►
From YouTube: Cloud Foundry Community Advisory Board [February 2018]
Description
Video from the monthly Cloud Foundry Community Advisory Board (CAB) meeting. Learn more at https://www.cloudfoundry.org/
Agenda:
* Calendar and tooling
- Now every 3rd Wed of the month (see above)
- See Zoom info above ^^^ in and the #CAB Slack channel
* CFF update
- CF Day at KubeCon 2018 in Copenhagen. CFP available now and deadline is 03/16 (hard)
- CF Summit Boston code for contributors available now
- CF Summit Boston - 392 registrations as of Feb 20. Please spread the word with your teams and on your social feeds
* PMC Projects updates
- Runtime PMC [Dieu]
-- Envoy is now added to containers
-- ContainerD inception
-- Garbage collection (docker image layers etc) improvements
-- CF Release
--Diego 2.0 coming
- BOSH [Dmitriy and Danny]
-- Continue to work on faster deploys and updates
-- Various updates for stemcells
- Open Service Broker API [Matt McNeeney] (https://www.openservicebrokerapi.org/)
-- New co-chairs
-- v2.14 update
- Extensions [Max] (https://github.com/cloudfoundry/pmc-notes/blob/master/Extensions/2018/02-26-2018.md)
-- CF-Abacus update (presented today)
-- Review of updated proposals for SAP’s App Lifecycle and Pivotal’s MS SQL Broker
-- Various updates from other projects in drumbeat from last month here
* Community projects and presentations
- CF Documentation Process and How you can contribute? by Benjamin Tarnoff and team, Pivotal. (https://docs.cloudfoundry.org/)
-- Contributor Guide (http://docs.cloudfoundry.org/concepts/contribute.html)
*CF Abacus Demo / Update by Hristo Lliev and team, SAP (https://github.com/cloudfoundry-incubator/cf-abacus)
B
A
By
the
way,
anybody
that
knows
how
to
figure
out
how
why
Google
Calendar
keeps
sending
a
new
invite
when
you
make
a
chain
to
like
public
calendar,
because
I
think
in
my
calendar
person,
founder
I,
have
something
like
six
version
of
the
CF
cap
month.
People
and
I
should
really
have
to,
and
the
reason
is
because
I
keep
making
some
small
chain
that
somebody
says
invite
me
so
I
invite
them
in
add
one
more.
So
if
you
know
how
to
fix
this
I'd
be
very
grateful.
A
B
A
A
B
A
Okay,
so
thank
you
for
coming
to
the
cat
ball.
Second,
one
of
the
year
there
is,
we
have
two
speakers
so
been.
Throwing
up
from
Peduto
will
be
discussing
the
CF
Docs,
so
Anna
and
I
had
a
chat
with
him
and
and
you'll
see
it's
really
good.
It's
a
short
meeting
short
discussion
about
CF
Doc's
attorney.
You
know
make
sure
you
try
to
listen
to
this
because
we
need
help.
So
so
then,
we'll
talk
about
that.
The
other
part
of
the
presentation
it's
going
to
be
here
still
from
SAT.
A
This
is
the
team
in
Sofia
and
they
actually
have
a
lot
to
discuss
so
we'll
try
to
give
them
as
much
time
as
possible.
I
told
about
20
to
25
minutes,
hey
juice
on
my
VM,
so
we
usually
get
started
with
kind
of
overview
of
the
different
PMC's.
So
this
try
could
be
short
but
obviously
give
us
much.
You
know
updates
that
you
want
I'll
just
kind
of
time
it
so
that
we
can
make
sure
the
speakers
are
some
time.
So
maybe
we'll
start
with
Shana.
A
C
I,
don't
have
the
registration
numbers
quite
on
hand,
but
I
think
we
are
definitely
on
track
similar
to
what
we
were
in
the
in
terms
of
the
registration
numbers
as
it
pertains
to
North
America
last
year.
I
do
have
one
quick
announcement
in
terms
of
Cloud
Foundry
days
we
just
announced,
or
we
just
watch
the
website
live,
but
we
are
gonna
have
a
Cloud
Foundry
day
at
cloud
native
con
Cuba
in
Copenhagen,
so
I
will
drop
the
link
in
the
cab
called
meeting
notes
and
also
here
in
the
zoo,
Channel.
C
But
the
event
is
on
May
1st,
so
the
Cloud
Foundry
day,
which
will
be
run
by
the
foundation,
is
on
May
1st
in
Copenhagen.
The
call
for
papers
is
now
open.
The
one
thing
that
the
events
team
wanted
us
to
emphasize
to
the
community
is
that
this
time
there
will
not
be
an
extension
for
the
CFP
deadline,
because
we
just
don't
have
the
time
to
manage
that
extension
leading
up
to
the
event,
so
the
CFP
hard
deadline,
close
date
is
March
16th.
