►
From YouTube: CNCF Serverless Working Group Meeting - 2018-03-01
Description
Join us for KubeCon + CloudNativeCon in Barcelona May 20 - 23, Shanghai June 24 - 26, and San Diego November 18 - 21! Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.
A
A
A
D
A
Gabe,
are
you
online
yet.
B
A
Right
cool,
let's
see,
Chad
I,
think
I
see
you.
So
that's
good
enough.
Let's
see
I'm
s
favio
I've
heard
before
Marky.
There.
G
H
A
Hey
Mark
all
right,
thank
you.
Let's
see
what
else
we
got
David
Lyle
some.
A
Type
Justin
butchering
that
no.
D
A
I'm
here,
okay,
thank
you.
Okay,
I
know:
I'm
gonna
butcher
this
one
yo
Marv
byom,
are
you
there
yeah.
A
A
A
D
A
B
A
Kathy
there
I
guess
not
yet
I
think
we're
just
missing
those
two.
Oh
wait!
A
minute!
Stanley!
Are
you
there
Oh?
Can
you
hear
me?
Yes?
What
is
that
is
that
Stanley
yep
but
Stanley
excellent?
Do
you
move
the
call?
How
am
I
just
not
remembering
you
from
last
time
no
I'm
knew
the
call
was
come
to
us
I'm
with
V
ohm
and
Oracle.
Oh
okay,
yeah
right.
C
A
L
A
A
A
Many
people
typing
all
right,
tell
you
what
we're
gonna
get
started
and
finish
this
up
later
on
all
right,
let's
jump
right
into
it
quickly,
just
to
touch
on
one
of
the
outstanding
action
items
Austin.
Would
you
like
to
talk
to
the
question
or
issue
you
have
around
the
website,
so
we
can
resolve
that
relatively
quickly.
B
Sure
the
Linux
Foundation
has
completed
the
domain
transfer.
They
are
now
the
owners
of
cloud
of
NCO
and
org.
However,
the
DNS
information
got
a
bit
messed
up
in
the
process,
so
those
domains
to
occur
interior
olve
to
anything
and
they
were
previously
linked
to
a
Squarespace
website.
At
the
same
time,
in
the
repository
we've
been
investigating,
just
for
launching
our
own
website
just
hosted
on
github
I'm
thinking,
instead
of
reaching
out
to
the
Linux
Foundation
and
asking
them
to
adjust
the
DNS
information.
B
Instead,
they
should
just
point
it
to
our
github
website
and
a
github
website
is
currently
deployed.
Thanks
to
mr.
Dan,
Barker
did
a
great
job
of
getting
this
up
and
running.
I
think
we
should
take
a
look
at
it.
It
doesn't
have
to
be
perfect
by
any
means,
but
I
think
it's
better
than
nothing
right
now
and
we
could
quickly
iterate
on
the
website
to
to
refine
it
shortly
after
we
just
fixed
the
domain
issue.
B
G
A
So
I
believe
the
overall
question
here
for
the
group
is
everybody
okay
with
using
what
is
a
damn
Barker
in
Barker's?
Current
version
of
the
website
doesn't
start
a
point
quite
forward,
assuming
we'll
just
make
it
eration
'he's,
going
as
usually
progress.
Is
that
right,
awesome,
correct,
okay,
any
questions
or
concerns
about
that
because
somethings
better
than
nothing
you.
M
Know
I
mean
I'm
all
for
like
starting
somewhere
I
just
want
to
point
out
that
we
have
this
consistent.
Almost
every
week,
people
get
confused
between
what
we're
talking
about
in
events
and
like
just
time,
series
data.
Some
people
configure
events
like
message.
Anything
async
on
the
internet
seems
to
be
an
event
and,
and
so
so
I
think
like
starting
with
something
but
like
making
big
noise
about
it.
We
should
figure
out
how
to
clarify
that
in
this
language,
I.
B
A
A
G
A
A
A
B
Thing
we
should
discuss
real
quick
is
I,
think
we'll
be
in
a
good
position
if
we
keep
making
great
progress
to
really
publicize
this
event.
At
cloud
native,
con
Europe
I
will
have
the
website
up.
Hopefully,
we'll
have
like
a
more
solid
version
of
the
specification,
and
if
we're
able
to
reach
reach
those
goals,
would
it
be
more
valuable
to
or
something
we
should
do?
In
addition
to
the
face
to
face,
they'd
have
some
sort
of
open
house.
