►
From YouTube: Kubernetes SIG K8s Infra - 20220302
Description
A
Good
morning,
good
afternoon,
good
evening,
depending
on
where
you
are
thank
you
for
joining
for
the
city's
infra
march,
2nd
bi-weekly
call,
my
name
is
eddie
zaneski
and
I'll.
Be
your
host
today.
Just
a
quick
reminder
before
we
begin
this
meeting
at
all.
Kubernetes
meetings
abide
by
the
cncf
code
of
conduct,
so
please
be
excellent
to
each
other.
With
that
do
we
have
any
new
folks
on
the
call
that
would
like
to
introduce
themselves?
C
Oh
sorry,
it's
me
hi
guys,
so
I'm
new
here
and
I'm
new
to
actually
open
source.
So
I
just
passed
cleared
my
cka
certified
kubernetes
administrator.
I
come
over
the
weekend,
I'm
new
to
kubernetes
and
the
company
that
I
work
for.
We
are
like
not
going
fully
full-fledged
with
the
kubernetes
yet,
but
we
are
exploring
the
kubernetes
option.
I've
got
opportunity
to
get
this
certification
and
explore
the
open
source
world.
So
I'm
here
with
you
guys
and
yeah
welcome.
A
Welcome.
Thank
you.
Yeah
thanks
for
joining
us
cool,
so
diving
into
our
reoccurring
topics.
Do
we
want
to
walk
through
the
the
billing
report.
B
I
I
think
we
want
to
do
that
because
we
basically
beginning
of
the
the
month
of
march
all
right.
So
let
me
share
my
screen.
B
B
So
I
paste
I
paste
in
the
chat,
basically
the
progress
that
enabled
that
service,
and
basically
this
is
done
for
the
bucket
hosting
the
binaries
for
for
caps.
So
I
enabled
google
cdn
service
for
that
to
basically
be
able
to
first
understand
how
things
are,
how
this
is
working
and
also
see
if
we
can
basically
reduce
the
bill,
relate
to
egress
for
this
bracket.
B
B
B
B
So
that's
going
to
be
busy
in
average
of
the
course
we
have
for
this
project
in
the
future.
B
B
E
D
B
D
Okay,
thank
you
yeah,
so
I
ended
up
picking
adolfo
and
asking
for
hey.
Do
we
have
any
statistics
on
how
many
things
get
downloaded,
which
versions
yeah.
B
D
B
D
F
Thank
you
very
much.
Oh
any
chance,
you
would
still
be
able
to
show
your
screen
please
for
the
link
in
the
doc
yeah.
F
So
I'm
just
looking
for
a
lgtm
for
this
thing,
it's
the
context,
is
it
pulls
in
data
from
aws's
cost
explorer
and
then
makes
it
make
it
eventually
will
make
its
way
to
a
data
studio
report
like
we
just
saw,
and
this
this
allows
us
to
bring
in
those
costs
that
we
aren't
currently
tracking
from
aws
into
how
we
manage
the
the
billing
for
the
periods
of
time.
B
Sorry
me
and
tim
already
approved
that
pro
request
is
just
near
lg
tm,
but
I
still
have
some
questions
about
about
the
pro
request.
I'm
basically,
I
understand
the
id
you
bring
in
and
what
you
want
to
do.
My
main
concern
is
basically
the
information
we
want
to
extract
from
that,
because
you
don't
filter
for
amazon
account
right.
F
It's
pulling
in
data
from
the
the
highest
level,
cncf
amazon
account
and
then
from
there
there's
each
sub
project
which
has
been
pulled
in
and
you
can
see
all
of
the
the
cost
per
project
kind
of
stuff.
B
I
think
that's
my
my
main
issue,
that's
my
first
issue
and
also
I'm
concerned
about
the
costs
for
hosting
that
information,
because
currently
there
are
basically
huge
data
set
inside
the
gcp
project
and
that
costs
us
something.
G
For
the
moment,
there's
very
little
in
there
and
we
don't
expect
that-
and
this
is
part
of
the
broader
strategy
where
we're
going
to
bring
in
at
the
moment
we're
just
looking
at
aws
at
the
cost
within
gcp
every
month
and
when
we
start
distributing,
we
want
to
be
able
to
show
where
okay,
this
is
what's
going
on
in
our
google
credits,
and
this
is
what's
going
on
in
our
amazon
credit.
