different
capabilities
just
pretty
into
the
tools
and
not
have
application
developers
available.
L
J
M
H
I
see
in
the
space
there
is
the
I
want
everything
obstructed.
I
don't
want
to
do
anything.
I
just
want
to
take
my
code
that
may
have
worked
everywhere
else
over
to
the
platform
and
builds
it
done
and
you're
going
to
have
your
other
side
of
the
house,
where
you're
going
to
have
those
that
want
to
Tinker
they
want
full
customization
and
being
able
to
cater
to
both.
Personas
is
the
challenge,
especially
that
we're
seeing
now
when
we
work
with
platforms
and
platform
engineering
is
conveniently
flexible,
yeah.
K
That's
key
like
because
we're
in
the
kubernetes
space
I
think
that
the
flexibility
is
something
we
are
not
aiming
to
create
a
platform
that
is,
that
is
completely
obstructing
everything
away.
We
want
your
deep
jokes,
so
that's
why
I'm
talking
about
capability
in
this?
Rather
do
you
want
to
build
your
containers
in
your
clusters?
What
do
you
need?
I
mean:
what
configurations,
what
kind
of
projects
are
going
to
do
that
for
you?
You
should
be
able
to
say
I
want
that
and
then
that
things
gets
installed,
a
bunch
of
things
gets
installed.
A
D
J
Yeah
I
was
just
going
to
say
that
comment
Works
in
Gary,
but
in
practice
I've
never
found
that
to
be
the
case.
I
think
it's
almost
always
had
to
be
fun
back
on
the
Gear
Up
developers
and
operators
and
that
kind
of
provision
in
terms
of
who
these
consumers
are
versus
their
application
developers,
who
want
to
be
abstracted
versus
real,
like
a
set
of
people
who
really.
B
G
Yes,
yes,
oh
yeah,
I
think
I
can
appreciate
it.
That's
what
we're
gonna
say:
I
mean
like
what
well
I
think
he's
just
left
now,
but
isn't
kind
of
that
argument
about
like
I
just
want
Ingress
I
just
want
blue
green
I.
Don't
care
about
the
options.
It's
like
that's
what
yeah
I
mean
this
is.
This
is
really
like
what
vendors
provide
right.
They
provide
their
services.
It's
like
you
want
to
work
with
red
hat
cool.
This
is
how
we
do
the
things
you
want
to
work.
G
You
know
you
want
to
do
stuff
on
AWS,
not
that
those
can't
work
together,
but
you
know
here's
how
you
do
this.
Here's,
how
Azure
does
our
thing-
and
you
know
you
want
code
fresh.
We
do
it
this
way.
Etc
I
think,
ultimately,
the
people
that
are
maintaining
these
tools.
We
want
to
keep
them
as
we
want
to
think
about
what
the
scope
is:
keep
them
as
well
scoped
as
possible,
make
them
keep
them
as
flexible
as
possible
and
sometimes
I
think
over
time
you
find
out.
Oh,
the
scope
is
too
wide.
B
D
The
issue
with
the
like
all-in-one
solutions
for
some
people
is
that
an
app
that
may
be
like
I,
lovely,
green
I,
don't
want
to
think
about
or
I
want
future
Vibes
I
want
to
think
about
it.
But
there's
someone
in
your
organization
that's
going,
but
there
are.
We
are
a
PCI
Compliant
secure.
We
have
to
meet
these
requirements.
They.
D
Before
you
can
turn
on
a
feature
flag,
but
you
have
to
provide
that,
and
so
it's
like
there's
that's.
Why
I
think
your
point
about
getting
personas
down
and
discoverable
to
your
point
by
people
who
have
the
different
needs.
It's
really
important
because
there's
I
think
a
big
upfront
platforms
are
that's
where
they're
becoming
as
they're
describing
more
exchangeable
for
parts,
because
you're
finding
that
there
are
people
in
the
organization
need
to
exchange
things.
F
This
is
the
third
model
the
hands
are
getting
used
to
provide.
You
can
always
create
a
spine,
which
is
only
a
data.
You
can
make
your
own.
You
can
use
Dynamic
reading,
but
you
need
to
show
the
compliances
of
how
you're
going
to
work
on
those
all
those
kind
of
things
that
we
saw
that
in
our
solution.
You
have
a
good
way
to
get
user
access
to
production,
which
is
compliant
with
everything
we
need.
F
So
you
need
to
replicate
that
right
now,
but
and
that's
that's
like
for
us-
that's
what
it's
meant
when
the
name
platform
came
from
I,
don't
think
that
tag
is
directly
to
say.
This
is
the
right
thing
to
do,
or
this
is
the
right
thing
to
do,
but
it's
a
great
place
to
gather
the
the
landscape
inside
your
working
group
and
say
that
the
detox
working
is
a
good
example
because
they
have
understood.
We
have
flux
in
there.
I
About
is
like
even
more
than
just
like,
recommending
the
one.
It's
also
how
to
think
about
the
problems
right
and
I.
Think
like
if
you're
going
to
investigate
some
open
source
tooling,
like
what
questions
you
should
be
answering
like.
You
know,
kind
of
tying
that
to
our
requirements
as
well,
I
think
is
important,
like
hey
like
the
vendor
lock-in
thing,
or
this
other
other
things
as
well.
G
G
A
A
All
right,
let's
I,
want
to
give
everyone
a
break
for
people
that
have
been
here
for
the
past
two
hours.
I
know
that,
but
let's
start
with
the
working
groups,
that's
our
next
topic
and
I
I
want
to
let
we
don't
talk
about
artifacts.
First,
let's
talk
about
that.
First
I.
Don't
want
to
talk
about
platforms.
F
A
F
M
M
D
D
A
The
next
part
of
the
meeting-
thank
you
so
much
our
presenters.
That
was
I
mean
to
me.
That
was
really
great.
It's
not
just
about
what
you
all
talked
about,
but
it
gives
us
an
opportunity
to
just
hash
some
things
over
as
a
group.
Next,
we
want
to
go
over
some
of
the
tags
work
in
progress,
in
particular
we're
gonna,
we're
gonna,
hear
from
a
group.
That's
been
we've
been
floating
the
past
few
weeks,
so
we
want
to
hear
their.
F
F
F
C
C
C
Sure,
let's
see
yeah,
it
was
I,
wasn't
I
was
supposed
to
go
and
then
I
people
want
to
confidence.
Okay,
I'm,
not
involved
and.
A
F
H
So
we
know
that
the
world
of
our
native
Computing
is
full
of
artifact.
The
artifacts
are
all
over
the
place
when
you
think
about
it,
and
artifact
is
a
container
image.
An
artifact
is
a
binary
representing
a
jar,
a
tendency
online
before
you
name
it,
unfortunately,
with
every
single
type
that
we
have
out
there
that
have
evolved
over
time.
Each
one
has
its
own
properties.
Each
one
has
its
own
format.
Each
one
has
its
own
way
of
transporting
from
one
or
the
other.
How
do
you
make
make
you
know.
F
H
Make
it,
how
do
you
actually
keep
your
hands
upon
it
for
all
of
them
and
they
keep
getting
bigger
and
bigger,
as
more
and
more
of
these
tools
continue
to
evolve,
how
do
we
try
to
handle
and
use
them
in
Cloud
native
computer,
because
I
work
with
a
lot
of
organizations
as
possible?
I've
been
I'm
I'm,
basically,
an
article
four.
F
F
F
H
Always
going
to
be
here's,
my
implementation
here,
here's
my
tool
of
choice
and
it's
impossible.
So
how
can
we
get
our
communal
content
and
one
of
the
things
we
can
do
is
look
at
ways
of
managing
it,
because
right
now,
I
don't
think
we
have
a
good
way
of
managing
different
type
of
artifacts
within
kubernetes
within
the
cloud
native
space,
and
this
working
group
really
is
to
get
our
hands
around
of
artifacts
and
the
ability
to
try
to
standardize
not
only
how
we
manage
them,
how
we,
how
we
store
them,
how
we
search
for
them.
H
There
is
one
benefit
recently
that
came
out
regarding
artifacts
and
it's
being
able
to
potentially
store
them
in
a
canonical
way.
Oci.
