►
From YouTube: Birds of a Feather (BoF): SODA: The Path To Data Autonomy - Steven Tan & Anjaneya "Reddy" Chagam
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
Birds of a Feather (BoF): SODA: The Path To Data Autonomy - Steven Tan, Futurewei & Anjaneya "Reddy" Chagam, Intel
Data autonomy is the ability to control data anywhere, anytime. It is about storing, running and managing data for cloud native, virtualization, and legacy environments both on-premise and in the cloud. The SODA Foundation is embarking on a mission to deliver open data autonomy for end users. To tackle this formidable task, SODA has brought together a community of global innovators to collaborate and contribute to this open source project.
https://sched.co/VnU1
A
B
A
Okay,
if
you
look
at
what
is
the
problem
that
we
are
trying
to
solve?
Typically
in
most
of
the
enterprise
environments,
when
you
look
at
the
storage,
you
normally
see
islands
of
storage.
Maybe
you
have
an
open
stack,
based,
storage
stack.
Maybe
you
have
a
database.
You
know
big
data,
related
storage
stack.
Maybe
you
have
enterprise
line
of
business.
You
know
databases,
you
know
in
a
different
infrastructure,
so
you
typically
see
a
siloed
infrastructure
as
well
as
siloed
data
management.
A
You
know,
solutions
on
top
is
what
you
normally
see
in
a
typical.
You
know
deployment
now,
ideally
what
you
want.
This
is
the
feedback
from
end
customers.
Is
you
need
to
be
able
to
store
using
a
common
data
management
platform?
You
need
to
be
able
to
run
different
types
of
workloads.
These
could
be
analytics
workloads.
You
know
database
workloads,
you
know
the
streaming
workloads.
You
need
to
be
able
to
store
data
in
a
very
seamless
fashion.
You
need
to
be
able
to
run
the
workloads
different
types
of
workloads
in
that
data.
A
Using
that
data
and
you
need
to
be
able
to
you
know,
provide
the
storage
to
different
types
of
infrastructure
environments.
It
could
be
virtualized,
it
could
be
containerized
or
it
could
be
traditional
bare-metal
type
of
you
know
deployment.
That's
essentially
what
you
want
to
do.
For
that
to
happen,
you
do
need
a
common
data
management
platform.
This
data
management
platform
need
to
bring
in
different
islands
of
storage.
A
Implementations
together,
you
need
to
be
able,
to,
you
know,
have
the
integration,
whether
it
is
cloud
storage
or
on
tram.
You
need
to
be
able
to
integrate
those
two
pieces
as
well.
You
need
to
have
a
way
to
expose
storage
to
different
workloads
and
frameworks.
It
could
be
AI
frameworks,
it
could
be
compute
frameworks
like
contains
containerized.
You
know
infrastructure
and
you
need
to
be
able
to
actually
bring
all
these
pieces
together,
both
from
a
storage
perspective,
workload
perspective
as
well
as
consumers,
which
are
really
the
you
know,
the
compute
side
of
the
infrastructure.
A
Now,
let's
look
at
the
kubernetes
today,
even
if
you
look
at
you
know,
forget
about
the
OpenStack
forget
about
VMware
infrastructure
and
bare
metal.
Even
if
you
look
at
the
cuber,
it
is
infrastructure.
Today,
what
do
you
normally
see?
You
normally
see
different
storage
backends
that
can
get
integrated
with
through
CSI
driver
plug-ins.
A
However,
when
it
comes
to
actually
looking
at
some
of
the
common
storage,
you
know,
functionality
like
can
I
take,
snaps
can
I
take
clones,
can
I
actually
on
board
the
volumes
that
are
actually
sitting
in.
Let's
say
in
public
cloud
that
is
not
Kuban.
It
is
implemented,
aware,
type
of
storage
volumes.
How
do
you
actually
ingest
them?
How
do
you
actually
do
automated
tearing?
How
do
you
do
backup?
You
know
you
can
take
snaps,
you
can
store
it
on
on
tram,
maybe
options
after
certain
amount
of
time.