So
that's
so
he's
looking
forward
to
where
we're
going.
D
Yeah
well
in,
in
other
words,
can
we
basically
if
we
want
to
the
thing
that
is
relevant
to
us
right,
yeah.
G
The
way
the
aws
accounts
are
structured
at
the
moment
we
can't
split
it
out
and
the
the
the
specific
account.
So
this
is
kind
of
sandboxed
I'll
see
how
it's
gonna
go.
There's
no
rule
that
it
has
to
stay
this
way.
So
it's
trying
to
get
it
up
and
see
see
that
we
get
it
working
with
what's
in
there
and,
as
we
know,
with
the
account
for
aws,
is
not
set
up,
as
it
should
be
for
production
to
host
any
registers.
G
D
I
think
we
should
just
merge
and
trade,
so
let's
do
that.
B
B
Is
that
so
I
didn't
realize
that
basically,
there
is
a
connector
inside
that
you,
you
export
the
cost,
explorer
data
inside
what
amazon
called
amazon
redshift
and
you
connect
to
dataceo
to
that,
and
you
basically
have
that
information,
because
what
you
want
is
basically
information.
You
need.
You
want
to
visualize
something
you
don't
necessarily
need
to
export
that
to
bigquery
and
eddie.
You
basically
need
to
create
a
new
dashboard
for
that.
F
B
So
I
think
the
next
step
is
is
try
to
filter
per
amazon
account.
Try
to
see
if
we
can
just
basically
get
the
information
for
one
account
and
do
the
one
you
create
for
sig
release,
slash
release
engineering
and
we
can
basically
be
able
to
filter
per
account,
because
I
don't
really
see
value
in
get
all
all
the
cost.
Information
from
the
other
cnc
project.
D
Yeah
yeah:
let's
not
worry
about
that.
I
know
as
long
as
it
doesn't
add
us
too
much
cost
to
what
we
are
doing.
I
think
it's
it's!
Okay!
It's
marginal
cost!
From
from
what
compliance.
G
E
A
B
D
A
E
D
So
can
we
hang
on
for
doing
this
until
until
there's
a
couple
of
things
that
I
wanted
to
talk
about,
and
then
we
can
do
this?
D
Okay,
okay,
because
this
is
gonna
suck
up
the
rest
of
the
time
right.
So,
okay,
so
is
there
anything
else
on
the
discussion
agenda?
No.
D
Okay,
so
the
first
question
I
have
or
no
is:
are
we
ready
to
do
the
registry
dot
case.
B
Retire,
yeah
not
yet
because
we
are
missing,
I
mean
metrics
are
missing
inside
the
proxy
and
who's
who's.
Doing
that
and
ben
bender
elder
is
working
on
that.
Currently
he
got
pulled
inside
proof
review
last
week,
so
we
were
kind
of
but
he's
what
he's
currently
working
on
that
I
mean
that's
the
we
we
talked
last
week
about
that
and
we
agree.
We
should
basically
be
able
to
at
least
have
two
or
three
metrics
inside
the
proxy
and
roll
out.
I
mean
release
a
a01
tag
of
the
proxy
and
basically
do
the
rollout.
B
We
want
to
have
matrix
for
that.
It's
not
a
requirement,
but
it's
an
interesting
things
to
get
because
at
least
we
want
to
have
basically
a
metric
that
express
the
l
check
to
gets
gci
dot
io
and
we
we
may
have
basically
set
up
a
nice
alert
based
on
that
and
basically
see
oh,
if
somehow,
the
proxy
can
pull
an
image
for
kids
on
each
other.
We
should
do
an
action.
D
Okay,
so
let's
let's
do
this?
D
If
we
don't
hear
back
from
ben
by
early
next
week,
yep,
we
should
just
go
ahead
with
it
right
like
what
is
there
tag
it
and
and
create
the
registry.kids.area,
because
I
don't
want
to
wait
right:
okay,
okay,
if
it
can
help
us
with
something,
we
should
just
go
ahead
and
do
it,
and
I
was
able
to
make
good
progress
on
running
an
e2e
test
where
all
the
images
are
pulled
from
registry
sandbox
and
so
far
I
haven't
found
any
problems.
D
Yet
there
were
a
few
few
images
that
were
still
getting
pulled
from
somewhere
else
where
they
shouldn't
be.