B
A
A
B
A
Feather,
okay,
tell
you
what
I
will
take
the
action
I
am
then
to
find
out
if
we
can
get
a
birds
of
a
feather,
slacks,
slash,
open
house
kind
of
thing,
the
axe
I
give
such
a
task.
Is
there
any
objection
to
me
going
off
and
see,
but
you
set
up
a
birds
of
a
feather
to
answer
questions
from
the
community.
A
A
The
great
and
well
solidifies
that
could
what
here
is
as
we
get
closer
but
yes,
I,
like
I,
like
the
ax
different
pressure
on
us
to
move
and
get
something
in
place
for
that
conference,
driven
development,
exactly
all
right.
So
so
is
there
any
then
objection
to
me
going
forward
trying
to
get
a
birds
of
a
feather
set
up
a
coupon
and
the
face
to
face
this.
I
Is
Rachel
I?
Don't
have
an
objection
to
the
birds
of
a
feather,
but
I
do
want
to
know
like
what
the
goal
of
the
face
to
face
is
like
I.
Don't
want
to
exclude
people
who
aren't
face
to
face
by
I'm
excited
to
like
meet
more
of
the
groups
from
like
more
the
people
from
this
group.
So
if
we
could
like,
if
we
could
iron
out
what
our
goals
for
the
face
to
face
would
be,
that
would
be
great
for
me.
Yes,.
A
I
agree,
and
that
was
just
what
I
kind
of
was
alluding
to
earlier
when
I
said
at
some
point
before
the
face
to
face
will
have
to
decide,
for
example,
whether
it's
an
official
meeting
where
we
can
actually
prove
pr's
or
not
because
be
honest.
If
we
only
have
seven
people
I'm,
not
sure,
that's
enough
of
a
quorum
to
say
yes,
we're
gonna
approve
PRS.
So
then
we
may
decide.
A
Okay,
the
agenda
is
gonna,
be
let's
see
if
we
can
hash
through
some
of
the
open
issues
at
the
time
to
see
if
we
can
leverage
the
face-to-face
time
to
get
better
design,
discussions
going
and
that
may
be
all
it
is-
is
just
increased.
Communicate.
Are
another
communication
channel
but
not
necessary
an
official
like
a
PR
approving.
D
L
L
B
A
B
I
know
that
there
are
a
lot
of
large
companies
who
have
been
collaborating
on
the
specification
and
I'm
just
curious
how
these
large
companies
would
like
to
characterize
their
involvement
in
this
effort,
not
that
we
have
to
figure
this
out
right
now,
not
that
we're
asking
them
to
officially
endorse
this,
but
if
they
could
give
us
some
some
insight
into
how
you
know
we
should
characterize
their
involvement.
That
would
be.
That
would
be
really
helpful
before
you
know
before
we
do
any
type
of
press
push
around
cloud
native
con.
B
M
D
A
I'm
trying
to
throw
out
the
best
way
to
characterize
that
or
the
next
steps
on
your
suggestion
there
I'm
saying
cuz
I,
think
it's
a
good
one.
Is
that
just
something
I,
don't
think
give
up
issues
appropriate
to
track
that
necessarily.
It
may
be
more
of
a
question
of
reminding
it
representives
on
the
call
to
sort
of
put
forward
some
sort
of
statement
or
something
that
would
save
in
terms
of
what
they
feel
comfortable
people
saying
about
their
involvement.
And
then
we
need
you
to
sort
of
gather
that
someplace
and
like
the
Google
Doc.
M
A
very
specific
proposal
in
the
contributors
list
that
seems
to
be
like
have
gotten
a
lot
of
positive
responses
and
I
didn't
like
relevant
for
Lisa.
Finding
that,
because
I
wasn't
sure
I
mean
it
just
seemed
to
be
lower
priority
than
other
things.
But
that
could
be
my
intent
with
that
was
so
that,
instead
of
just
being
random
people
on
slack
who
may
or
may
not
match
up
with
github
names,
it
was.
M
It
was
sort
of
helpful
to
understand
what
people
were
doing
with
the
spec,
which
helps
sort
of
understand
people's
comments
and
and
create,
like
shared
understanding
of
what
we're
all
trying
to
achieve,
and
so
I
pulled
that
contributors
list
initially
from
the
doc,
but
that
Austin
had
put
together
and
then
and
then
people
added
in.
Like
oh,
please,
add
me,
and
so
that
could
be
a
way
that
you
know.
Then
people
can
do
PRS
if
they
want
to
update
how
the
context
for
their
contributions
so.