C
C
The
schedule
for
Cloud
Foundry
summit
is
coming
through.
Well,
we
will
have
several
events
on
day
0,
like
the
hackathon
user
day.
We
are
also
looking
at
one
other
technical
project
hours,
project,
Alcazar's
kind
of
a
day.
That's
still
in
the
works
we
will
announce.
We
will
announce
that
shortly
as
well
and
I
think
that
most
of
the
agenda
is
on
the
website
already
on
the
website.
Already,
if
you
have
any
questions
or
any
other
kind
of
concerns
or
feedback
in
terms
of
summit,
Boston
drop
us
a
note.
A
A
A
People
just
sure
when
I
sent
him
an
email,
so
I'm
kind
of
happy
about
that.
I,
don't
know
how
long
that's
gonna
last
point:
since
there's
acquired
two
people,
we
had
Megan
on
Google,
so
that
was
great
and
then
both
of
these
guys
they're,
Eric
and
Jules,
where
it's
part
of
it.
So
thank
you
for,
though
so
anyway,
I'll
try
to
put
some
video
when
I
get
a
chance.
B
A
B
A
B
B
B
D
This
so
I
guess
the
the
two
things
big
quite
tiny
thing
is
on
Monday
we're
incepting
on
putting
in
container
D,
and
so
you
have
a
kind
of
consistent
operator
experience
hopefully
actually
used
to
container
be
show
time.
Some
nice
deployment
scenarios
and
we've
also
hope
to
release
in
the
next
couple
of
days
as
soon
as
I
push
the
button
and
finish
accepting
things
we
think
we've
improved
the
garbage
collection.
China
has
been
a
pain
point
for
a
lot
of
people
for
a
long
time.
D
I
think
we
finally
figured
out
a
low-hanging
fruits
way
of
doing
the
garbage
collection
that
that
should
tide
us
over
for
a
little
while,
and
so
we
can
do
something
much
better,
but
basically
look
out
for
the
release
notes
on
the
next
garden,
because
that'd
be
some
opting
that
you'll
have
to
do.
But
hopefully
the
upshot
should
be
that
we
can
take
care
of
all
the
sizing
of
the
cache
that
we
have
and
operators
will
no
longer
have
to
guess
fairly
unguessable
values,
which
was
the
old
situation.
D
D
D
D
Think
at
the
moment,
they're
kind
of
letting
all
flowers
clue,
and
then
they
put
the
co
in
the
same
way
that
we
look
on
it.
So
we
can
have
lots
of
runtimes
at
the
back
and
you
don't
really
care
that
much.
They
have
CRI,
so
you've
got
C
all
right.
Doctor
I
just
seem
like
it's
going
away
and
then
cRIO
and
see
how
I
container
D
and
maybe
eventually
people
standardized
of
one
of
the
other.
D
Obviously,
in
Cloud
Foundry
we've
got
a
bit
of
a
more
curated
situation,
so
our
team,
a
garden
theme,
has
decided
that
we
kind
of
like
that
contain
of
the
approach
so
provisional
Ealing
something
we
haven't
done.
Yet
we've
looked
them.
You
think
it's
the
right
approach
for
us.
Let's
share
a
bit
more
code
and
if
you
could
pull
the
stories
that
we
like
we're,
gonna
try
and
go
down
that
path.
If
it
turns
out,
doesn't
work
we'll
back
out
and
go
back
to
something
that
looks
a
lot
more
like
creo
Inception
is
to
is.
D
So
we've
done
some
investigation
already
we,
despite
for
a
so
we
have
pocs
that
look
like
they
work,
a
look
pretty
promising,
so
we
we
feel
fairly
confident
the
moment
us
it's
going
to
be
a
promising
direction.
Obviously
we're
agile
and
you
know
if
it
says
that
we
hit
a
roadblock
we'll
back
up.
We
don't
think
it's
a
massive
amount
of
work
for
us
to
pull
it
in,
because
I
actually
match
this
kind
of
a
component.
We
already
have
that
pre-exists
because
we
really
could
garner
one
seat
before
container
the
existed
sure.
D
So
we
have
some
components,
that's
very
similar,
but
actually
we
took
a
very
similar
approach
to
container
B,
so
it
look
quite
similar.
So
it's
quite
easy
for
us
to
pull
in.
We
think
so.
It
feels
like
the
right
move,
but
I
don't
think
it
will
be
everything.
Operators
already
notice
much
in
the
same
way
that
one
day
we
switched
on
gardener
and
see
for
everyone
and
God
Linux
went
away
and
it
was
mostly
visible
to
people
who
are
using
CF
release.