I
have
a
thread
going
on
in
sick
testing
channel
where
I'm
logging,
either
issues
or
filing
prs
to
clean
them
up
so
so
far
registry
sandbox
looks
good
right,
so
that
is.
That
is
one
thing.
Thanks
for
all
your
reviews,
there
are
no,
so
then
the
other
one
is.
E
D
Yes,
that
is,
that
is
the
most
important
one
right
so
you've,
given
two
options
there
right.
B
B
H
And
also
you
have
a
question
yeah,
so
this
is
to
set
up
a
cdn
next
related
to
the
projection
of
the
blobs,
and
things
like
that.
B
B
Basically,
we
have
for
everyone
in
the
car.
We
have
three
kind
of
artifact.
We
need
to
distribute
and
I
want
to
apply
for
a
cdn
program
to
make
sure
we
can
distribute
the
binaries
and
the
system
packages.
If
you
do
the
migration,
which
means
we
have
flowed
the
infrastructure
management
for
that
and
the
cost
to
another
cdn
provider.
So
the
question
now
is
which
ones
is
is
better
because
both
offer
some
kind
of
pla
of
kind
of
pricing
for
open
source
project.
B
I
don't
know
if,
if
it's
fully
free
fastly
say
it's
free
cloudflare
is
say
you
get
a
plan
as
an
oversight
project,
so
we
need
to.
I
want
to
basically
at
least
start
the
conversation
with
both
of
them
and
see
which
one
match
with
our
technical
constraint,
because
cloudflare
onboarding
is
very
different
from
fastly
onboarding,
just
from
a
technical,
technical
point
of
view
right,
so
I'm
applying.
Basically,
I
I
can
by
default
I
will
go
very
fastly
because
it's
it's
mainly
used
by
all
the
pressure.
I
mentioned
the
apache
foundation,
the
current
ending
everyone.
B
B
So
we
can,
we
can
go
with
that,
but
I
also
want
to
explore
cloudflare
because
clutter
has
a
bigger
network
in
term
of
distribution
and
they
have
compression
mechanisms
that
can
be
interesting
to
explore.
So
I
basically
want
to
at
least
start
the
conversation
with
them
and
see
what
what
we
can
do
with
one
of
them.
G
If,
okay,
if
a
false,
falsely
says
they're,
free
and
without
flair,
there
is
some
cost.
If
there
is
some
cost
out
of
what
buckets
we're
gonna
pay
for
that
is
there.
B
I
mean,
if
the
article
I
put.
Basically
you
say
it's
free,
so
we
are
not
supposed
to
pay,
but
there
may
be
some
kind
of
condition
on
the
dimension
is
free.
We
need
to
check
that
because
they
may
fastly
can
say.
Oh
you
don't
you
we
have.
We
basically
have
limitations
in
depth
of
total
bandwidth
per
month.
If
we
want
to
get
in
a
person's
project,
it's
not
matching
the
page.
G
G
B
D
About
trying
to
talk
to
them
and
get
an
information
about
what
is
involved,
you
know
what
are
the
restrictions
like
arnold
said
and
typically
there'll
be
some
kind
of
marketing
promotions.
Then
we'll
have
to
figure
out
like
which
kind,
how
much
logo
here
on
a
website,
or
something
like
that.
D
D
This
is
not
a
request
to
like
sign
us
up
for
fastly,
but
this
request
to
talk
too
fastly
to
get
enough
information,
and
then
we
make
a
decision
right
like
so
break
it
up
into
two
parts,
because
it's
easier
for
the
steering
to
give
you
permission
to
go
talk
too
fastly
than
if
you
say
hey.
We
need
to
sign
up
apply
for
one
of
these
programs.
G
I
don't
know
sorry
if
we
tell
tells
falsely
what
the
what
is
the
shape
of
the
things
we're
doing,
because
we
have
all
the
data
to
show
them,
then
they
can
say.
Oh
there's
no
way
we
can
carry
this
or
absolutely.
A
Go
ahead,
eddie
I
was
yeah.
I
was
gonna
echo.
All
of
that
we
should
that
conversation
should
be
have
had
formally
with
steering
and
chris
from
the
cncf,
because
fastly
is
going
to
be
really
generous,
but
we
are
double
dumping,
a
massive
production
workload
on
them
so
yeah.