How
many
of
you
are
familiar
with
ocir
attacks?
It
is
a
good
artifact.
It's
a
good,
a
good
reference
use
case.
It's
not
the
end-all.
Bl
I've
had
long
conversation
with
Josh
to
say
that
we
want
to
potentially
add
starting
with
OCR
artifacts,
because
people
look
at
computer
images.
Look
at
s-bombs,
we
look
at
Helm
charts,
they
can
all
be
stored
in
including
registries.
You.
G
Might
say
it's
not
the
end-all,
be
all
yet
right
right,
because
because
oci
has
an
extensions
mechanism,
there's
ways
to
there's
ways
to
improve
upon
it
without
having
to
go
through
the
same
process
that
we've
had
to
go
through
four
years
to
to
get.
You
know
the
the
artifact
spec
distribution,
spec.
H
H
H
So
the
idea
is
that
we
want
to
bring
together
a
lot
of
the
folks
who
are
currently
working
on
those
get
artifact
and
the
definition
of
reference.
You
know
and
reference
specification
standpoint
with
app
delivery,
because
people
need
to
be
connected.
They
don't
want
to
create
these
and
bespoke
instances.
So
we
want
to
use
this
word
as
a
way
that
we
can
bring
all
these
different
players
to
the
party.
H
Some
of
the
goals
that
we're
looking
to
do
offhand
is,
you
know,
creating
a
direct
artifact
common
data
model
between
different
types
of
artifacts
that
are
out
there
or
we
can
go
ahead
and
search
for
artifacts
I.
Look
at
you
know
who
went
home
had
a
feature
where
you
can
search
different
charts
that
are
within
a
repository.
You
can
do
it
somewhat
with
the
docker
API,
but
it's
very
limited.
A
D
G
C
Robert
currently
yeah,
so
we
also
I
wonder
why
there
will
be
like
the
very
small
step
actually
going
there
by
actually
searching
very
photos.
Yeah
yeah,
you
can
search
by
artifact
type
there,
but
that's
the
only
thing,
but
we
are
already
hitting
problems
with
that
because
you
may
have
multiple
artifacts
that
are
from
the
same
type
and
using
this
genetic
Ayana,
for
example,.
G
C
How
many
of
those
and
I
need
only
one
of
those
typical
example
is
we'll
be
presenting
the
model,
vulnerability
management
with
that
and
yeah.
We
have
more
than
one
vulnerability
report.
Actually
you
want
the
latest
one,
so
the
referral
type
by
now
the
filtering
API
will
not
connect
to
this
capability
at
all.
The.
H
Refers
API
for
those
who
aren't
as
familiar
with
mostly
identification
is
a
way
to
add
additional
metadata
to
the
Manifest
that
allows
you
to
provide
references
of
the
source
and
the
who.
G
C
H
F
H
A
G
Have
like
so
we're
basically
like
right
on
the
edge
of
either
being
completely
blocked
from
doing
this
or
completely
full
steam
ahead,
and
it
seems
to
be
hinged
on
whether
or
not
the
oci
org
itself
is
a
better
place.
You
can
have
a
working
group
than
you
know,
then
cncf
and
personally
I
make
personally
I
think
it
would
be
great
to
do
under
oci.
My
big
concern
is
that
I
mean
there's
going
to
be
an
oci
meeting
tomorrow,
right.
G
Yes,
and
no
yes,
we
have
yes,
we
we
did
go
to
the
last
oci
meeting,
okay
and
had
some
conversations
and-
and
there
were
definitely
two
minds
in
the
room
from
people
who
have
been
there
for
a
long
time.
You.
A
G
And
you
know
if
oci
wants
to
maintain
a
working
group
and
we
can
have
some
of
the
same
numbers
go
across
the
two
of
them
fantastic.
You
know,
but
we
know
that
we
can
move
ahead
and
be
really
productive.
Make
really
productive
working
groups
here,
we've
done
it
before
we're
doing
it
currently
and
we
know
we're
doing
it
under
Tag
app
deliberate
right
now.
So
so,
ultimately,
we
would
really
like
to
do
this
and
then
work
with
such
a
working
group
in
the
UCI
org.
B
G
Or
the
actual
implementation
for
the
actual
reference
itself,
we're
actually
building
that
reference
and
going
through
their
process
and
not
trying
to
bypass
them
in
any
way.
So
that's
the
main
thing
we
try
to
impress
upon
them
and
so
I
think
we
just
that's
a
really
good
point
to
keep
that
in
the
language.
A
G
Know
not
at
all,
and
that's
the
thing
is
if,
if
right,
if,
if,
if
that
working
group
becomes
extraordinarily
Nimble-
and
we
just
you
know,
working
groups
with
ncncf
are
meant
to
be
short-lived,
they
are
only
around
as
long
as
they
need
to
be
in
order
to
fulfill
their
their
work,
products
that
they're
set
out
to
fulfill,
and
then
they
should
be
dissolved.
Yeah.
F
F
H
I'll
have
to
talk
about
today.
Well
again,
obviously,
some
stakeholders
using
oci
distributions
different
parties
that
are
out
there.
We
want
to
go
ahead
and
also
look
at
other
Packaging
Systems
aside
from
oci,
so
other
Cloud
native
ones,
that
we
are
already
out
there,
the
kind
of
brings
in
the
whole
ecosystem.
Together
we
open
that
oci,
but
obviously
we're
not
going
to
exclude
it.
H
We
want
to
advocate
for
a
common
format
to
enable
search
and
discoverability
that's
important,
but
especially
when
we
talk
about
the
spiritual
reference
supply
chain
being
able
to
search
and
discover
the
different
type
of
assets
that
are
out
there
very
important,
then.
Finally,
that's
where
we
would
report
it,
but
bring
together
some
prototypes,
some
of
the
activities
that
we
are
focusing
on
as
part
of
this
one
and.
G
To
be
clear
that
the
developing
prototypes,
the
goal
of
that
is,
that
should
fall
outside
of
the
working
group
itself
and
be
done
within,
say.
A
sub-project
of
oras
is
a
really
good
example
of
where
we
can
do
that.
We
already
have
a
maintainer
like
the
whole
group
is
just
like.
Yes,
let's
do
it
in
fact.
That's
where
this
we
should
probably
give
credit,
except
that
Steve
Lester
get
put
together
a
whole
group
of
folks,
including
at
you
myself
and.
H
You
know
so
there
are
a
lot
of
variants
in
parties
and
one
of
the
goals
that
every
one
of
us
who
had
kind
of
been
on
this
episode
for
a
while,
is
that
we
can't
just
do
anything
about
them.
We
need
everyone
up,
because
I
think
we
can
build
some
pretty
cool
things
and
enable
a
lot
of
people,
that's
what
kind
of
all
thinking
about,
but
don't
have
the
ability
to
actually
put
in
the
words
or
into
practice.
F
F
H
I
have
a
bit
of
a
little
career
mode.
Ulterior
motives
that
is,
that
the
work
we
started
in
auras
could
be
that
first
jumping
off
point
to
start
making
candidable
outcomes,
because
we've
already
thought
about
it.
We
already
have
a
lot
of
interested
individuals
and
we
actually
have
some
code
that
we
can
start
to
reuse
as
well.
So,
let's.
C
A
G
Yes,
exactly
and
Josh,
you
were
very
helpful
so
far
in
champion
championing
this
within
the
tag,
because
you
know
working
groups
do
need
a
sponsor
and
you
you
are
sponsoring
this
and
have
helped
to
put
to
bring
in
some
other
examples
of
packaging
formats.
C
H
H
A
H
Process,
this
is
really
our
thoughts
around
with
project
delivery,
artifacts
working
group.
We
look
forward
to
everyone's
contribution
interest
and
thank
you
for
taking
the
time
to
allow
me
to
prevent,
with
your
not
only
host
the
supporting
group,
but
we've
been
looking
for
a
nice
home.
This
is
a
ideal
location
that
has
good
visibility
to
bring
in
all
those
interesting
parties
and
especially
giving
me
the
time
in
the
very,
very,
very
tight
schedule
be
able
to
step
on
it.
C
A
Yeah,
so
the
idea
was
in
this
section
talk
about
all
of
the
tags.