This
should
be
mostly
you'll,
just
get
it
for
free
Dimitri.
D
Will
say
so
our
current
plan,
at
least
to
begin
with,
is
if
you
will
have
the
options
opposed.
So
the
idea
is
we're
going
to
maintain
our
own
thing
to
begin
with,
especially
because
containment
he
doesn't
have
window
supports
and
then,
if
you
put
a
bot
property
with
the
container,
these
sockets
on
to
garden
then
will
delegate
to
container
B.
D
A
B
A
B
A
E
A
Danny
gave
a
really
nice
talk
yesterday
on
some
of
the
new
stuff
happening
on
Bosh
I'm
hoping
he
gets
a
chance
to
give
it
oh
yeah
summit,
so
otherwise
you
can
pay
him.
I
thought
it
was
really
good.
Thank
you
for
that.
Benny,
alright!
So
maybe
Matt
Matt,
meaning
he's
the
new
West
PMC
chair
for
open
serviceworker
and
Rochelle
Davis
was
co-chair.
B
I'm,
just
hiding
somewhere
yeah
so
with
Alex,
lay
changing
roles,
Doug
and
I,
and
our
taking
on
his
own
responsibilities,
chairing
the
open
service
broker.
Api
group,
no
major
updates
really
on
that
we're
still
targeting
the
214
hoping
to
get
out
in
the
next
couple
of
weeks.
Really
there's
a
few
new
interesting
features
in
there
to
do
with
a
fetching
service
instances
and
bindings.
So
you
can
do
things
like
fetch
any
config
parameters
you
use
to
configure
those
things
and
also
supporting
asynchronous
bindings,
which
is
going
to
be
useful
for
a
few
broad
crosses.
There's.
B
A
new
kind
of
major
initiative
going
on
is
to
do
with
service
instance
actions.
So
how
a
broker
could
in
like
a
programmatic
way
using
an
opening
key
I,
don't
expose
the
actions
you
can
perform
on
a
service
instance,
and
you
guys
can
kind
of
build
things
around
that
to
do
is
common
life
cycle
stuff
like
backup
and
restore
and
starting
and
stopping
instances.
B
A
A
A
So
first
you're
gonna,
hear
about
the
package.
I
think
there's
quite
a
bit
going
on
there.
So
you're
doing
the
team
I'm
gonna
discuss
this
Stephen
from
pivotal
has
various
updates.
So
you
can
see
from
our
release
notes
that
our
suspension
quickly,
he
has
a
bookshelf.
So
the
local
project
is
moving
strong
and
then
also
he
has.
B
A
Or
even
say,
statistics
with
yeah
you're,
absolutely
right
and
then
another
quick
thing
to
mention.
So
of
course
you
can
go
to
the
notes
and
get
all
the
updates
there's
quite
a
bit.
So
don't
just
assume
that
I'm,
just
highlighting
with
you
the
of
course
the
Kubo
projects
moving
along
quite
well
and
George
has
been
very
good
at
giving
very
detailed
feedback
on.
What's
going
on
there.
So
they've
been
you
know,
cutting
releases,
you
know
constant
periods.
A
So
if
you're
interested
in
that
project
make
sure
and
and
then
final
thing
I
wanna
mention
is
that
service
fabric
they
have
washed
it
all
those
support.
So
basically
they
all
sort
of
modernized
for
a
Bosch.
So
if
you
were
interested
in
that
project
in
order
to
dry
definitely
check
it
out,
okay
and
then
there
are
a
couple
things
coming
down.
The
pipeline
hopefully
emerge
we'll
see
who
we
get
to
schedule
it.
A
It's
getting
to
a
point
now
where
we
can
only
really
do
two
talks
for
cattle,
but
there
is
more
than
two
at
a
time,
so
we'll
have
to
figure
it
out
what
we
go
in
the
future.
But
that's
that's
about
it
for
extensions
alibied
here.
So
let's
go
to
the
I,
guess
the
main
item
for
the
week,
so
we
have
been
torn
off
from
digital
and
then
after
that,
Mitchell
then
has
a
shorter
presentation.
A
F
Perfect
yeah,
so
thanks
so
much
for
having
me
I'm
Ben
I'm
on
the
pivotal
Doc's
team,
I've
just
got
a
very
brief
presentation
want
to
give
people
a
little
bit
more
context
on
how
the
Cloud
Foundry
Doc's
are
made.
Some
of
the
challenges
we're
currently
facing
with
those
and
hopefully
how
you
all
can
can
help
so
I'm
gonna
share
my
screen.
F
All
right
can
you
all
see
that
yes,
brilliant
all
right,
so
this
is
a
very
short
presentation.