I
think
we
should
have
like
the
broader
kubernetes
leadership
involved
in
that.
D
Absolutely
so,
as
so,
the
way
I
look
at
it
is
arnold
is
going
to
talk
to
them
and
there
will
be
a
proposal
which
we
are
going
to
circulate,
and
then
all
of
us
have
to
sign
on
off
on
it,
including
sig
release,
and
then
we
turn
things
on
or
whatever.
We
need
to
do.
B
D
B
H
And
also
yeah,
just
a
quick
one,
so
also,
I
think
it's
important
to
keep
in
mind
that,
just
as
we've
seen
with
the
image
mirroring
things,
we
have
very
few
hands
on
the
promoter
right
now
so
changes
there
take
a
while
so
yeah,
it's
a
complicated
setup
on
that
side.
It
can
take
some
time.
D
B
H
D
Yeah
got
it
so
I
know
when
you
email
them
just
cc
me
so.
B
I'm
in
the
loop
is
not
a
email.
Someone
inside
steering
need
to
file
for
in
the
firm
inside
the
page
I
or
I
give,
and
then
we
contact
you.
You
want.
D
Any
question
yeah
that
that's
all
I
had
is
there
anything
else
important
that
is
blocking
anyone.
I'm
basically
asking
the
rhian
and
caleb
here.
D
Okay,
okay
sounds
good,
so
let's
go
back
to
the
the
annual
report.
I
do
need
to
drop
off
to
pick
up
my
son,
so
that's
why
I
was
trying
to
hustle
you
up
quickly,
so
I
can
go
thanks
a
lot
everyone,
and
so
I
know
when
you
make
an
update
to
this
issue,
then
I
will
work
on
it.
Okay,
thank
you.
Bye.
A
A
A
B
B
B
B
E
B
I
G
Do
we
want
to
add
some
some
analysis
of
distribution
per
problem,
yeah.
B
B
A
A
Steven
hasn't
yelled
at
you,
yet
so
it's
okay,
I'm
fine,
okay,
so
okay
initiatives
that
we
are
working
on
that
aren't
being
tracked
in
caps.
A
lot
of.
B
Migration,
migration
for
migration
to
from
manglish
migrate
away
from
get
gcio
to
community
earned
resources
to
community
owned
resources.
This
is
not
a
kept
basically
where
collaboration
with
secret
is
about
container
image
signature.
F
B
No,
no!
It's
it's
not
the
same
thing:
the!
What
ep
made
it
a
policy
about
distribution
but
the
proxy.
It's
a
different!
It's
a
different
approach
at
some
point,
they're
going
to
converge,
but
what
ap
open
is
basically
have
a
policy
about,
basically
how
we
want
to
distribute
content
image
to
other
cloud
provider,
but
currently
we're
working
on
basically
set
up
on
ownership
of
the
proxy
for
kzg,
and
it's
now
under
sick
release,
because
we
we
yeah
his
secret
is
not
kids
and
frog.
B
Yeah,
okay,
that's
the
kept
basically
ryan
open,
but
there
is
a
pull
request
open
for
that
associated
press
that
basically
the
3055
the
first
one.
You
just
missed
that.
A
B
A
Okay,
what
other
big
initiatives
that
aren't
tracked
in
caps.
E
Ownership
of
code
search-
yes,
yes,
tracking
issue
for
that
too,
somewhere,
yes,.
B
Yeah,
where
to
all
still
yeah
exactly
honestly,
jim
is
working
on
that
you
don't
there's
no
blocker
on
that.
Currently
jim.
I
think
everything
that
match.
So
we
need
to
just
build
the
image
and
start
the
deployment.
I
B
Ping
me
on
in
the
on
slack.
I
would
tell
you
it's.
A
A
A
B
B
B
B
A
Okay,
yeah
promotion,
okay,
yeah,
not
much.
A
A
I'm
gonna
leave
it
off
for
now,
but
we
can
add
it
in.
We
got
nine
minutes,
let's
get
through
the
the
stuff
that
we
can
do
real,
quick,
okay,
sub
project
owners
auditing.
So
we
have
a
list
of
all
of
our
sub-projects
that
we
own.
A
B
A
B
Well,
cancer
programs
are
owned
by
sick
country
backs,
and
teams,
mainly
oci
proxy
is
mostly
mostly
it's
mostly
getting
from
in
case
of
io
is
basically
getting
fresh,
so
we
have
at
least
two
six
technical
leads
for
sub-projects.