I
mean
I
can
talk
briefly
about
platforms.
If
we
want
there's,
not
there's
not
that
much
to
say,
except
I
wanted
to
use
the
opportunity
to
tell
people
how
to
contribute
so
specific
and
I'll.
Stop
sharing.
G
Just
a
quick
question
as
you're
changing
the
slides
here
would
be
a
good
place
to.
Let
me
already
put
the
issue
out
in
mailing
lists
and
on
GitHub.
Where
would
be
a
good
place
to
hear
feedback
from
folks
about
reasons
why
they
might
not
think
it's
a
good
idea.
You
know
how.
A
C
A
Brought
it
up
there
but
yeah,
we
need
to
bring
it
up
again
and
then
we
need
to
finalize
it
and
actually
I
was
saying
to
Jen
before,
like
that's.
Actually,
you
guys
are
on
impetus
for
us
to
put
some
structure
to
how
we
form
working
groups.
A
G
G
Yeah,
let's
bring
it
up
at
the
next
meeting
and
let's
yeah,
we
need
feedback
from
you
need
feedback
from
COC
and
so
that
that's
why
I
mean.
G
Yes,
exactly
and
that's
what
basically
like
both
Amy
and
Chris
a
said,
the
exact
same
thing
got
back
on
email
and
they
were
just
like
yeah
everything
you
did
was
you
know
we
did
the
correct
thing.
I.
C
G
Yeah
well
Thomas
mentioned
something
today.
I,
don't
want
to
speak
about
him
because
the
other
room,
but
it's
more
like
something
he
said
to
me.
Something
I
can
say
it
is
that
he's
not
sure
that
this
is
the
right
tag
or
this
and
that's
and
then
I
said
well.
What
would
be
then,
ultimately
definitely
looking
for
feedback,
and
his
only
thought
was
what
about
the
OCA
org
I'm
like
okay?
Well,
that's
bringing
that
back
to
that
issue
and
I.
G
Think
we've
already
covered
that
pretty
clearly
both
with
talking
with
the
oci
award
and
just
saying
it
can
be
a
yes
and
it
doesn't
have
to
be
another.
G
G
I
C
G
Else,
the
biggest
issue
to
yeah,
okay
cool
the
biggest
issue-
that
I
heard
is
another
distraction
and
I
just
want
to
mention
real
quick
is
that
there
has
been
some
there
have
been
some
working
groups
that
have
just
kind
of
gone,
inactive
and
I.
Think
it's
important
to
impress
upon
the
people
who
are
in
this
tag
that
that
that
is
certainly
not
going
to
be
the
face
for
this
there's
an
extraordinary
amount
of
energy.
Behind
this.
G
My
I
think
our
concern
is
to
make
sure
that
we
don't
bike
shed
around
the
process,
that
we
find
the
correct
process
of
a
correct
process,
move
through
it
in
a
way
where
it
can
be
fair,
transparent.
It
will
then,
and
all
of
that
which
this
is,
is
and
then
just
get
moving,
so
that
that
energy
doesn't
just
yeah
foreign
inertia.
You
know
maybe.
B
G
G
That
will
be
I,
think
good,
good
feedback
for
all
the
other
working
groups
is
like
well.
When
will
you
know
you're
finished
you
know,
and
how
do
you
know
what's
your
success?
Metrics
I
mean
something
we
can
follow
up
on,
but
just
to
say
that
there's
certainly
enough
of
that.
I
think
in
this
case,
yeah.
F
A
Oh
yeah,
so
platforms
working
group
we've
got
the
paper
out.
I
want
to
show
you
briefly
what
comes
next
for
the
paper
struggling
with
the
mouse
all
day
here.
A
The
way
that
this
it's
structured
now
is
the
two
papers
that
we
maintain
are
kept
at
the
top
level
of
the
repo
and
they
have
a
latest
and
V1.
Both
of
them
only
have
the
one
at
this
point,
and
these
this
latest
directory
is
Sim
linked
into
the
website.
A
The
website
is
the
Hugo
website,
I've,
never
gotten
to
walk
anybody
through
this
before
so
I
might
as
well.
Let
you
know
if
you
want
to
contribute
it's
just
Hugo
everything
that
we
publish
is
in
the
content
section.
So
white
papers
is
actually
just
two
siblings
to
those
back
there
where's.
C
To
get
a
little
bit
more
visibility
on
the
website,
which
makes
sense
yes,.
A
H
It's
funny
because
it
kind
of
went
through
a
little
process
of
like
so
cartographers
started,
had
a
website
for
the
cloud
maturity
model,
and
so
we
basically
just
forked
that
we
followed
the
pattern
and
then
they
ended
up
standardizing
on
this.
For,
like
other
sites,.
A
B
A
Did
I'm
this
was
great.
This
is
a
good
experience,
so
yeah
we've
got
this.
This
whole
thing.
Actually,
if
you
want
to
right
now
in
the
about
section,
there's
just
Ouija
platforms,
because
there's
somebody
here
that
really
likes
Ouija
platforms,
but
if
you
want
to
add
like
open,
git,
Ops
or
reference
here.
C
A
Very
basically
know
we
just
tried
to
get
it
going
whatever
we
needed
to
do,
but
this
is
where
the
white
paper
was
published.
Actually,
so,
if
you
have,
you
know,
if
you
contribute
to
the
latest
version-
and
you
know
once
it's
accepted
it'll
show
up
here
momentarily
that's
the
idea
and
then
eventually
you
know
we'll
cut
a
v11.
H
A
And
then
there's
a
Blog
so
and
by
the
way
so
folks
can
contribute,
like
you
want
to
contribute
other
articles
to
the
blog
from
get
Ops
or
artifacts,
or
anything
like
that.
I'll
just
put
it
in
here,
I've.
C
Wanted
to
add
a
post,
you
may
have
some
sort
of
like
someone
that
looks
at
it
and
said
yeah.
This
seems
like
like
the
right
thing
for
for
the
channel.
We.
A
C
H
It's
something
that
nobody
knows
anything
about.
It's
a
it's
a
markdown
file,
yeah
I'm,
just
thinking
both
both.
C
If
I
wanted
to
do
something
on
perhaps
in
both
open
garage
and
then
as
well
yeah-
and
you
know
me
doesn't
matter
but
understand
that,
like
others,
people
are
asking
me
how
to
send
in
or
if
you
have
anything
thought
about
around
that,
or
is
it
just
basically
do
a
full
request?
I.
A
N
I
think
so,
I
just
came
from
sorry
I'm,
Alex
I
just
came
from
the
TOC
meeting,
and
one
of
the
things
that
I
was
reflecting
upon
was
the
left.
Is
that
there's
a
real
impetus
towards
more
contributors,
more
stickiness,
to
stay
involved
in
the
tag
most
people
come
in
as
transactional
right.
You've
got
something
you
want
to
do
you
do
it
you
lead
I,
would
love
for
us
to
have
blogs
that
people
come
back
to
and
like
hey,
they
post
some
interesting
stuff.
N
You
know
every
so
often
so
I
think
we
would
really
welcome
not
just
thought
leadership,
but,
like
anecdotal
accounts
of
where
you've
used,
operators
or
platforms,
or
even
better
platforms
and
operators
and
get
Ops
in
Earnest
right,
and
that
would
be
a
great
thing
to
do.
Then
you
could
publicize
it
on
our
Twitter
and
stuff
like
that.
N
The
work
in
progress,
Loosely
tied
but
we'd,
be
happy
to
celebrate.
I
think,
like
hey
sure,
your
company's
done
something
cool
yeah.
Why
not
come
here
as
long
as
it's
like
a
marketing
exercise,
you
know
come
and
talk
to
us
about
it,
and
then
we
could
even
have
spin-offs
like
you.
Could
then
interview
you
about
it
because
then
XYZ
so
I
think
something
we've
been
historically
not
bad
at,
but
we've
just
not
directed
our
energy
into
it.
Right.
A
Of
the
things
like
Mauricio
was
here
earlier,
when
we
were
talking
about
like
short-lived
projects
or
prototype
things
that
you
throw
out
after
the
demo,
and
we
were
saying
we
want
to
encourage
more
people
to
come.