We
can
kind
of
run
through
it
pretty
quickly,
there's
a
very
brief
demo
portion
and
that
I'd
love
to
hear
any
questions
you
might
have
all
right.
So
here
is
the
state
of
the
world
to
kick
things
off
as
I'm
sure.
All
of
you
know
the
Cloud
Foundry
Doc's
currently
live
at
Doc's
Cloud,
Foundry
org.
F
You
haven't
seen
them
recently.
This
is
what
they
look
like,
who
reads:
the
Cloud,
Foundry
Doc's,
just
kind
of
understanding
who
our
audiences
are.
As
far
as
we
understand
it,
we're
looking
at
three
audiences
we're
looking
at
operators.
Obviously
they
need
to
know
how
to
install,
configure
and
manage
CF
we're
looking
at
developers
who
need
to
know
how
to
run
and
manage
their
apps
on
Cloud
Foundry,
and
then
we're
looking
at
a
third
category
that
you
could
call
tire
kickers
and
others
who
want
to
know
how
Cloud
Foundry
works.
F
Anyone
who
uses
Cloud
Foundry
they
sign
the
contributor
License
Agreement
they
can
contribute
to
the
documentation,
but
in
practice
right
now,
pivots
right
about
90%
of
the
docs,
and
you
might
ask.
Why
is
that
a
problem?
So
what
we've
discovered
this
is
coming
out
of
conversations
that
we've
been
having
on
the
pivotal
Doc's
team
is.
This
is
what
we're
worried
about,
which
is
that,
when
the
Cloud
Foundry
Doc's
depend
overwhelmingly
are
primarily
on
pivotal,
we're
worrying
that
the
accuracy
and
completeness
of
the
dachshund
suffer
as
a
result,
and
it's
not
because
we
don't
enjoy.
F
Writing
those
dogs.
It's
not
because
we're
not
passionate
about
Cloud
Foundry,
but
it's
because
we
don't
have
as
much
time
as
we
like
to
devote
to
the
Cloud
Foundry
Doc's
right.
We're
spending
a
lot
of
time
on
PCF
on
PKS,
on
these
new
commercial
products
and
what
we're
worried
about
is
less
usable,
Doc's
right
that
don't
have
that
accuracy
and
completeness
which
may
lead
to
fewer
people
using
Cloud,
Foundry
being
successful
with
Cloud
Foundry
and
again
without
those
Doc's
people
can't
use
CF.
So
that's
what
we're
worried
about
from
the
pivotal
Doc's
team
perspective.
F
F
F
F
There
are
various
ways
to
contribute.
The
easiest
way
and
we've
tried
to
streamline
this
process
as
much
as
possible
is
to
use
a
pull
request
right.
So
every
topic
and
the
Cloud
Foundry
Doc's
lives
on
github
as
a
markdown
file
and
at
the
bottom
of
every
topic
you
can
actually
find
a
link
that
will
take
you
to
the
corresponding
topic
in
the
corresponding
github
repo,
and
you
can
do
a
pull
request
right
through
there.
F
The
thing
I
would
emphasize
here
we're
going
to
do
a
demo
briefly,
but
I
think
what
we're
trying
to
emphasize
as
much
as
possible
is
for
folks
not
to
worry
too
much
about
style
and
structure
in
the
language.
You
can
just
throw
the
raw
data
at
us
over
the
wall.
We
just
want
it
to
be
correct
and
complete
as
much
as
possible,
all
right
so
I'm
sure
many
of
you
have
done
pull
requests
on
our
Docs
before
we
thought
it
would
be
useful
just
to
do
a
very
brief
demo
of
how
you
might
do
this.
F
So
this
is
the
contributing
documentation
topic.
As
you
see,
there's
actually
a
ton
of
information
in
here
you
can
kind
of
go
as
deep
as
you
want.
We
wanted
to
provide
as
many
resources
as
possible
for
people,
but
at
the
very
bottom,
like
all
the
topics,
it
has
create
a
pull
request
or
raise
an
issue
on
the
source
for
this
page
and
github
right.
So
let's
say
we
were
reading
this
topic.
We
found
something
that
was
incorrect.
We
click
at
the
bottom
and
we
can
edit
this
in
github.
F
Thanks
typo
and
you
won't
be
able
to
commit
directly
so
it'll
look
like
that,
and
then
we
can
create
our
pull
request
and
we'll
immediately
be
notified
of
that,
because
that
populates
directly
to
our
tracker
and
I
have
a
link
for
that
tract.
If
you
like
to
look
at
it
later
in
the
presentation,
the
CF
Docs
public
tracker
so
we'll
get
that
right
away.
F
So
that's
a
pull
request,
pretty
simple
process:
when
does
it
make
sense
to
update
the
docs,
so
we
basically
have
two
scenarios
right:
we
have
maintenance.