You
want
to
put
it
in
a
cloud.
A
A
Typically,
when
you
look
at
the
data
management
pipeline,
you
have
the
ingestion
once
you
do
the
ingestion,
you
go
through
certain
amount
of
transformation
through
data
engineering
effort,
and
then
you
essentially
consume
the
data.
Let's
say
it
is
data
analytics
as
one
of
the
you
know,
consumer
parts
of
the
pipeline.
So
if
you
look
at
the
underlying
technologies,
there
are
wide
variety
of
technologies.
A
Even
if
you
look
at
just
the
ETL
portion,
you
could
be
just
you
know:
storing
the
unstruck
streaming,
the
unstructured
data,
it
could
be
structured,
sequel,
type
of
data
or
it
could
be
object
type
of
data.
You
should
be
able
to
actually
consume
different
types
of
data
and
you
need
to
be
able
to
transform
do
the
data
cleansing
and
all
that
stuff
and
you
need
to
be
able
to.
You
know,
store
it
in
a
very
unified
way
and
then
consume
that
that's
really
the
data
management
pipeline
that
we
are
looking
at
from
an
implementation
viewpoint.
A
So
the
reference
architecture
for
the
data
management
platform
that
we
plan
to
do
through
this
open
source
community
effort
called
soda,
is,
if
you
look
at
the
bottom
side,
you
essentially
have
different
storage
backends,
whether
it
is
a
scale
out
storage
like
open
source
implementations
like
Mineo,
SEF
or
whether
it
is
scale
up.
You
know,
proprietary
storage
tags.
We
should
be
able
to
manage
them
using
the
driver
frameworks.
That's
a
key
theme
and
the
core
foundation
services
include.
How
do
you
stitch
together
profile
based
consuming
storage
resources?
A
So
you
could
be
consuming
the
storage
resources
based
on
performance
characteristics
based
on
you
know,
capacity,
characteristics,
all
types
of
characteristics
that
you
can
actually
stitch
together
through
profiles
and
you
need
to
be
able
to
consume
them.
You
know
through
the
profile
based
workflow,
orchestration
and
policy
based
you
know
mechanism,
and
then
we
need
to
have
plugins
for
different
orchestrations
tax,
just
not
cuban.
It
is
only
but
you
need
to
be
able
to
have
plugin
for
vmware.
So
if
you
have
a
storage
back-end,
if
it
is
compatible
with
soda
plug-in,
it
should
work
with
OpenStack.
A
It
should
work
with
VMware
and
it
should
work
with
kubernetes
containerized
platform.
That's
really
the
intent
of
what
we
are
trying
to
do
this
through
this
unified
data
management
platform.
At
the
same
time,
we
want
to
be
able
to
support
some
of
the
emerging
use.
Cases
like
edge
edge
use
cases
are
dramatically
different
where
you
want
to
collect
the
data,
but
normally
when
you
collect
the
data,
you
are
not
actually
storing
all
the
data
into
the
core
data
center
back-end
storage
infrastructure,
but
rather
you
are
collecting
the
data
you
are
doing.
A
Some
sort
of
you
know
massaging
the
data
and
then
shipping
some
portion
of
the
data
back
into
the
core
back-end.
So
we
need
to
be
able
to
support
the
edge
data
use
cases
and
we
need
to
be
able
to
support
multi
cloud
integration
as
well,
so
I
should
be
able
to
actually
take
a
snapshot
of
on-prem,
scale-out
storage,
backed
content
and
I
need
to
be
able
to
back
it
up
into
the
public
cloud
and
I
need
to
be
able
to
do
that
through
some
sort
of
policy.
A
You
know
notion
where
I
want
to
be
back
it
up
on.
At
a
regular
interval,
maybe
once
in
a
month
and
into
a
cloud
of
my
choice,
so
you
should
be
able
to
describe
those
things
on
the
data
lifecycle,
management
through
policies
and
it
needs
the
framework
underlying
framework
need
to
be
able
to
execute
orthogonal
to
what
public
cloud
that
you
are
choosing
and
how
you
need
to
back
it
up.