A
I'm,
not
okay,
so
I
think
there's
two
different
discussions.
There
there's
the
discussion
of
how
do
we
list
contributors
or
authors
of
the
specification
I
think
that's
one
discussion
and
I
thought
that's
what
your
PR
was
more
focused
on.
This
second
part
of
the
discussion
seems
to
be
more
about
how
are
people
going
to
use
a
spec
or
how
do
they
want
the
participation
as
denoted,
or
you
know
when
one
of
the
one
said
about
their
participation
in
the
working
group
and
I,
don't
think
that's
appropriate
to
go
into
the
spec
itself.
M
It's
in
the
spectrum
posit
or
e
to
give
context
for
people
who
are
so
there's
different
kinds
of
contributions
suspect
like
it's,
not
all
wordsmithing
right.
All
the
contributions
don't
necessarily
come
in
a
form
of
a
give,
a
pull
request,
not
to
mention
that
some
of
the
contributions
came
before
this
repo
existed,
and
so
the
idea
is
that
people
could
contribute
a
statement
about
their
planned
usage
of
the
SPAC
right
as
a
way
of
contributing
to
the
effort
right
and
that
that
contributed.
A
I
suggest
that
we
open
an
issue.
That's
up
to
discuss
this
too
I,
don't
want
to
necessarily
dive
too
deep
on
this
one
right
here,
but
I
think
Austin
your
original
proposal
about
trying
to
figure
out
what
each
company
wants
to
say
about
the
participation
is
a
good
one.
We
just
need
to
figure
out
the
best
way
to
gather
that
information
and
then
have
it
someplace
available,
so
people
won't
be
making
stuff
up
about
other
companies
I.
M
Think
you
have
a
particular
concern
about
pyaare
around
the
event,
and
maybe
you
could
add
that
to
the
roadmap,
because
I
I
think
that
we
need
to
hit
a
milestone
first
and
then
have
that
conversation.
So
maybe,
if
it's
on
the
roadmap
at
an
appropriate
milestone,
that
might
be
a
good
way
to
address
that.
B
Yeah
absolutely
and
the
roadmap
doesn't
have
any
dates
on
it,
and
this
is
definitely
a
date
focused
thing,
but
but
let
me
take
it
as
an
action
item.
I
think
you
brought
up
actually
a
good
point.
Sarah
I
believe
that
there
is
a
there's,
a
middle
ground
here
for
four
of
these
companies,
and
then
it's
like
something
in
between
there's
something
less
than
official
endorsement.
B
They
could
it's
credible
more
of
a
casual
acknowledgement
that
they're
they're
collaborating
on
this
effort
and
hopefully
their
respective
companies
get
some
points
for
being
you
know,
community
focus,
good
community
focus
citizens,
but
I
think
you
know
what
I
characterize
this
in
the
context
of
how
do
we
do?
How
do
we
describe
this
within
the
context
of
PR,
around
Cube,
Khan
or
cloud
native
Khan
I?
Think
this
actually
just
touches
on
the
larger
issue
that
that
contributes
list
is
meant
to
solve.
It's
like
what.
B
How
do
we
characterize
these
contributors
and
how
could
we
do
this
in
a
way,
that's
kind
of
informal
and
not
perceived
as
an
official
endorsement
I
think
if
we
could
just
solve
that
problem,
then
we
could
kind
of
use
that
characterization
when
we
go
to
take
this
to
the
press,
and
that
is
hey.
These
are
contributors
they're,
not
officially
endorsing
this
project.
We've
just
agreed
to
informally
kind
of
collaborate
on
this
and
and
see
where
it
goes.
B
A
N
N
B
A
A
A
A
good
question-
I
guess
we
could
do-
is
do
a
PR.
Where
oh
wait
a
minute.
We
don't
have
to
have
that
in
here.
I'll
tell
you
what
but
I
was
gonna
say
I
got
another
PR
to
modify
the
reading
a
little
to
say,
not
uploading
things,
here's
a
spot
to
add
links
or
references
to
places
someplace.
You
know
someplace
else.
That
way.
People
can
do
a
PR
against
to
read
me
to
add
their
stuff.
Yeah.
N
A
A
All
right
cool,
so
easy
ones
done.
Thank
you
guys
very
much
all
right
so
back
to
the
presentations
or
discussions
around
what
is
source
now
on
last
week's
call.
We
ended
with
Iran
finished
up
this
presentation,
but
because
we
ran
out
of
time,
I
put
a
little
bad
that
we
didn't
really
get
a
chance
to
ask
any
questions
in
case
they
had
any.