Ensuring
existing
Doc's
are
accurate
and
complete.
This
might
be
a
good
onboarding
activity,
particularly
if
folks
are
coming
in.
Maybe
two
points
yep
for
the
first
time
pushing
an
app
to
see
it
for
the
first
time
doing
something
for
the
first
time
they
might
be
looking
at
the
docs
and
kind
of
reading
them
more
closely.
F
F
Ensuring
new
Docs
are
written
when
new
features
are
developed,
and
the
idea
here,
of
course,
is
that
the
team
that
owns
the
feature
owns
the
docs
right
so,
as
sabeti
was
putting
together
the
plan
for
CF
deployment,
how
they're
gonna
roll
that
out,
he
reached
out
to
me
and
we
talked
about
how
those
Doc's
are
going
to
get
developed,
what
the
operator
needed
to
know.
So
that's
another
option
right
for
something
like
CF
deployment.
F
This
is
where
all
that
github
activity
populates
as
well,
we're
also
slowly
starting
to
cultivate
CF
Doc's
evangelists
in
member
companies,
so
I'm
the
point
person
for
this
again.
If
you're
at
a
company,
that's
not
pivotal
and
you
want
to
push
documentation.
Practices
within
your
organization
drop
me
a
line.
We
can
talk
about
what
we're
thinking
right
now
and
how
we
might
widen
the
contributor
pool
for
docs
all
right,
so
I
think
that's
the
presentation.
I
think
we
have
a
little
time
for
questions
like
I.
Think.
B
A
F
So
MK
docks
is
something
we're
actually
using
for
the
the
CFC,
our
Kubo.
So
if
you've
looked
at
that
site,
it's
like
a
standalone
dioxide.
So
the
answer
is,
we:
are
you
still
using
bookbinder
for
most
of
our
docs
just
one
case
of
the
CFC,
our
Docs
we're
using
MK
Docs.
So
it's
an
ongoing
conversation
I
think
we're
actually
trying
to
get
the
pub
tools
team,
which
was
the
team
that
built
the
binder
originally
spun
back
up.
F
E
F
It's
possible
I
mean
the
services
folks,
really
like
it.
The
struggle
with
us
with
MK
Doc's
is
that
there's
a
few
things
that
it
doesn't
seem
to
support
natively
that
we
would
need
a
development
team
to
build
it
out
for
us,
so
support
for
multiple
versions
of
the
docs
content
sharing
right
in
terms
of
how
you
would
do
single
sourcing
of
material
partials
template
variables.
So
there's
a
few
things
we'd
probably
want
to
cross
off
the
list
before
we
can
roll
out
MK
Doc's
more
generally.
E
B
F
A
Okay,
well,
thank
you,
Ben
and
I'll
say
one
thing
again
not
to
try
to
compare
ourselves
to
other
communities,
but
maybe
I've
been
looking
at
you
know,
communities
and
during
the
kubernetes,
their
documentation
is
very
extensive
and
it
doesn't
seem
to
be
coming
from
one
particular
company.
So
I
think
that's
a
good
thing.
My
coincidence,
even
recently,
I
saw
it
some
books,
lady
I'm,
talking
about
how
we
can
increase
contribution
to
their
documentation.
A
So
obviously
that's
why
Ben
is
asking
here:
alright,
let's
do
as
a
concerted
community
and
effort
concerted
effort
to
to
help
so
that
pivotal
doesn't
have
to
carry
the
brunt
of
developing
food.
So
thank
you
for
that
then
appreciate
it.
Ok,
so
we're
doing
good
on
time.
So
if
you
guys
have
questions
for
the
previous
discussions,
so
as
we
set
up,
let
me
kind
of
introduce
the
next
one.
I
wanted
to
have
a
lot
of
time
for
him,
because
I
started
talking,
there's
quite
a
bit
he's
gonna
cover
and
there's
also
live
demos.
A
So
here's
to
you
can
take
over
and
you
can
start
sharing
your
screen.
I'll.
Just
briefly
mention
that
the
project
abacus
is
is
you're,
gonna
hear
about
is
a
metering
engine
both
on
foundry.
It
started
at
IBM,
but
SAT
sort
of
took
it
over
and
what
you're
gonna
hear
is
not
only
version
one
and
what's
going
on
there,
but
also
exciting.
What's
going
on
with
version
two
and
I
think
you
might
definitely
want
to
hear
about
that
and
hear
stories
part
of
the
team
in
Bulgaria,
so
he
has
an
entire
team
of
them
here.