Talk
to
us
about
those,
even
if
they're
not
going
to
maintain
in
the
long
term,
tell
us
what
you've
learned.
You
know,
let's,
let's
learn
from
the
smaller
things,
so
that
fits
with
us
too.
That's
what
I
was
thinking
like
if
you,
if
someone
puts
in
a
use
case
where
they
try
to
put
together
platforms
and
operators
and.
F
G
Is
to
Define
contribution
criteria
like
that
specifically
content
criteria
and
now
I'm
kind
of
wondering.
Maybe
that
would
be
good
to
open
that
up
for
discussion
within
the
tag,
because
there
it
might
be
good
to
standardize
that
across
the
working
groups,
potentially.
N
G
Just
reviewed
it
and
tried
to
simplify
it
and
and
yeah
like,
maybe
that
would
be
good
to
share
so
that,
like
each
working
group
doesn't
have
to
like
come
up
with
their
own
yeah.
D
It
was
tried
to
find
a
way
to
how
do
you
get
this
thing
published
and
like
we
have
it
I
don't
know
if
we
end
up
actually
creating
the
documents
around
it,
but
we
were
thinking
about
putting
PR's
into
the
tag
around
that
for,
like
future
working
groups
working
into
a
future
working
groups
wanting
to
get
things
published.
What
is
the
process
and
like
how
do
you
make
sure
you
get
the
right
set
of
reviewers
on
it
and
the
right,
like
cncs
support
for
it
marketing
and
all
that.
I
N
Those
papers
are
getting
quite
a
lot
of
inertia
and
since
we
wrote
the
the
operator
white
paper,
it's
become
sort
of
a
touchstone
reference,
I
think
a
lot
of
companies
who
are
sort
of
looking
to
something
neutral
in
that
space
and
I'd
love
to
see
that
for
the
the
captains.
What
they've
as
well.
A
Actually,
one
of
the
things
I
wanted
to
ask
in
our
session
on
Thursday
is
how
many
people
have
tried
to
apply
all
the
open,
get
Ops
principles.
I,
don't
feel
successfully
applied
them
all.
I,
don't
I
mean
they
definitely
go
to
there
and
look
at
them
and
get
inspired
and
when
I'm
just
wondering
is
it?
Do
they
actually
apply
them
all
the
way
through,
but.
C
It's
a
touch
base
that
we'll
be
discussing
as
well
with
those.
Obviously,
what
they're
kind
of
saying
is,
if
you,
if
you
haven't
applied
all
the
principles,
does
that
matter?
Well,
it
doesn't
really
because
you
do
here,
you
know
in
that
sense,
if
you
want
to
apply
just
half
of
them,
we
are
sure,
but
in
our
eyes
that
wouldn't
be
yet
ups
that
wouldn't
be
invited
on
your
way
to
become
so.
You
know,
there's
a
difference
there
as
well
and
I
I.
You
know
getting
that
type
of
information.
G
Some
of
that
and
and
that
would
be
great
to
repost
on
the
tag
like
website
like
you're
sending-
and
you
know
it
might
be
something
that
we
can
think
about
across
the
different
tags-
the
the
maturity
I
guess
you
know-
I
probably
already
been
talking
about
this,
but
but
the
maturity
model
that
we
wanted
to
also
use
or
get
UPS
was
something
that
other
companies
like,
for
example,
wework
has
already
been
doing
with
clients
for
like
years
and
years,
but
they
could
use
refinement
too
and
and
by
the
way
we
work
sending
more.
G
But
so
this
is
not
like
a
plug
for
them.
It's
just
just
like,
for
example,
like
bringing
the
something
that's
bringing
that
experience,
but
the
idea
was
to
to
not
just
say
like:
are
you
doing
good
I'm
sure?
Aren't
you
it's
more
like
for
each
since
it's
principle-led
that
we
could
have
a
maturity,
kind
of
like
scale,
just
like
sort
of
like
self
self
assessment
type
of
scale
or
on
a
per
principle
basis?
Okay,
maybe
something
like
that
would
be
valuable
in
a
platforms
too,
because
you
said
you
wanted
to
be
principled
and.
D
They
can
imagine
that
being
a
right
turn
to
where
the
paper
currently
is.
A
straw.
Man
currently
yeah,
like
kind
of
being
completely
thrown
out
for
a
very
different
format,
I
think
the
one
of
the
characteristics
of
the
current
straw
magic
that
I
liked,
which
I'm
hearing
you
talk
about
here.
The
get-offs
is
wanting
to
acknowledge
and
call
out
where
you're
not
doing
it
and
I.
D
Think
that's
one
of
the
reasons
why,
while
maturity
model
makes
me
growing
a
little
bit
being
able
to
put
a
level
one
and
describe
the
the
detracting
effect
of
that,
behavior
is
actually
really
powerful,
rather
than
only
displaying
here's.
What
good
looks
like
because
people
will
be
like
yeah,
that's
what
I'm
doing
right.
D
There
explicitly
calling
out
the
behavior
of
stored
in
a
repository,
but
is
applied
by
Pipeline
on
a
schedule
or
not
on
a
schedule
or
on
commit
or
not
behavior
and
being
able
to
acknowledge.
These
are
the
pros
and
cons
of
that.
These
are
the
effects
that
occur
in
your
organization.
If
you
do
it
this
way,
I
think
is
our
role
that
organizations
organizations
choose
if
that's
good
for
them,
because
for
some,
that's
me
like
that's
exactly
the
behavior
we
want.
Thank
you.
I
C
It's
absolutely
yeah.
G
For
sure
for
sure,
and
the
only
reason
I
brought
it
up
now
in
this
context
is
not
to
not
to
try
to
make
that
platform
this,
but
to
say,
hey,
we
may
be
able
to
I
think
the
topic
is
really
helping.
The
tag
help
to
unify
things
for
the
working
group.
A
A
I
would
love
for
us
to
go
a
level
deeper
in
them
and
find
synergies
and
find
ways,
mainly
my
motivations,
to
make
it
easier
for
users,
like
the
ones
that
I've
put
up
so
far
are
the
secrets
one
which
is
service
bindings
I,
feel
like
that's
a
difficult
issue:
that's
not
a
lot
of
money
in
it,
and
so
it's
you
know,
there's
like
20
different
ways
to
get
a
secret
from
the
service
to
the
actual
workload
so
that
it
could
connect
back
and
most
of
the
time
you
know
somebody
figured
it
out
when
they
first
set
it
up
and
there's
a
bash
script
that
manages
it
all.
A
You
know
even
think
about
it,
but
it's
it's
worrisome
like
that's,
things
are
going
to
leak
and
a
lot
of
different
patterns,
so
I
put
that
up
there.
Another
couple
that
are
up
there
sustainability.
We
wanted
to
maybe
talk
about
what
that
capability
would
look
like
in
a
platform
came
down
to
optimization
multi-tenancy,
that
the
reason
why
that
didn't
make
the
paper
is
because
we
felt
like
platform
users,
that's
not
a
capability
they
care
about,
like
they
shouldn't
even
know
their
multi-tenant,
but
for
platform
Builders,
it's
pretty
relevant.
A
So
that's
been
a
topic
that
we
wanted
to
go
for.
Oh
yeah,
Alex.
We
started
the
most
fantasy
white
paper,
yeah
yeah,
but
then
we
realized
that
well,
at
least
in
my
opinion,
we
realized
we
should
describe
the
platform
first
and
then
we
can
describe
why
you
need
multi-dependency
in
the
platform,
so
we're
kind
of
at
that
stage
there
artifact
storage,
which
we
just
talked
about
observability,
has
its
own
work
group,
but
portal
that
I'd
put
down
to
talk
about
this.
This
is
another
thing.
A
That's
on
my
mind,
is:
is
breaking
down
portals
into
multiple
capabilities:
yeah
I,
guess
I'm!
Just
sharing
these
like
if
somebody's
passionate
about
service
bindings
or
is
passionate
about
artifact
storage.
Thank
you
folks
or
any
of
the
domains
that
we've
identified.
You
know
and
wants
to
bring
together
projects
in
that
area.
I
think
that's
another
thing
that
the
working
group
could
do.
C
And
I
think
there
are
a
lot
of
topics
back
on
across
several
of
the
work
members.