B
My
comment
is
that
I
really
enjoyed
the
design
goals
that
sarah
has
contributed
and
I
do
believe.
As
we
look
at
it
aspect
of
this
effort
at
first,
we
should
settle
on
those
design
goals
and
that
will
really
kind
of
help,
rein
in
our
focus
and
kind
of
get
aligned,
and
then
look
at
these
all
these
other
presentations
and
descriptions
of
things
just
from
a
similar
perspective.
Okay,.
A
K
Think
I
relate
the
discussion,
is
sort
of
what
are
we
really
defining
because
are
we
defining
the
event
or
served
the
way
to
transfer
an
event
where
which
doesn't
need
to
understand
the
event
itself
and
I?
Think
that
leads
to
the
points
about
source
and
source,
ID
and
event
type
etcetera,
because
some
of
the
examples
that
contain
just
as
an
example
the
bucket
name
and
the
file
name
that
was
sent
in
the
s3,
you
know
event
as
a
source.
I
would
categorize
something
like
that
as
specific
to
a
class
of
event.
K
You
know
because
there
may
be
another
event
that
doesn't
relate
to
an
s3
bucket
and
a
file,
so
I
think
we.
The
first
thing
we
need
to
define
is:
what
are
we
trying
to
define?
Is
it
a
way
to
transfer
events,
or
are
we
trying
to
define
the
events
themselves
because
I
don't
think
there
is
enough
clarity.
I
would
like
to
minimize
the
amount
of
metadata
that
we're
sort
of
putting
in
context
and
anything
that
sort
of
events
specific
will
be
defined
in
the
event
in
the
event,
schema.
Okay,.
A
A
What
is
source
first
then
get
to
Sarah's
draft
design
goals,
and
we
then
we
could
have
a
discussion
about
what
you
just
talked
about
your
on
is
you
know,
are
we
talking
about
just
the
shape
of
the
event
or
other
things
around
it
like
transmitting
the
event,
because
that
isn't
topic
on
the
agenda
and
I
want
to
get
to
that
just
in
a
timely
fashion.
So
first
focus
just
on
my
original
question.
A
A
We're
actually
I
understand,
finish
Sarah
I'm,
not
trying
to
write
to
sidetrack
that
it's
just.
We
have
to
have
time
for
everybody
to
give
the
presentation
on
what
is
source
and
so
I
want
to
focus
on.
Are
there
any
questions
for
related
to
your
ons
proposal
so
that
we
can
move
forward
and
get
to
Cathy
and
your
proposal
or
your
presentations?
This.
I
I
A
I
think
I
think
everybody
agreed
to
need
to
define
what
we're
defining
I
think
that's
true,
right,
ok,
so
so
so
I'm
not
hearing
anybody
to
say
raise
any
questions
for
your
on
based
upon
its
presentation.
So
let's
move
forward
now,
so
we
can
get
through
the
other
two
presentations
that
we
have
on
tart
are
on
schedule.
So
Kathy
you
want
to
walk
through
your
presentation.
Yes,.
L
L
Okay,
so
just
a
second.
L
L
Yeah,
okay,
so
can
you
see
it
now.
L
Well,
I
think
you
know.
I
just
gave
an
example
of
you
know
the
situation.
You
know
how
we
should
define
the
source
on
so
here's
an
example
of
the
burglary
detection
example.
It's
a
hole
monitoring
case
like,
for
example,
there
would
be
two
events
like
one
event
is
like
a
sensor
detect
a
motion
and
another
event
is
or
window
opening
event,
and
this
UN
will
go
through.
L
You
know,
for
example,
the
motion
event
with
sugar,
maybe
a
picture
or
video
clip
to
be
stored
into
the
cloud
storage
and
then
that
will
trigger
a
notification
to
a
service
platform
which
will
trigger
a
function,
so
this
function
to
do
the
very
analysis.
Similarly,
similarly,
you
know
the
window
opening
event
or
trigger
maybe
a
cloud
messaging
system
to
send
another
message
to
the
another
event
to
the
service
platform,
which
will
also
trigger
a
function.
So
now
we
have
this.
L
So
this
this
analysis
will
based
on
the
two
events,
but
we
can
see
that
the
event
you
know
originally
originated
from
a
home
sensor
and
there
you
know
from
a
Soviet
platform
point
of
view.
It
looks
like
it's
originated
from
cloud
storage
or
the
cloud
messaging
system,
and
so
here
is
you
know
it
could
also.