A
G
Okay,
so
let
me
start
the
presentation:
I
want
to
talk
about,
as
Mark
said
Ilocos
the
current
state,
the
problems
we
have
and
why
we
actually
want
to
start
version,
two
and
I'm
going
to
talk
about
why
at
all,
using
something
like
a
vocals
while
we
need
it,
what
is
the
current
architecture,
a
short
demonstration
about
what
we
have
now,
then
the
broker
project?
Why
we
started
it
demonstration
on
one
of
our
systems,
then,
as
I
said,
problem,
sent
version
to
how
we
basically
want
to
approach
these
problems.
G
So,
let's
start
with
the
first
thing:
why
abacus?
Well?
Because
when
you
use
your
applications
on
desktop
or
mobile,
you
basically
consume
our
computer
storage
services
and
resources
in
the
code
being
that
Cloud
Foundry
kubernetes
or
yes
whatever,
and
you
need
basically
to
meter
how
much
these
resources
are
and
what
are
the
resources.
So
you
should
be
able
to
feed
some
billing
engine
and
Indian
get
an
invoice
and
monetize
all
these
resources.
So
you
can
benefit
from
this,
and
abacus
is
actually
the
metering
part
of
this
pipeline.
G
So
we,
what
we
do
is
basically
meter
how
much
resources
you
are
using
and
we
are
doing
some
accumulation,
accumulation
and
aggregation.
So,
let's
move
to
some
short
architecture
sites.
There
are
a
lot
of
components
in
abacus
because
we
use
pipeline.
So
there
are
a
lot
of
documents
flowing
into
our
pipeline
and
they're
basically
accepted.
Then
we
do
some
modification.
We
apply
normalization.
For
example,
if
you
want
to
say
to
us
that
there
are
new
database,
that
is
10
gigabytes
and
someone
actually
created
an
instance.
G
G
We
basically
accumulate
across
different
documents
and
get
you
the
current
state
of
the
database.
You
probably
you
already
had
another.
Thank
you.
Thank
you
goodbye
database,
so
we
most
probably
will
sum
this
up
and
we'll
get
current
state
of
20
gigabytes,
and
then
we
basically
aggregate
the
growth
organization
and
spaces.
So,
in
the
end,
what
you
get
out
of
abacus
is
basically
a
report
that
says
how
much
resources
you've
been
using
in
terms
of
organizations
basis
brakes
breakdowns.
G
We
also
care
for
services,
breakdowns
and
things
like
that,
and
since
this
is
quite
complex
structure,
you
can
think
favicon.ico
blackbox
and
it's
basically
fitted
with
usage
by
different
components.
That
can
be
something
quite
current
and
providers.
We
have
two
of
these:
we
bicycle
Futaba
coos
with
data
about
cloud,
foundry
applications
and
co-founder
services.
This
is
somehow
generic
and
comes
with
the
project,
but
there
can
be
service
providers
that
want
some
more
complex
data
and
more
complex
calculations,
so
they
have
their
own
logic
or
probably
run
time
providers.
G
You
can
probably
measure
your
own
stuff
with
have
a
goose
and
get
some
reports
in
the
end.
So
this
is
the
use
user
document
we
have
and
tit
can
be
basically
split
into
several
parts
when
the
resource
consumption
happened,
who
is
actually
using
the
resources?
What
we
are
using,
we
have
three
fields
about
that.
Basically,
that
can
be
database.
We
can
have
the
plan
and
the
service
name
and
how
much
so
this
thing
here,
then
can
be.
G
G
So
let
me
start
a
because
the
things
you
see
here
are
all
these
small
boxes
we
started
locally.
Of
course
you
can
push
it
to
Cloud
Foundry
some
applications
or
you
can
build
ball
deployments
out
of
this,
and
we
have
our
demonstration
application
just
to
be
sure
that
everything
works.
We
also
use
it
for
tests
and
what
we
see
here
is
a
lot
of
stuff,
but
I
want
to
show
you
these
three
lines.
G
Basically,
we
submitted
10
gigabytes
usage,
1080i
calls
and
102
heavy
vehicles,
so
you
can
submit
multiple
measures
and
build
report
out
of
these
things,
and
this
long
conk
JSON
file
is
actually
the
report
with
breakdowns
of
the
different
plants,
abacus
plants,
the
different
spaces,
different
consumers,
which
is
basically
the
applications
and
tindy,
and
we
have
some
numbers
that
aggregated
for
all
these
three
documents.
So
this
is
basically
how
abacus
works
using
me
to
document
and
India,
and
you
get
report
out
of
it.
G
So
this
is
nice,
but
we
recognize
that
we
need
a
bit
more
because,
let's
see
why?
Because
we
had
to
create
our
client
and
plan
basically
saying
how
this
document
should
be
processed,
how
to
convert
from
one
unit
to
the
other,
how
to
make
the
accumulation
aggregation
and
either
you
do
it
manually,
which
is
not
very
nice,
especially
if
you
careful
out
of
these
ones.