C
So
in
our
open
we
talk
a
lot
about
the
secrets
we
get
us
because
that's
this
woman
solving
exhibition
and
there's
a
lot
of
ways
to
do
it.
And
how
do
you
do
that?
And
what's
that?
What
are
the
benefits
between.
C
C
It,
for
instance,
like
the
platforms
working
group,
we
started
to
start
working
on
food
like
Secrets
happenings.
C
C
H
C
G
B
G
To
this,
I
want
to
like
talk
with
you
about
I,
want
to
add
my
ideas
for
some
kind
of
product,
whether
it's
blog
like
me
or
whatever,
Security,
in
the
case
of
in
the
case
of
get
Ops
impact
on
environmental
sustainability.
It
like
actually
involves
you
know,
projects
like
Kepler
and
Coke
lots
of
servers
like
running
basically
running
and
measuring
energy
and
trying
to
like
understand
like
the
state
of
it
now,
and
what
what
it
could
be
able
to
use
it
in
certain
ways.
C
C
G
A
C
G
A
big
topic,
and
maybe
like
the
different
pieces
of
the
outcomes
that
land
in
different
places,
you
know
what
I
mean
like
something:
that's
focused
just
on
like
the
questions
of
people.
How
do
I
do
get-offs
with
this?
Okay?
Maybe
that
makes
sense
to
really
highlight
you
know
in
the
open
gas
project.
Perhaps
we
Bubble
Up
also
to
the
to
the
tag
log
like.
B
C
G
Because
that
could
be
useful
for
okay,
if
you
click
down
into
project
repo
and
then
just
click
click
the
teams
at
the
very
bottom.
That's.
B
G
A
Okay,
I'll
take
a
look
at
that.
We
want
to
keep
us
going
here.
So
looking
at
the
notes,
so
we
cover
platforms,
they
covered
artifacts
covered,
get
Ops
Jen.
Do
you
want
to
talk
at
all
about
operators.
M
Can
you
just
make
quick
things?
So,
as
you
know
like
we,
we
have
to
operate
white
paper
and
it's
it's.
It
was
an
interesting
story
because
we
at
that
time,
like
it
lost
momentum,
and
it
was
like
months
without,
like
anyone
touching
and
then
we
like
picked
up
momentum.
M
That's
when
I
joined
the
tag
and
joined
the
operator
working
group
and
interestingly,
it's
like
it
was
one
of
those
Paths
of
like
someone
who
had
been
in
contact
too
much
with
like
the
cmsf,
but
I
just
started
working
with
operators,
and
then
I
was
like
looking
for
information
to
to
literature
about
it
or
something
to
learn.
And
then
I
saw
the
meeting.
M
I
looked
at
the
calendar
and
this
operator
working
group
meeting
and
then
I
attended
there
and
then
that's
like
so
then
I
was
like
full
of
energy
and
then
like,
like
working
with
Omar
and
homas
and
Alex
then,
and
everyone
who
contributed
like
together,
white
paper
up
there
and
one
of
the
things
is
just
that,
like
as
Alex
just
said
like
it
became
some
like
a
point
of
reference
such
day
when
I
search
I
see.
M
You
know
that
the
operator
white
paper
there,
so
it's
like
really
cool,
but
so
one
thing
that
we,
since
the
platform
is
now
white
paper
and
like
this
thing
of
like
capabilities
of
platforms,
it
will
be
really
nice
to
get
a
fresh
version
of
the
operator
like
white
paper
out
there
and
with
opinions
and
perspectives,
especially
like
matching
like
the
platform
capabilities
and
also
perhaps
a
definition
and
our
like.
M
Like
definition
of
operators
and
the
thing
is
yeah
last
time,
one
of
the
downsides
was
that
it
still
was
quite
difficult
to
get
contributors.
Like
some
people
attended
the
meetings
we
tried
to
like
distribute
work,
but
it
was
ended
up
like
sometimes
I,
was
doing
a
lot
of
the
work.
So
I
would
like
to
hear
from
you.
We
opened
the
the
slack
Channel,
which
is
study
of
delivery,
work
where
operator
white
paper,
and
they
say
all
these
words.
M
It's
separated
by
lashes,
if
you
have
a
story
or
like
a
like
some
something
you
would
like
to
tell
like
is
your
opinion
on
the
evolution
of
operators
and
how
that
yeah
like
how
they
are
used
today
and
how
you
would
use
it,
how
you
use
it
today.
That
would
be
interesting,
so
we
can
get
started,
there
are
open
and
you
can
open
two
requests
to
the
there
is
a
open
operator.
White
paper
lead
to
links.
A
B
M
M
Let
us
know
on
the
chat
and
yeah
we'd
like
to
get
this
going
again
and
have
the
same
like
processes
getting
a
continuous
updates
like
the
platforms
white
paper
then,
but
it
would
be
nice
to
have
a
new
Fresh
look
at
new
implementations
and
because
the
usage
has
changed
quite
a
lot
in
the
last
two
years,
because
that's
being
almost
two
years
now
that
the
white
paper
version
one
was
released
so
yeah
I'll.
A
Say
I
think
I
mentioned
this
to
you:
Jen
I'm,
really
interested
in
the
different
operators
of
operators,
meaning
operator
lifecycle
manager
from
the
operator
framework
or
your
Cloud
chosen
method
for
adding
extensions
to
the
platform
like
there
are
a
plurality.
There's
a
company
called
plural.
That
does
this
too.
Actually,
there's
a
bunch.
A
F
A
No,
you
don't
bring
it
in
not
into
your
cluster
but
like
like
represented
in
your
I
mean.
Would
you
bring
it
into
the
box?
Okay,
I
guess
the
question
yeah
I
think
I
was
right
now
like
comparing
like
operator
lifecycle
manager
where
I
can
go
in
and
choose
I
want
these
operators
in
my
cluster
or
I
can
go
into
Azure
and
I
can
say:
AZ
AKs,
add
flux.
A
N
N
Leverage,
just
to
add
a
little
bit
to
what
Jim
was
saying.
I
started:
writing
the
V2
of
the
white
paper.
It's
already
available
on
the
website,
which
does
have
a
definition
of
what
an
operator
is
I,
just
sort
of
did
that
in
the
holiday
period,
I
also
re-categorized,
all
the
existing
operators
with
the
kind
of
strengths
of
everything
from
you
know
the
python
based
operators
Opera,
you
know
it's
a
key
Builder
operator
framework
Etc,
but
I.
Think
the
important
thing
to
differentiate
is
the
implementation
operators.
N
Wholesale
is
extremely
primitive
right,
they're,
usually
event
based
to
reconcile
it.
That's
not
really
what
an
operator
is.
It
should
be
a
synthesis
of
human
behaviors
and
I
think
that
this
is
an
interesting
concept,
especially
with
local
modeling
such
as
llama,
which
you
can
apply
into
your
models,
which
are
now
seen
with
local
mlai.
So
I
think
the
definition
of
operators
is
going
to
change
dramatically
in
the
next
12
months.
N
So
there's
some
interesting
stuff
that
we
can
really
invite
people
say
Hey,
you
know
help
us
write
the
future
of
what
operators
look
like
you
know.
I've
also
introduced
things
like
Juju,
which
is
not
a
very
well
known
operator
pattern,
but
it
is
a
pattern
and
I
think
the
the
more
widespread
coverage
we
have
of
operators,
the
more
it
will
show
how
disparate
the
landscape
is.
N
For
me,
I
think
what
would
be
really
interesting
to
take
a
a
page
from
the
book
of
the
platforms
group
is
to
define
the
kind
of
the
heuristics
of
an
operator
and
what
are
the
things
that
make
it
an
operator
and
to
say
right.
If
you're
close
to
this
you've
probably
got
an
operator.
No,
it's
value
in
that
right.
Companies
value,
they
can
say
right,
we're
aligning
to
this
idea.
What
are
the
behaviors
an
operator
can
do
right.
So,
yes,
please
do
get
involved.
It's
a
very
fast-moving
subject.
G
Actually,
really,
it
was
sort
of
larval
at
that
time
it
wasn't
really
totally
formed,
but
if
you're
doing
a
a
new
version
of
that
or
since
you're
doing
a
new
version
of
that,
should
we
connect
on
that
of.