The
event
from
the
regional
originated
inside
originating
source
could
also
travels
through
some
middleware
device,
so
how
we
should
define
the
event
source
and
source
time
in
society
and
even
type
I.
Think.
L
Currently,
our
specification
is
not
very
clear
about
this
I'm
not
going
to
propose
just
you
know,
say
what
is
a
source,
and
this
is
open
for
discussion.
You
know,
should
we
put
this
or
should
we
define
the
source
as
the
sensor
from
the
you
know,
for
example,
in
this
case
at
the
home,
or
should
we
define
the
source
as
a
the
cloud,
storage
or
cloud
messaging
system,
which
you
know
interface
directly
with
a
service
platform?
That's
my
question.
L
I,
don't
know
what
other
people
think
I'm
thinking
in
me,
I
think
in
the
sauce
rabida
from
a
service
platform.
Point
of
view
should
be
the
cloud
storage
or
the
cloud
messaging
system,
not
any
middleware
devices
or
the
sort
at
the
information.
The
event
information
itself
should
have
a
way
to
to
identify
the
like.
You
know
the
how
to
sate
the
original
original
event
original.
The
sensor
information,
sir
information,
their
sensor
data
there-
that's
my
thought,
I,
don't
know
what
other
people
think
about
this.
K
Yeah
I
agree
that
what
I
was
saying
before
is
that,
if
we're
starting
to
go
inside
into
the
sensor
itself,
each
sensor
will
have
a
different
meta
data
and
it
will
be
very
ambiguous
and
will
take
us
five
years
to
define
it.
I
think
we
need
to
just
define
sorry
what
I
call
the
envelope
which
is.
There
is
the
guy
reporting
the
event
and
here's
the
schema.
K
L
L
L
Action
or
you
know,
you
know
the
source
will
be:
oh,
it's
the
sensors
or
whatever
the
region,
and
that
really
you
know,
originated
inside,
but
it
one
current.
The
description
of
our
spec
I
think
clear.
You
know
different
people
can
interpret
it
different
way.
That's
why
I
think
we
have
so
many
so
much
discussion
discussions.
You
know
which
you
know,
people
are
you
know
kind
of
like
and
not
sync
up
very
well
also,
there's
a
sauce
hype,
but
that
depends
how
we
define
the
event
source.
L
L
So
as
I
think
we
are
talking
about
this
event,
source
or
source
type,
or
they
were
from
Soviets
point
of
view-
might
a
case.
The
events
or
should
be
the
like
in
this
example,
should
be
the
cloud
storage
or
the
cloud
messaging
system,
which
you
know
interact
directly
with
a
service
platform
to
send
us
I
mean
to
send
that
sugar
our
event
to
that
service
platform.
If
we
go
this
way,
then
you
know
the
source
type
we
can
define,
you
know
different
source
type
right,
there's
a
story,
source
type.
L
L
K
L
L
The
other,
that's
fine,
you
know
it's
nice
way,
you
know,
but
you
can
have
you
know
so
sauce
time.
Out
of
aside
a
schema
to
you
know,
there
is
ample
you
different
source
type
for
record
for
storage.
The
storage
is,
don't
have
different,
schema,
different
fields
or
different
metadata
information.
Right
frizam,
it's
going
to
happen.
It
may
be
column,
but
if,
like
Imam.
K
M
Of
chime
in
with
a
clarifying
question,
I
think
in
this
picture,
there's
three
different
or
potentially
four
different
sources
and
it
for
me.
From
my
perspective,
cloud
storage
is
a
very
different
kind
of
thing
than
a
messaging
system,
whereas
a
messaging
system
seems
like
it's
transporting
an
event
from
here
to
there.
Whereas
a
storage
is
like
okay,
the
sensor
generated
an
image
and
the
storage
is
saying:
Oh
am
I,
gonna
create
that
image
am
I
like
replacing
and
like.
M
There
may
be,
like
storage
specific,
like
events
that
have
to
do
with
that
object
being
created
in
storage,
which
could
be
then
a
different
semantic
event
than
the
original
like
I
have
a
sensor.
That
did
a
thing
right.
It's
not
the
same
like
transporting
a
message
from
here
to
there.
In
my
view,
right.
K
L
Yeah
sorry
I
think
you'll
clarify
some
of
my
thoughts.
Yeah
I
think
the
look
if,
if
we
say
the
source
is
cloud
storage
or
comma
City
system
or
any
other,
you
know
system
which
interact
directly
with
a
service
platform.