Here
the
resource
providers,
code
or
you'll
do
some
automation,
so
we
decided
to
create
a
broker.
The
basically
automates
these
tasks
for
for
us
and
lets
directly
go
to
the
demonstration.
G
G
G
G
So
it's
a
bit
raw
still,
but
we
intend
to
improve
it
a
lot
in
the
next
version
and
what
we
actually
have
is
simple
measure,
and
so
you
can
change
the
measure.
You
can
change
the
metric
names.
You
can
change.
Actually
the
formulas
we
have
for
calculating-
and
we
have
three
formulas
like
metering,
accumulation,
aggregation
and
summarize
I
will
not
go
into
details.
Probably
that
would
be
too
much.
What
I
wanted
basically
to
show
is
that
we
now
have
environment
to
our
application
and
you
can
directly
access
your
account
with
this
credential
and
password.
A
G
Yep,
okay,
so
we
have
it
working
and
from
the
report
you
can
basically
do
a
development
like
change.
These
formulas.
Try
out
different
things
and
the
one
thing
we
are
going
to
try
is
basically
to
submit
document
and
see
if
we
can
read
it
with
our
application,
so
I'm
going
to
submit
it's
automatically
generated.
It
knows
what
the
name,
sir,
if
change
something
it
will
be
changed
as
well
here,
so
we're
going
to
submit
100
API
calls
so
and
I
will
set
that
it
accepted
the
usage
and
done
some
calculations.
If
we
now.
B
G
G
And
I
want
to
post
one
more
time:
okay,
everything's
working,
getting
back
the
report.
We
should
have
200
and
that's
because
we
have
default
for
most
that
simply
do
some,
but
there's
nothing
stopping
you
from
defining
or
complex,
like
multiplications
minimum
maximum.
Whatever
you
basically
want,
so
you
can
do
pretty
complex
stuff
with
abacus.
G
Why
I
said
because
we
want
to
automate,
how
do
we
ultimate?
We,
you
saw
that
we've
done
create
and
bounty
instance
through
the
service
broker
standard
for,
but
behind
that
we
reach
you
a
and
create
all
quant.
So
you
can
get
a
token
and
thought
authenticate
before
have
occurs
and
we
also
create
a
sample
plan
in
abacus.
When
you
do
some
changes
in
this
UI,
we
update
the
plan,
so
we
automate
everything,
so
you
simply
can
submit
your
usage.
G
That's
why
we
want
to
have
this
service
broker
and
also
we
have
a
lot
of
internal
usage.
We
don't
want
to
take
care
about
manual
steps
or
dealing
with
guys
that
simply
want
to
deploy
their
solution
and
start
measuring
it.
So
that's
why
we
have
this
service
broker
and
I
know.
Should
we
leave
time
for
questions
here
or
should
we
continue?
A
G
G
We
have
problems
with
our
databases
because
we
have
really
a
lot
fuse
each
like,
but
minimum
400
requests
per
second
with
a
small
document,
so
we
need
to
aggregate
accumulate
them,
so
the
database
gets
pretty
huge.
We
also
have
problem
with
our
model
and
our
database
document
is
getting
too
big
because
we
keep
a
document
with
aggregation
accumulations.
The
report
can
grow
tremendously.
You
already
saw
that
for
only
three
documents,
it's
to
be
quick
to
screens.
G
G
Not
only
the
data
but
also
the
execution
and
son
guarantee
by
this
transaction
like
support,
and
so
this
somehow
prevents
abacus
from
scaling.
Well
it
we
also
have
to
take
care
of
this
custom
application
level
charting.
So
it's
pretty
hard
to
implement
a
highly
available
system
and
boogly
grain
deploy
not
impossible
but
hurt,
and
that's
why
we
are
having
hard
times
operating
the
solution
and
the
use
it
one.
Well,
we
have
multiple
metrics,
which
basically
means
that
we
can't
really
validate
that
the
document
you're
sending
is
really
consistent.
G
You
might
be
missing
one
or
two
measures
and
we
have
no
way
to
tell
if
this
is
or
if
this
is
intentional,
or
you
just
made
a
mistake,
and
so
we
are
using
a
JavaScript
code
to
basically
enable
great
flexibility,
but
this
comes
with
a
price
we
have
kind
of
sandbox,
but
no
real
sandbox
in
JavaScript,
so
we
are
currently
prone
to
do
now.
Services.
We
are
running
if
we
are
running
untrusted
code.
This
sandbox
we
currently
have
is
all
simply
because
you
are
writing
actual
code.