A
A
All
right
anything,
the
more
specifics
we
can
provide
the
better.
So
what
do
we
need
documented.
B
The
service
will
propose
to
the
community
because
it's
still
not
unclear
what
the
company
the
community
need
to
do.
We've
been
with
gcp
or
amazon.
We
need
to
basically
document
the
safety,
the
different
services
we
propose
to
community
and
how
we
can
how
to
consume
them.
B
If
anyone
I
mean
anyone
that
have
been
around
for
a
long
time
can
answer
that
question.
I
will
appreciate
that
because
I'm
not
sure
I
have
been
this-
I
mean
I'm
mainly
no
many
stuff
inside
this
ripple,
so
I'm
not
sure
what's
needs
to
be
improved
or
not.
What's
the
priority
for
distro,
I
know
one
thing:
my
main
priority
is
documentation
right
now
for
this
year
and
artifact
management.
G
I
think
the
the
aaron's
department
might
departure
might
be
a
thing
that
that
should
be
mentioned
on
the
project
health
because
he
was
really
carrying
a
massive
amount
of
the
weight.
So
I
think
that's
definitely
something
worth
mentioning
at
project
health
at
the
top
level.
B
A
H
Exactly
that's
what
we
were
talking
about
earlier
today,
or
not
so
sometimes
folks,
with
some
infra
background,
want
to
jump
in
and
help,
but
sometimes
it's
not
clear.
So
there
are
issues
that
one
that
seem
to
suggest
that
that
they
want
to
move
things
from
terraform
to
from
bash
to
terraform,
but
sometimes
it
seems
like
those
those
efforts
are
like
frozen.
H
When
when
I,
when
I
was
opening
the
issue
for
the
project,
I
wasn't
in
I've
been
in
this
group
like
for
two
years
now,
and
I
wasn't
even
sure
if
I
should
open
it
in
terraform
or
not
yeah.
We.
B
B
Whatever
tooling
we
are
using.
This
is
the
main
issue
we
decide
to
choose
terraform
because
is
the
more
it's
the
most
stable
and
natural
tool
in
the
in
the
ecosystem
right
now
long
term.
It
can
be
something
else.
We
see
progress
in
cross
plane.
We
see
progress
in
poland,
we
can
go
with
that.
I
mean
I'm
fine,
but
the
main
goal
is
oh:
how
we
introduce
end-to-end
tests
for
infrastructure
management?
That's
the
that's
the
issue.
B
If
we
can
have
so,
I
mean
cops
doing
that
very
well
using
terraform,
but
I
I
justin
barbara,
opened
an
issue
recently
about
exploring
different
tool,
proposing
the
kubernetes
resource
model
for
infrastructure
management.
B
B
The
next,
I
would
say
four
months,
is
make
sure
we
have
a
solution
that
helped
us
register
the
cost
burden
and
her
help
seek
release,
be
sure
they
got
an
image
assigned,
because
I
think
I
see
somewhere
there
is
a
promise
to
make
sure
communities
have
124
sign
with
cosign.
So
I
think
my
my
priority
right
now
is
that
just
because
problem
is
significant,
we
can
open
issue
and
start
the
conversation
about
hey,
let's
migrate,
to
from
batch
to
terraform.
But
for
me
it's
a
conversation.
H
Just
I
think
just
so
I
I
guess
the
meta
problem
here
would
be.
What
do
you
need
help
with?
Maybe
expressing
the
needs
of
the
city
more
yeah.
B
I
will
try
to
rephrase
we
had
an
issue
about
this,
but
it
got
lost
in
the
limbs
of
github.
I
would
try
to
basically
open
an
issue
and
rephrase
that
in
maybe
a
more
refreshed
perspective,
I'll
try
to
do
something
like
that,
but.
B
The
one
thing
I
want
to
say
is:
we
use
opinion
infrastructure
because
we
use
cloud
providers.
So
not
everyone
can
embody
on
this.
If
they
don't
understand,
gcp,
that's
another
constraint
or
they
don't
understand
amazon,
which
is
for
me
open
your
infrastructure.
So
we
may
even
basically
say
give
access
to
someone
to
say:
oh
write
that
there
for
that
are
from
code
for
doing
this.