I
think
different
sources
are
going
to
when
they
send
the
event
to
the
service
platform
that
you,
when
event,
will
have
very
different
information
carried
in
it,
and
you
know
different
vendor
properties.
L
Different
vendors
probably
will
have
different
different
ways
of
organizing
those
information
or
those
yeah,
I,
say
information
associate
with
that
event
and
okay,
so
the
different
representation.
That's
another,
that's
a
separate
issue,
but
I
think
you
know.
The
key
issue
is
different
sources.
The
the
information
carried
for
in
the
Army
from
those
carried
in
that
event
in
those
events
will
be
different.
L
M
L
Yeah,
you
can
say
that
way,
but
sometimes
some,
but
you
know
it
depends.
You
know
some
messaging
system.
For
example,
if
there's
a
window
open
event
right
the
from
that
sensor,
the
information
or
the
format
of
that
event
could
be
different
from
the
event
or
that
sent
from
this
cloud
messaging
system,
because
the
chromaticity
system
has
its
own
to
say,
has
his
own
way
of
you
know,
define
of
you
know
deciding
what
information
to
be
put
into
that
event,
and
what
is
the
format
of
that
event?
It
would
be,
could
be
different
right.
M
J
I
chime
in
on
something
here
really
entry
have
something
interesting
here
that
I'd
like
to
discuss,
just
because
the
messaging
system
is
actually
a
good
example
here,
cuz
you're
right
that
yeah,
it's
usually
just
a
transporter
thing,
but
sometimes
it
does
raise.
Events
like
we
have
a
messaging
service,
it's
a
lot
like
Kafka
and
we
let
you
put
a
time
and
window
trigger
on
it.
Where
will
write
events
to
storage
rather
than
you
having
to
read
a
stream,
and
so
that's
funny
because
it
actually
ties
these
two
together.
J
K
There's
a
way
to
clarify
this
saying:
the
source
of
the
event
is
they're
sort
of
the
resource.
They
reported
the
event.
Okay,
because
the
messaging
thing
is,
you
know,
just
think
about
email,
your
game,
the
two
in
the
email
is
not
your
email
server
is
the
guy
that
sent
the
email,
which
is
the
most
basic
messaging
system.
But
if
the
email
system
is
sending
you
a
notification,
then
it's
becoming
the
the
resource
that
reported
an
event
yeah.
J
I
think
that's
correct,
I
mean
that's.
What
I'm
trying
to
say
it
with
ours
is
our
intention
on
our
event,
for
that
is
you're.
Not
thinking
about
I
mean
you're
interested
in
the
storage
that
was
created,
but
you
do
want
to
know
hey.
This
actually
came
from
the
messaging
system
and
yes,
there's
storage
information
in
the
data,
but
that's
not
really
relevant
to
yeah.
M
I
think
why
are
me
bring
up
a
really
interesting
point,
which
is
that,
like
it
could
be
that
the
storage
system
is
it
in
emitting
events,
but
that
most
people
who
are
using
your
whole
system
most
of
developers,
would
ignore.
That
would
not
listen
to
the
storage
events,
because
those
are
really
like
under
the
hood
and
not
relevant
to
the
like
the
semantic
meaning
of
the
events
and
so
and.
A
So
so,
just
that
of
just
trying
to
watch
the
clock
here
so
I
can
go.
There
are
two
big
things
I
want
to
get
to
on.
Today's
call
so
Kathy
is
there.
Is
there
anything
really
big
that
you'd
like
to
to
say
cuz?
We
kind
of
need
to
wrap
this
up
and
let
Sarah
and
then
Thomas
do
their
their
their
pitch.
Okay,.
L
I
think
it
looks
like
we
agree
on
Dino
the
source.
The
event
source
will
be
the
the
system
that
you
know
sand
directly
interact
with
a
service
platform
which
send
the
event
to
that
service
prophet,
whether
it's
a
storage
or
messaging
or
is
other.
You
know
that
many
other
you
know
system
to
interact
with
the
service
powerful.
H
F
A
L
So
my
point
is,
you
know,
I
think
the
source.
You
know
it
should
be
the
component
that
interact
directly
with
the
service
Pavel
yeah.
A
C
L
So
yeah,
that's
a
good
question.
I
think
he
knows.
So
that's
what
I
said
you
know
before
I
mentioned
before.
So
if
we
define
the
source
as
a
cloud,
storage
and
messenger
system
in
this
example,
right
other
example
will
be
other
things,
but
you
know,
but
that
event
should
have
some
metadata
which
specify
the
original
information
of
the
sensor
or
the
opposite
of
the
yeah.