C
Course
yeah,
that's
so
important,
I'm
going
to
write
about
that,
also
how
they
were
opening
or
something
inside.
A
On
over
communicated
all
right,
we're
going
nice
and
over
everything,
but
we're
up
to
we're
done
with
the
the
official
stuff.
Now
we
just
have
a
few
lightning
talks
left
me
and
Alex
are
signed
up,
but
I
think
that
we
can
prioritize
ourselves
lowest
wants
to
talk,
wants
to
share
perilous,
yes
scene
is
he
here.
E
A
A
F
Let
me
good.
A
I
Just
getting
I
can
go
yeah
where's,
the.
A
C
C
I
So
so,
thanks
for
Josh
for
giving
me
this
opportunity
so
I'm.
I
A
cncf
Sandbox
project,
it's
called
perilous
and
the
idea
is
to
basically
be
able
to
give
developers
secure
access
into
kubernetes
Customs
right.
So
I'll
skip
this.
So
when
we
talk
about
secure
access
to
kubernetes
clusters,
there
was
a
survey
that
was
done
by
Shadow
server
and
it
turns
out
84
of
kubernetes
customers
were
accessible
via
the
Internet
meaning
you
know,
there's
definitely
a
security
challenge
that
a
lot
of
customers
are
so
basically
right
and
when
we
look
at
a
lot
of
the
traditional
approaches
for
security.
I
Typically,
you
know:
there's
they
try
like
a
Bastion
server.
They
try,
maybe
like
a
VPN,
but
these
are
quite
clunky
Solutions
right
asking
to
a
security
to
open
a
port
22
the
attack
surface,
since
once
you
get
into
the
Bastion,
the
entire
private
network
is
accessible.
I
So
these
are
not
exactly
you
know,
companies
do
it.
They
might
even
use
like
a
jump
post,
for
example,
I'm
going
faster
by
the
way,
just
looking
yeah
so
and
vpns,
of
course,
are
expensive
to
operate
right.
So
there's
a
question
of
like
you
know
when
we're
looking
at
the
requirements
right.
I
Firstly,
how
do
we
give
developers
access
to
kubernetes
clusters
that
may
be
living
behind
like
a
firewall,
for
example
right,
and
how
do
you
make
sure
they
get
just
in
time
access
so
from
a
security
point
of
view
they're
able
to
log
in
and
when
they're
don't
they're,
not
using
the
cluster
anymore
they're,
you
know
they're
the
Privileges
are
taken
away.
I
Is
it
just
in
time
thing
right
for
zero
trust,
and
how
do
you
make
sure
you
give
your
platform
games
the
security
teams
out
of
a
centralized
audit
Trail
so
because
that's
something
they
have
to
show
a
lot
of
time
from
the
compliance
point
of
view
from
a
I
am
auditing
and
security
point
of
view
right.
I
So
when
we
talk
about
perilous,
it's
a
open
source
project.
It's
by
raffae
Rafa
is
a
kubernetes
operations,
company
and
management
technique,
and
some
of
the
core
capabilities
that
we
have
are.
Basically
you
know
the
ability
to
create
custom
roles
for
different
developers
and
the
ability
to
fine-tune
that
prefer
project
or
on
a
per
cluster
basis
right.
I
So,
for
example,
in
a
production
cluster,
maybe
they
have
read-only
access,
but
in
a
developer
cluster,
maybe
they're
able
to
you,
know,
spin
up
a
new
space,
for
example
right
and
another
challenge
is
integrating.
You
know
your
idps,
for
example
like
if
you
have
like
a
octa
or
Azure
ad.
How
do
you
map
that
to
kubernetes
R
back?
That's
been
a
historical
challenge.
Our
back
is
something
that
companies
have
had
to
deal
with
for
a
long
long
time.
I
I
And
then,
and
then
yeah,
we
have
like
a
GUI
to
be
able
to
do
this
as
well,
and
the
way
it
works
is
you
know
just
just
to
kind
of
talk
a
little
bit
about
the
design
is,
basically,
you
can
use
Cube
cuddle
or
you
can
use
a
browser-based
Cube
cuddle
shell,
that
I'll
show
in
the
demo,
and
you
know
the
perilous
server,
that's
something
you
install,
but
it
doesn't
require
any
inbound
connections
to
your
clusters.
It's
all
that
happens.
I
Is
you
install
the
perilous
operator
once
you
import
your
cluster
into
perilous
and
then
it
just
has
outbound
443,
and
the
process
is
like
when
you
log
in
as
a
developer
as
a
user.
Firstly,
you're
authenticated
depending
on
your
local
user
single
sign-on
and,
like
you
are
back
like
hey
like
do
you
have
access
to
this
cluster
or
not?
I
And
then
what
happens
is
if
you
have
access?
Basically,
we
create
a
service
account
just
in
time
and
basically
use
the
cube
apis
then
able
to
respond
using
that
service
account
and
once
you
close
your
tube
cuddle
access
or
you
close
your
browser,
shell,
we
remove
that
service
account
as
well,
and
we
keep
of
course,
like
an
audit
trail
of
all
this
activity,
so
I'll
log
into
the
demo
here
right.
I
So
we
you
know
the
way
you
start
I'm
logged
in
as
an
admin,
so
you
can
create
different
projects
and
in
each
project
you
can
import
a
cluster
and
the
cluster
import
process
is
super
simple,
like
you
can
save.
I
You
know
if
it's
coming
from
a
public
cloud
like
eks
or
like
in
my
example,
for
example:
I
use
digitalocean
because
it's
cheaper
or
you
can
use
like
openshift
Rancher,
anything
else
right
and
so
I'm
not
going
to
show
the
demo
import
or
the
cluster
report
process,
because
that's
something
I
already
did
it's
super
simple
though
you
just
import
it,
and
then
you
install
the
perilous
operator
using
a
cube
cuddle
suppose.
So
you
have
to
go
into
that
yeah,
because
we
don't
want
the
we
don't
want
to
inbound
any
anything.
I
Maybe
it's
only
an
outbound
connection
right
and
so
then
what
happens?
I
Is
you
know
so,
for
example,
here
what
I'm
going
to
do
is
map
different
users
and
I
have
different
groups
right
so
I
have
imagine,
I
have
a
developer
group
and
an
operations
group
right,
and
so
the
Development
Group
they're
only
able
to
access
certain
projects
and
they
have
only
a
namespace
admin
privileges,
meaning
they're
only
able
to
access
certain
namespaces
that
they
they're,
given,
for
example,
by
the
platform
right,
whereas
the
operation
operations
group
they
have
cluster
admin
privileges,
so
they're
able
to
create
namespaces
they're
able
to
create
other
resources
on
the
cluster
and
what's
cool,
is
that
it's
all
tied
with
kubernetes
are
back.
I
So
we
have
all
these
different
types
of
roles
that
you
can
use.
Cluster
admin
project
admin,
namespace
admin.
So
if
I
were
to
create
a
new
user-
or
let's
say,
let's
just
take
an
existing
user,
if
I
wanted
to
assign
the
user
to
a
project,
I
can
assign
different
Privileges
and
I
can
do
that
on
a
per
project
basis.
So
again,
like
the
prod
a
product
cluster,
they
might
only
have
a
namespace,
read-only
or
Dev
cluster.
They
might
have
cluster
admin
privileges.
G
B
A
I
Exactly
and
it's
all
mapped
in
in
perilous-
and
you
know
also
gave
like
certain
like
break
class
access,
for
example,
or
if
you
need
to
like
you,
only
want
the
cube
config
to
last
for
a
certain
time,
so
they
have
to
like
refresh
and
download
a
new
one.
You
can
you
can
do
that
as
well.
A
D
I
So
now,
yeah
exactly
so
so
now,
I'm
logged
in
as
a
developer
and
the
developer
only
has
access
to
certain
projects
here
right
in
this
case
I'm
just
using
the
default
one
and
so
I'm
gonna
do
a
cube
cuddle
in
in
this
and
I'll
show
you
like
the
differences
right.