B
But
if
the
people
don't
understand
those
the
mechanisms
of
the
cloud
provider
is
going
to
be
extremely
difficult
for
someone
to
unboard,
that's
also
another
constraint
and
I
don't
even
think
talk
about
trust
between
the
the
different
owner
of
the
different
sub
project.
So
it's
a
complicated
conversation
and
we
are
out
of
time.
So
I
don't
know
if
you
want
to
continue
that
right
now
and
welcome
back
gabriel.
A
So
yeah,
I
so
the
right
so
stepping
back
is
this
this
article
that
we
talk
about
a
lot
in
the
leap
circle
and
it's
snacking.
Have
you
you've
probably
heard
me
share
this
before
tim's
the
one
who
originally
shared
it
to
me,
I'll
paste
it
in
the
chat
and
the
idea
is
that
I'll
just
share
my
screen
and
anyone
that
has
to
drop
feel
free,
I'm
just
going
to
ramble,
but
we
have
right.
A
So
if
we
have
this,
this
grid
of
impact
of
you
know,
effort
to
impact
and
where
everyone
always
wants
to
land.
Is
this
low
effort
high
impact
quadrant
right
like
that,
is
the
magic
money
maker?
If
you
get
work
that
lands
in
low
effort
high
impact,
like
that's
great,
everyone
wants
to
stay
out
of
the
high
effort,
low
impact
quadrant
right.
No
one
really
wants
to
be
there,
and
so
it's
these
other
two
quadrants
that
are
worth
talking
about,
and
so
the
top
one
is
our
strategy
and
our
mission
right.
A
This
is
where
our
long-term
vision
is
high
effort.
High
impact
right.
This
is
where
the
leads
need
to
be
like
focused
on
their
time,
laying
the
foundation
for
everyone
else,
and
this
bottom
left
corner
is
the
dangerous
one,
and
this
is
this
is
what
this
article
calls
snacking
and
so
the
low
effort,
low
impact
thing
is
something
you're
like
oh,
like
that.
Only
took
me
20
minutes
of
work
right
like
you
did
something
it
made.
A
You
feel
good
because
you
got
something
done,
but
it
was
a
really
low
impact
type
of
task
and,
as
leads,
we
need
to
make
sure
that
we're
migrating
to
that
top
right
corner
and
that's
that's
the
really
hard
part,
and
that's
the
one
thing
that
I
struggle
with
too
on
the
6cli
team,
because
I
don't
have
anyone
to
be
reviewing
and
triaging
cube
control
issues
right.
So
that's
what
90
of
my
time
is
is
just
going
through
cube
control
reports.
A
We
need
to
figure
out
a
way,
and
this
goes
for
the
whole
project
right
to
grow
people
into
that
top
right
quadrant,
and
some
of
that
does
start
with
the
bottom
left
work,
and
so
that's
where
we
can
think
about
scoping
out
more
of
that
terraform
work
or
I
know
just
it.
It
makes
me
sad
to
see
people
swing
through
like
us,
one
or
two
sig
meetings
not
have
anything
to
work
on
and
then
kind
of
just
like
disappear
and
not
come
back
so,
as
leaves
that's.
B
And
that's
I
want
to
say
one
of
the
complexity
of
kids
in
france
is
trust,
delegation
and
trust
because
we
hack
we
are
concerned.
We
are
the
plumbing
under
if
all
that
greatness.
So
it's
going
to
be
difficult
to
basically
and
borrow
people
delegate
and
trust
them
because
they
are.
They
are
risk
related
to
that
and
that's
why
I
want
to
think
about
have
end-to-end
end-to-end
tests
for
infrastructure,
so
we
can
automate
all
of
that
and
basically
start
delegation.
B
B
We're
fine,
I
mean
that's
the
goal,
and
this
is
not
the
easy
thing
to
do.
Jobs
took
years
to
make
that
happen,
so
we
need
to.
I
think
that
and
that's
why
I
want
to
focus
on.
Basically,
the
ghost
problem
we
have
and
the
conversation
we
started
with
secret
is
before
we
before
we
start
that
conversation,
so
we
we
make
sure
okay,
we
can
carefully
use
the
three
million
budget
for
this
year
and
when
we
have
room
about
that,
we
can
start
the
conversation
about
those
kind
of
delegation.
Please
right
now
reduce
the
cost
yeah.