The
motion
sensor
or
the
door-open
sensor
so
I
have
some
other
slides
which
bring
up
another.
A
C
A
A
So
then,
so
what
I
want
to
do
Sarah
is:
have
you
guys
try
to
wrap
this
part
of
the
of
the
discussion
up
at
the
within
ten
minutes?
So
five
minutes
left
only
for
the
sole
purpose,
then
of
saying
Sarah,
you
have
a
PR
out
there
about
the
design
goals
and
like
everybody,
to
review
it's
because
next
call
we're
gonna
dive
deep
into
that
I
would.
M
A
M
Can
you
see
my
screen?
Yes,
we
can
see
I'm.
This
is
just
a
kind
of
a
very
high-level
how
Google
cloud
functions
fits
into
this
world.
You
have
a
in
this
example.
A
developer
calls
the
Google
API.
It
is
to
do
what
we
colloquially
call
trigger
configuration
right,
so
you
say
basically
say
I'm
interested
in
a
fire
store
document,
create
message
and
I'm
want
to
connect
it
to
my
action.
In
this
case
it
was
a
plot
function
that
will
do
post
to
slack.
M
That
would
then
and
writes
to
a
very
specific
path
that
would
then
trigger
a
Google
Cloud
function
that
gets
that
notification
in
that
data.
So
in
this
example,
the
source
is
fire
store,
which
is
a
Google
service,
and
there
are
many
of
these
and
then
the
action
is
functions
and
we
think
of
the
event
as
just
describing
what
happened
in
fire
store
and
that
that
data
changing
and
then
it
gets
transmitted
to
the
action.
E
The
it's
similar
to
Cathy's,
but
it
looks
like
you
know:
you
have
an
originating
source
coming
from
maybe
a
cell
phone
or
some
server
to
the
fire
store.
So
so
they
could
be
other
upstream
events
that
are
actually
triggering
the
commit
to
of
the
the
document
create
in
fire
store.
Yes,
I'm
sorry
seems.
M
Could
be
modeled
that
way,
our
proposal
is
that
we
consent
be
at
these
I
mean,
and
in
fact
our
current
implementation
is
that
these
Google
service
is
the
source,
because
what
the
event
that
we
are
transmitting
is
we
have
done
a
database
mutation
which
is
different
from
a
user
clicked
a
button
or
a
user
spoke
into
their
phone.
That
would
be
a
very
different
event
and
so
I
I
did
prepare.
M
Like
sort
of
this
contrasts
the
common
eventing
architecture,
which
is
sort
of
loosely
based
on
your
insight,
your
own
styie
graham's,
where
the
source
goes
through
some
kind
of
relay
gateway
to
an
action.
So
if
we
wanted
to
consider
the
source
being
user
spoke
into
their
phone,
then
we
could
treat
fire
sores
or
relay.
M
However,
that's
not
how
we've
modeled
the
the
event
and,
given
that
we're
modeling
the
event
is
a
database
mutation
that
leads
us
to
say
the
source
is
the
Google
service
that
does
that
transaction,
and
so
so,
basically,
I
think
there's
sort
of
two
architectures
that
were
mostly
Congress
discussing
one
is
the
sort
of
IOT
case
or
the
case
where
you've
got
some
remote
device
where
it
relays,
through
a
number
of
things,
to
the
action,
and
then
we've
got.
This
is
the
the
other
event
in
architecture.
K
Two
diagrams
are
don't
really
conflict
from
my
perspective,
because
what
I
refer
to
as
a
real,
a
sort
of
a
transparent,
really
even
your
message
probably
goes
with
some
message:
queue
or
you
know,
lambda
asynchronous
messages
under
the
hood
go
through
sqs
or
something
like
that,
so
that
relay
is
essentially
not
an
entity.
You
know
unless
it
report
it
wants
to
serve
proxy
and
event.
Yeah.
M
K
In
general,
but
if
you
look
at
all
the
discussion,
I
think
we
roughly
have
a
consensus
about
it.
The
only
thing
is
to
try
and
think
put
good
things
that
are
belong
belong
to
the
actual
event
description,
like
you
know
the
bucket
or
the
file
name
in
the
event
itself
and
trying
to
minimize
the
amount
of
metadata,
because
if
we
open
the
door
for
more
metadata,
everyone
will
come
with
a
new
idea
for
another
certain
metadata.
So.