So
the
implementation
is
abstracted
completely
in
the
policy
based.
A
You
know,
plumbing
is
what
the
operators
will
end
up
using
to
be
able
to
do
these
things
more
seamlessly,
with
that
I
will
transition
it
to
micro
presenter.
Who
is
going
to
talk
about
the
community,
the
roadmap,
all
the
events
that
are
happening
to
give
you
a
flavor
of
what
else
is
going
on
in
this
community.
B
Okay,
so
this
the
project
that
ready
has
been
talking
about
so
the
foundation,
this
is
actually
a
project
that
has
not
been
launched
it.
So
this
project
is
going
to
be
launched
in
the
first
quarter
of
next
year
and
this
project
is
actually
a
project,
that's
being
transformed
from
open
SDS,
so
open
SDS
is
a
Linux
Foundation
project,
also
a
Linux
Foundation
project
that
was
initially
focused
on
the
software-defined
storage
control.
B
So
but-
and
these
are
the
supporting
companies
behind
the
project
the
but
the
things
that
we
there
are
a
lot
of
end
users
in
the
open,
SDS
community-
they
are
requesting
for
more
than
storage
control
there.
There
are
the
things
that
Reddy
was
mentioning
about
the
data
pipeline.
The
integration
of
the
different
kind
of
different
pieces
in
their
data
management
pipeline
has
to
be
integrated.
B
So
the
mission
of
soda
are
there
are
3-3
points
in
here.
One
is
to
actually
build
an
ecosystem
of
software
for
data
management
and
storage.
The
second
point
is
to
actually
facilitate
the
cross
project
collaboration.
So
the
thing
right
now
is
that
there
are
a
lot
of
projects
out
there.
There
are,
you
know,
solving
singular
I
mean
specific
spot
solutions,
both
problems,
but
again
things
like
what
radios
talking
about
myself
is
so
it
has.
This
I
mean
it's
great.
B
Yes,
it's
all
Yukiko
system,
but
it
doesn't
work
with
other
kind
of
storage
or,
if
you
need
what
needed
to
work
with
other
things,
you
have
to
actually
develop
pieces
of
things
to
integrate
them
together.
So
so
what
we
want
to
do
is
actually
try
to
get
everybody
together
and
then,
under
this,
our
project
itself
try
to
integrate
pieces
of
all
these
spot
solutions
together
and
the
third
point
about
this-
our
mission
is
to
actually
be
able
to
provide
end-users
with
quality
end
to
end
solutions.
B
So
again,
the
focus
here
is
an
end-to-end
solutions,
so
that
end
users
don't
have
to
do
the
work
of
like
trying
to
piece
everything
together.
Okay,
so
so
this
is,
this
is
actually
a
huge
task.
I
mean
it's
like
when
you
scale
Mountain
I
mean
you
can
do
it
alone,
I
mean
we
need
to
a
team.
I
mean
a
team
of
people
to
actually
do
do
that.
So
that's
why
we
are
here.
B
B
Unlike
a
lot
of
other
projects,
this
project,
we
ever
a
lot
of
active
end-user
participation,
so
we
have
like
companies
like
Toyota,
like
they
have
certain
requirements
that
they
tell
us
and
like
Vodafone
air,
so
the
needs
that
they
have
that
we
try
to
put
into
the
roadmap
and
stuff
like
that.
So
I
mean
the
the
roadmap
and
the
solutions
or
the
features
that
we
are
developing
ROG
real,
quick,
real
requirements
from
end-users.
B
B
So
what
we
want
to
do
is
actually
to
be
able
to
build
a
really
kind
of
ecosystem,
so
that
users,
when
they
use
it,
they
can
just
reuse,
it
I
mean
I,
I
mean
deploy
and
it
starts
running
and
things
different
components
or
from
different
solutions
from
different
vendors
can
just
work
together.
This
is
what
we
are
trying
to
achieve,
also
so
there
so
again,
so
that
is
gonna
be
more
like
a
and
umbrella
for
different
kinds
of
projects.