I
So
let
me-
and
this
is
the
admin
view
so
I'm
just
gonna,
so
as
an
admin
I'm
able
to
do
like
a
get
Ms,
for
example
like
I'm
all
able
to
list
all
the
namespaces
because
I'm
the
admin
right,
but
as
a
developer,
when
we
go
back
to
the
to
their
permissions
right,
we
said
that
for
the
default
project,
the
name,
it's
a
name,
space
admin
and
I
only
have
access
to
a
namespace
called
developer
MS.
So
when
I
try
to
do
that,
get
NS
command.
I
I
There
you
go
yeah,
so
so
yeah
I
only
have
the
ability
to
get
like
my
name
space
right,
so
that's
kind
of
like
the
you
know
the
the
security-
that's
that's
being
done
here
and
the
audit.
Basically
everything
is
audited.
So
if
you
go
to
the
audit
logs,
you
can
see
that
developer.
Sally.
You
know
tried
to
do
this
command
to
try
to
execute
this
command.
You
can.
You
can
filter
it
by.
I
You
know
last
one
month
last
24
hours
and
you
have
the
ability
to
export
this
as
well
from
a
security
audit
and
then
some
clients-
and
you
can
you
can
do
this
via.
Like
the
you
know,
the
the
commands
that
are
being
done,
like
apis
API
logs
as
well
yeah,
so
like,
for
example,
if
I'm
developer,
Sally
I
can
also
download
my
personalized
cubeconfig
and
then
use
that
as
well,
and
it
is
the
zero
trust.
The
same
thing
will
work
so
like.
I
If
you
try
to
do
a
cube,
cuddle
get
Ms,
it
won't
work
when
you
try
to
use
use
IQ
config,
but
because
it's
all
like
secured
and
accessing
the
the
API
endpoint
is
actually
the
perilous
endpoint
that
it's
referring
to
so
yeah.
So
that's
pretty
much
most
of
the
demo
like.
So
we
have
documentation
the
you
know
where
you
can.
You
know
configure
SSO
like
so
you
can
bring.
You
can
do
all
of
that
as
well
with
OCTA.
I
We
have
some
demo
videos
there
we're
going
to
be
at
the
open
source,
Pavilion
thingy
project
Pavilion,
sorry,
so
so
we're
going
to
be
there
we'll.
You
know
we'll
be
there
to
answer
questions
but
yeah
looking.
You
know
just
looking
for
feedback
looking
for
folks
to
try
this
out.
You
know
there's
part
of
the
platforms
you
know
as,
as
you
know,
there's
whole
this
whole
concept
of
like
there's
multi-tenancy,
there's
shared
services
platforms.
I
So,
if
you're
using
like
a
shared
cluster,
these
kind
of
zero
trust
principles
really
double
to
play
as
something
yeah.
I
A
E
A
E
I
can
think
about
this
right.
Ultimately,
everything
that
is
being
done
is
being
done
between
them.
Applications,
services,
workloads
and
identity
defines
who
are
doing.
What
something
doesn't
work
include
is
what
at
the
point
that
is
wrong,
so
there's
an
element
of
identity
and
in
the
kubernetes
round
we
sort
of
there's
a
lot
of
things.
I
mean
a
lot
of
things
about
it,
so
why
why
they
just
introduce
myself,
because
that
would
be
part
of
the
time
I
was
involved
with
earlier,
because
since
30
days
of
kubernetes,
with
defining.
E
E
E
And,
and
so
today,
when
you
deploy
applications
put
this
on
the
shoes
of
a
service,
you've
got
a
couple
of
deployments
a
couple
of
clusters
and
you
have
a
bunch
of
services,
consumers
across
the
network.
It
could
be
across
other
platforms,
it
doesn't
have
to
be
all
kubernetes
with
normal
people
kubernetes,
and
you
need
to
go
figure
out.
Okay,
look
I
need
to
have
a
service
abstraction
service,
abstraction
that
you
can
proxy
SketchUp
with
all
the
lighting
that
happens.
E
Blue
green,
you
talk
about
Canary,
you
talk
about
other
sort
of
constructs
like
that.
Well,
a
you
need
to
have
oftentimes
a
match.
That's
in
fact,
I
can
let
people
think
about
today.
It
doesn't
have
to
be.
But
yes
that's
what
people
think
about.
You
have
Cloud
infrastructure,
dlbs
MLBs,
whatever
load
balancers.
All
of
these
things
to
catch
up
and
when
you
have
this
mesh
data
plane
that
has
to
be
programmed.
So
you
need
a
mesh
controller
and
each
of
these
are
running
in
different
questions.
E
So
all
of
these
need
to
be
Federated
and
put
yourself
in
the
shoes
of
a
service
going
on
that
needs
to
actually
figure
out
how
all
of
these
things
work,
because
now
you're
managing
State
across
all
of
them
and
any
one
of
those
breaks.
It's
not
just
the
initial
deployment
you
shouldn't
forget
any
one
of
those.
Few
proxy
doesn't
sing
correctly
mesh
proxy.
Somebody
doesn't
say:
perfect,
okay,
applications!
E
The
reality
is
I,
don't
know
if
you
guys
agree,
90
of
resources,
science,
costs
and
all
that
other
infrastructure
and
In
fairness
to
platforms
like
Cloud
boundary
I've,
actually
been
in
redhead
openshifted
a
little
bit
of
some
things
like
CF
push,
I'd
say:
look
all
of
this
can
be
abstracted
of
it,
but
that
infrastructure
continuously
exists.
It's
not
just
the
fact
that
you're
saying
as
an
application
developer.
E
What
is
an
ass
is
a
little
bit
different
from
2015
and
different
from
when
Solomon
is
talking
about
hey,
that's
everything
in
a
containers
need
to
be
done
with
it.
What's
in
the
container,
it
might
be
a
dynamic
web
assembly
model
being
involved
in
the
network
right.
So
what
does
it
matter?
I
think
we
divide
first
and
secondly,
what
does
that
serve?
What
is
that
contract
that
have
that
has
with
us
in
viewers?
And
how
do
you
define
that?
And
in
our
view,
that
definition
has
to
be
linked
to
the
consumer's
identity.
E
E
Is
it
someone
that
I
need
to
ignore,
because
right
now,
I
don't
have
capacity
all
of
those
are
defined
by
I
get
it,
but
is
there
a
way
that
we
can
define
a
record
at
the
application
that
provides
these
service
contacts?
It
provides
a
way
for
the
service
owner
to
Define
these
service
contracts
and
not
just
not
worry
about
the
infrastructure
at
all.
Yes,
the
infrastructure
is
always
there.
You
need
to
get
things.
E
But
how
much
of
that
infrastructure
is
not
just
you
can't
ignore
it,
but
I'm
talking
about
the
main
kind
of
instructions.
So
that's
when
you
have
a
service
contract
that
says
this
identity
is
important
and
here's
the
service
level
objective,
that
entire
intermediate
infrastructure
from
the
consumer
to
the
service
needs
to
follow
that
contract
right.
E
A
E
I
have
I
have
three
requirements
and
two
clusters:
if
I
see
a
transaction
from
an
idea
from
an
identity
group
premium
customer
make
sure
this
SLO
has
created
evaluation
again,
which
means
right
now
it
goes
to
deployment
in
this
particular
location.
Despite
the
dks,
this
variable
works
right.
So
if
you
can
get
that
separation
of
concerns,
then
maybe
we
can
start
thinking
about
letting
the
application
order
not
worry
about
it
as
well.
E
Until
now
that
has
nothing
possible
because
we've
been
dealing
with
a
world
where,
if
you
need
to
do
this,
you
can
perform
the
network
right.
So
some
of
the
things
we're
starting
to
look
at
is
capabilities
that
might
be
possible
in
your
technologies
that
enables
some
of
this
and
I've
already
typically
LGBT,
because
that's
one
area
of
focus
for
us
in
a
few
related
organizations,
and
one
thing
that
enables
us,
because
now
for
the
first
time
ever,
identity
is
A,
Plus,
Class
citizen
in
the
actual
transport
of
this
we
all
use
http.
E
Pls
is
no
magnitude,
which
means
every
workload.
At
least
every
service
that
is
running
with
hbb
service
needs
to
have
an
x-597
.
interesting
now
becomes.