M
M
What
I'm
saying
is
I
believe
these
are
all
the
same
thing,
which
is
that
this
that
the
source
defines
itself
by
the
events.
It
emits
right.
So
in
this
picture,
because
the
event
is
a
database
mutation,
the
source
is
firestorm
and
then
I
Oh
T
case
right,
like
whatever
the
event
is,
determines
its
source,
and
you
may
have
like.
What's
not
pictured
in
this
first
thing,
is
that
the
heavy
green
arrow
from
source
to
action
actually
has
a
transport
in
between.
F
Okay,
so
follow-on
question,
so
you
have
a
have
the
the
phone
cause,
a
message
into
so
the
phone
something
happens
on
the
phone,
and
now
you
post
that
into
into
your
firebase.
Are
you
saying
that
the
the
event
that
is
now
causing
the
action
is
is
happening
because
of
firebase
and
your
effect
obscuring
the
original
event?
So
we
have
a
native
mobile
app.
M
Where
the
user
clicks
a
button,
if
there's
like
some
kind
of
like
graphical
user
interface,
events
that
I'm
suggesting
is
not
part
of
what
we're
talking
about
and
then
the
mobile
app
makes
an
API
call
to
fire
store,
which
is
a
cloud
product
and
that
API
you
could
call
that
an
event
but
I'm
saying
for
this
case.
We
are
not
modeling
that
as
an
event,
that
is
a
a
choice
of
the
disease
of
this
system.
Okay,.
D
M
M
Saying
that
the
I
could
draw
a
different
picture
right,
that
would
be
more
like
this,
yes
and
then
relay
would
be
owned
and
then
I
would
have
a
different
set
of
events,
and
we
may
do
that
in
the
fullness
of
time,
but
I
don't
think
that
we
are
going
to
the
extent
of
modeling
on
the
click
as
a
cloud
of
it.
Okay,.
A
So
Sarah
I
think
we
got
to
call
time
on
this,
just
because
I
don't
want
to
run
out
of
time
for
you
to
do
a
quick
introduction
of
your
design
goals
so
obviously
that
we're
gonna
continue.
The
discussion
is
about
source
and
stuff,
so
you
want
to
share
your
screen
for
your
design
goals,
PR
and
just
quickly
introduce
that
before
we
run
out
of
time,
sure
thing
all
right
go
forth,
so.
M
What
I
think
that
we've
been
talking
about
with
a
few
thank
you
Clements
and
Doug
for
chiming
in
on
different
things,
but
the
idea
being
that
this
is
not
events
are
often
what
I'm
trying
to
do
is
disambiguate.
What
we're
talking
about
from
in
the
industry.
People
talk
about
asynchronous
messages
as
events.
This
is
more
specific
than
any
messaging
system.
People
also
talk
about
some
time
series
data,
log
events
right
anything
logged
any
time
series
data
any
metric
is
an
event.
M
A
A
M
So
I'm
suggesting
that
the
event
and
what
I
think
like
so
this
design
goal
is
attempting
to
communicate
that
the
event
does
not
know
anything
about
how
it
would
get
to
its
destination.
So
it
would
not
include
the
topic.
However,
it
should
be
able
to
interoperate
it
in
a
system
where
you
could
have
a
rule
that
specifies
you
need
this
topic
to
get
to
this
destination
or
whatever
your
system
requires
to
route.
That
event
would
then
be
outside
of
the
event
and
communicated
to
the
source
in
some
way.
Okay,.
M
I
think
that
that
what
the
other
part
of
this
PR
and
thanks
for
asking
that
is
the
read
main,
where
I've
said
that
the
process
right
that
we're
first
we're
doing
the
common
attributes
and
then
the
architectures
and
then
how
the
events
are
transported.
So
my
current
thinking
is
that
that
comes
into
the
house.
Events
are
transporting
the
message
envelope,
but
the
key
thing
is-
and
we
may
need
to
address
some
of
these
in
parallel
in
order
to
separate
the
transport
from
the
event.
A
Hey
I'm
gonna
have
to
call
time
because
we
are
at
the
top
of
the
hour
and
I
know.
People
have
to
sum
over
other
calls,
but
next
week
we're
gonna
continue
both
these
top
hot
discussions
source
as
well
as
design
goals,
probably
design
goals
first
and
its
respective
source.
But
please
take
a
look
at
at
the
PRS
and
be
thinking
about
some
of
the
existing
PRS
out
there
for
how
to
redesign
the
source
stuff.
But
before
we
adjourn
I
wanted
to
do
for.