G
We
somehow
need
to
test
this
to
validate
it,
and
this
means
also
that
it's
pretty
complex
and
sometimes
unpredictable,
and
we
have
the
salt
called
time-based
metrics,
which
are
too
complex,
and
we
don't
really
recommend
them
using
it.
So
that's
another
of
our
big
problems
and
that's
why
we
basically
want
to
start
version
2
somehow
to
solve
all
of
these
problems.
G
G
They
don't
really
want
more
than
maximum
sum
or
division
multiplications,
and
we
created
poking
go
lonk,
two
reasons
for
that.
We
want
to
gather
all
of
our
small
applications
into
one
to
be
able
to
call
co-locate
them,
and
the
other
reason
is
basically
that
we
want
to
start
supporting
Cuba
Knights
as
well,
and
that
goes
pretty
well
with
the
open
service
broker,
where
we
can
get
no
longer
only
organizations
based
but
given
identities.
So
we
started
thinking
in
this
direction
and
we
have
a
proof
of
concept
for
all
this,
it's
created
by
my
colleague
module.
G
We
are
working
internally
with
docker
compose
simply
because
it's
convenient
but
there's
nothing,
stopping
us
pushing
these
to
either
cloud
foundry
or
kubernetes,
and
that's
basically
the
idea
to
be
able
to
move
to
different
either
infrastructure
or
paths.
So
what
we
have
is
three
abacus
applications,
the
collector
that
accept
the
usage
document
service
that
obstructs
the
database.
So
we
can
change
form
our
DBM,
probably
something
else.
Oh,
if
we
need
it
and
a
single
worker
that
basically
does
all
the
work
currently
and.
G
All
of
this
basically
looks
like
this:
we
have
some
clients
submitting
kills
each
collector
accepts
it
puts
it
into
the
document
service
and
stores
it
in
the
database.
Then
we
are
using
crabbit
mq
to
transfer
the
document
to
the
next
in
the
pipeline,
the
worker,
when
the
worker
finishes
the
jobs,
it
saves
it
in
the
document
service.
G
G
And
I
want
to
posts,
so
it
automatically
created
the
dummy
transform
that
does
some
for
accumulation.
It
has
usage
measure
pretty
dummy
name,
but
it
is
okay
for
testing
and
so
for
month
accumulation.
It
also
sums
what
we
are
going
to
send
and
I'm
going
to
need
this
ID.
So
when
I
say
collector
then
use
each
post.
G
G
We
have
one
post,
it
that's
okay
and
we,
if
we
post
a
second
time,
we
should
get
to
so
simple,
some,
not
in
Korea
fancy
and
what
is
actually
different
is
that
so
we
can
now
do
colocation.
We
are
now
from
because
we
use
a
lot
of
techniques
to
get
rid
of
our
problems.
We
are
at
least
six
times
faster.
That's
the
initial
measurements
we
use
really
less
database.
G
We
almost
managed
to
solve
all
of
the
problems
except
the
reporting,
because
the
reporting
is
modeled
against
the
previous
version
of
a
box
version
1,
and
we
want
to
start
capturing
this
problem
as
well
and
the
timelines.
Well,
we
want
still
to
continue
supporting
version
1
because
we
are
productive
with
it.
G
A
Yeah
perfectly
on
time,
so
maybe
we
can
take
a
few
more
minutes
for
questions.
Let
me
say,
as
we
prepare
questions,
this
is
massive
for
people
that
don't
understand.
There's
literally
pilla
bytes
of
data
that
has
to
be
processed
and
I
know
that
in
the
IBM
cloud
I
can
speak
for
that
it
sits.
In
that
level
we
have
to
back
up
the
old
video.
Obviously,
but
if
you're
gonna
church
people
you
gotta,
have
the
data
also
a
way
to
retrieve
it,
because
people
complain
and
we
need
to
be
able
to
prove
it.
A
So
there
are
lots
of
regulations
around
that
just
quickly,
as
here's
to
mention
v2
inceptions
can
happen
soon.
Last
time,
I
chatted
with
yes
I
think
they
wanted
to
do
it
for
March.
So
you
should
see
this
coming.
You
know
March
or
April.
The
team
at
ICP
is
sort
of
distributed,
but
we'll
figure
out
a
good
time.
So
if
you
want
to
be
part
of
it,
any
other
questions
in
the
other
comments.
A
No
all
right
well,
I
guess
it's
recorded!
Thank
you
here.
So
this
is
this
is
this
is
big
and
I
appreciate
your
time
and,
of
course,
then
thank
you
for
this
all
for
documentation,
try
to
submit
a
full
request.
That's
always
a
good
thing
about,
let's
be
better
than
the
rest,
all
right,
so
I'll
talk
to
everybody
again
in
March.
Thank
you
for
joining.
Have
a
good
idea,
so
take
care.
Bye,
just
choose
life.