What
is
that
identity
understand
that
significant?
You
have
a
science
certificate.
Well,
you
have
something
that
has
cross
associated
with
that,
and
this
is
when
I've
introduced
Technologies
like
stripping,
because
50
provides
a
nice
wheel.
Having
a
well-defined
it's
a
specification
that
says
when
we
assign
you
an
identity,
it
has
a
syntax
right.
You
have
a
trust
domain.
You
have
other
aspects
of
that.
E
You
expect
the
URI
such
as
namespaces
such
as
service
accounts
and
the
other
aspect
of
saying
press
domain.
It
has
to
be
included
in
some
Arabian
test,
and
so
part
of
that
is
how
you
establish
that
address.
Historically,
especially
if
you
look
at
app
delivery,
hey
Google
I'm,
just
going
to
use
something
like
search
manager
create
a
certificate.
E
But
how
do
you
trust
that
that
certificate
is
what
prevents
you
from
creating
a
significant
request
that
says:
I
have
the
scripture
having
assignment
and
what?
What
prevent
someone
remotely
from
saying
I'm,
going
to
trust
that
it
has
a
specific
impressive.
You
have
to
have
to
press
all
the
way
down,
so
increasing
leave
Inspire
with
the
expire
has
a
CSI
driver
search
manager
has
a
CSI
driver
as
50..
So
this
is
something
that
we
take
out
of
the
hands
of
application
developers.
The
workload
comes
up.
E
There's
a
CSI
driver
gives
you
an
identity
owner
has
said
here.
Is
your
species
that
identity
is
kind
of
that
certificate?
Has
the
white
camera
one
up
at
least
insert
manager
as
far
as
a
little
bit
longer
forget
it
early
and
something
that
the
application
owner
should
not
have
already
done
so
increasingly,
the
kinds
of
things
we're
trying
to
do
in
this
layer
is:
how
do
we
take
things
away
from
the
application?
Level
application
owner
should
not
be
doing
a
lot
of
themselves
is
a
attribute
of
that
transaction.
E
It's
an
attribute
of
that
connection.
The
application
owner
should
not
have
to
go
and
define
specific
authentication
policies.
This
is
a
question
of
what
is
connected
to
want
on
The
Wire
right.
So
that's
the
sort
of.
If
you
take
the
qualitative
speed,
I,
don't
know
it's
just
a
sort
of
way
of
thinking
about
this
and
I
I
actually
go
back
to
the
point
where
they
think
about
hey.
We
need
to
figure
out
multi-tenancy.
E
E
That's
something
completely
we'll
take
a
look
at,
but
the
basic
thinking
is
when
you
start
thinking
about
applications,
if
you
don't
have
a
contested
identity,
not
just
for
the
network
communications
Industries
for
things
like
code
signing,
it's
for
a
remote
web
assembly
calls
if
it's
for
any
number
of
things
and
spiffy
and
very
important,
as
far
is
just
the
it
is
essentially
solving
the
last
five
six
years
of
how
people
required
moving
forward.
The
world
is
changing
to
where
now
you
have
Dynamic
instantiation
of
your
application
with
was
equal.
E
He
was
writing
on
the
network
to
have
a
you.
You
need
a
different
identity
when
you're
talking
about
an
application,
whether
inference
engine
has
changed
based
on
new
data.
Another
application
behavior
is
different.
Is
that
so
there's
additional
decomposition
of
identity,
but
these
are
all
the
things
to
think
about,
and
user
identity
what
wouldn't
want
service
identity?
What
services,
what
transaction
identity?
What
happened
in
that
particular
transaction?
E
All
of
those
we
do
come
together
from
compost
be
composed
together
decompose
because
you
have
things
like
proxies
gateways,
other
things
which
are
when
you
don't
use
an
API
engagement.
You
are
essentially
delegating
Island
Navigator
for
your
service
right,
saying,
look
I'm,
giving
your
authorization
to
receive
and
gone
requests
and
send
them
all
of
those
look
at
their
point.
I
have
no
questions.
That's
why
that's?
Why
in
Ashfield
would
love
I.
A
A
My
takeaway
here
is
that,
like
we're
getting
identity,
a
trusted
identity
into
everything,
we
can
leverage
it
for
a
lot
more
than
we
are
yep.
I
hear
you
on
the
service
binding
stuff.
That
makes
a
lot
of
like
the
full
purpose
of
getting
a
secret
over.
There
is
to
prove
your
identity
that
we
have
it
built
into
the
platform.
Then
we
can
start
leveraging
that
built-in
identity
questions.
A
A
C
C
Now
with
due
to
the
fact
that
we
have
a
lot
of
customers
and
Community
users
who
are
still
using
UNL
and
have
a
niche
Factory
of
Legacy
system,
we
would
like
to
might
modernize
that
meaning
we
can
stand
out
or
move
to
grpc
graphql
power
of
the
human
driving
system.
And
that's
where
the
I
think.
Then
personally
and
we
are
going
back
to
my
drugs
and
Universal
approach
in
order
to
lose
and,
of
course,
that
the
opportunities
we
can
put
anywhere
in
the
laptop
and
clouds.
C
And
now,
if
we
are
dealing
about
speeding
up,
API
and
Cloud
native
application,
that's
where
my
drugs
motion
contains
you
again,
because
if
there's
still
a
lot
of
people
who
think
that
contract
and
specification
for
API
is
only
used
to
you
know,
generate
your
documentation
automatically.
We
are
doing
much
more
than
less
and
we
rely
and
we
like
to
use
examples
within
the
contact
and
specification.
I
mean
the
real
example,
something
we
travel
with
a
real,
meaningful
service
that
we're
going
to
provide,
and
we,
by
the
way
we
have
contributed
to
Oasis
API
specially.
C
Is
to
be
able
to
show
the
energy
panel
because
we
can,
as
a
provider,
already
have
contacted
as
optimization
with
your
consumers,
your
client,
your
partners
and
clearly
and
immediately
check
your
implementation
or
bringing
the
right
way
to
get
iterates.
You
can
change
some
paralysis
and
so
and
for
huge
amounts
of
wind
and
regarding
the
time,
especially
at
the
Enterprise
scale,
and
the
second
one
is
finally
using
this
micbox
approach.
You
are
able
to
paralyzed
with
the
product.
C
C
Intelligence
and
we
can
fully
automated
all
the
testing
stuff
for
each
version
in
which
you
have
multiple
API
version,
which
is
going
to
happen
at
7.5
Insurance
example.
We've
done
from
a
single
comment
check
that
you
are
not
doing
any
ranking
changes
and
we
are
going
to
introduce
any
regression
tests
and
recently
of
related
and
last
we'll
talk
about
Jesus.
C
It's
always
better
from
from
my
point
of
view
to
the
testimonials
from
Andrew
the
communities,
one
of
the
maintenance
jobs,
and
this
is
something
that
we
release
an
a
few
weeks
ago
from
the
Deviant
company
in
the
US,
which
is
a
I
mean.
C
There
are
three
three
well-rounding
us
for
transportation
and
Logistics,
and
we
clearly
explained
within
this
blog
post,
why
they
are
sold
in
microbes,
where
they
are
coming
from
them,
that
we
have
been
able
to
sell
and
also
the
benefits
and
there's
a
new
channel
that
benefits
for
one
of
the
very
difficult
business
tools
from
the
branch
market
and
personal
market
conditions
so
from
the
employers.
You
guys
would
just
read
this
because.
A
C
A
C
H
C
Right
has
been
a
huge
philosopher,
the
immunity
micro
devotions,
because
they
can
really
improve
an
example
within
the
specification
to
some
real
data,
something
which,
which
are
useful
I,
mean
before
their
application,
but
they
they,
you
know
from
gpk
open.
If
you
want
some
of
that
close
at
all,
so
we
can
share
their
Network
sandbox
too
many
users,
including
problems,
is
there
there's
a
bit
of
a?
Is
there
a
modernization
play
that
you're,
seeing
clear,
though
at
all
like
folks
that
are
still
stuck
on
soap,
but
want
to
allow
developers
to
use
or
knock.
A
That
and
converted
the
rest
yeah.
We
have
a
lot
of
people
use.
You
know
using
recently
over
to
migration.