B
The
original
open,
STS
project,
the
these
are
the
ones
that
I
in
blue
there
we
are
moving
on
to
solar,
so
they're,
like
the
core
itself,
is
actually
doing
the
orchestration
doing
the
scheduling
and
stuff
like
that
telemetry.
Well,
we
try
to
collect
the
metrics,
and
then
we
actually
integrate
with
Prometheus.
You
know
for
for
the
she
kind
of
our
control
and
then
automation.
B
Intelligence
is
something
that
a
lot
of
the
companies
that
we
talked
to
are
very
interested
in
on
how
to
actually
be
able
to
build
in
intelligence
into
the
platform
like
this
morning.
We're
talking
to
arm,
for
instance,
and
they
are
talking
about
like
how
we
can
actually
try
to
leverage
the
edge
storage
kind
of
control
to
actually
be
able
to
determine
life
are
the
kinds
of
data
that
you
want
to
send
to
the
cloud
or
sent
to
the
data
center
and
which
other
data
that
you
you
just
need,
the
other
edge
and
stuff
like
that.
B
So
this
are
the
kind
of
AI
or
you
know,
machine
learning,
kind
of
things
that
we
want
to
put
it
multi-cloud
piece.
This
is
kind
of
more
or
less
done
is
more
like
an
object
kind
of
storage
that
we
actually
can
a
data
control
that
connects
to
AWS,
Google
and
ledger.
And
besides
that,
we
also
connect
to
the
IBM
cloud
and
also
anything
that
talks
s3.
Okay,
for
object,
storage,.
B
Plugins
so
like
what
radios
showing
we
have
plugins
for
kubernetes,
we
have
plugins
for
OpenStack
and
right
now
we
are
developing
the
plug-in
for
VMware
wheesh,
a
lot
of
enterprise
users,
as
she
still
use
I
mean
all
those
crew
cool
and
it
is,
is
great
yeah
and
then
the
dashboard.
So
there
some
other
projects
that
we
are.
We
are
trying
to
always
try
to
start
in
next
year,
like
for
storage,
that's
nvme
storage
that
we
want
to
do.
That's
all
a
she
kind
of
storage,
I,
availability,
storage
that
relative
and
then
in
the
edge
side.
B
Many
companies
are
discussing
about
this
project.
So
how
what
we
wanted
to
do
over
there?
Okay
and
then
flick-
is
more
the
in
clock
control
like,
for
instance,
you
have
an
image
that
you
back
up
to
the
cloud
is
solved
and
the
snapshot
that
you
have
in
the
cloud.
How
do
you
actually
connect
it
to
today
and
AWS
kind,
of
instance,.
B
So
the
reason
why
we
are
having
this
is
there
a
lot
of
companies
when
they
do
purchasing
I
mean
if
they
need
new
kind
of
a
solution
and
so
on
they
have
to
write
pages
of
RFPs
and
so
on,
stating
all
kinds
of
different
tests
and
so
on.
So
the
end
users
are
asking
is
like:
if
I
don't
wanna
buy
some
kind
of
data
management
solution
can
I
just
have
like
I
just
say:
I
need
a
soda
or
any
kind
of
solution.
B
So
this
this,
what
we're
trying
to
help
the
end
users
is
so
so
we
expect
this
to
I
mean
there
are
three
parts,
the
conformance
part,
the
different
hardware
software
you
know,
or
cloud
services
to
be
able
to
interoperate
with
one
another.
I
mean
that
kind
of
conformance
testing,
and
also
the
performance
testing
to
baseline
a
product
or
solution
the
for
certification
for
service
providers
to
be
able
to
support
audio
end
users.
So
this
would
be
more
of
the
essence
out
there
and
then
also
securing
for
like
operators
and
then
developers
also.
B
So
we
do
work
with.
Like
I
mean
we
do
collaborate
with
other
kinds
of
organizations
like
you
know,
storage
performance
Council
to
do
all
these
things,
so
we
do
not.
We
do
not
expect
everything
to
be
done.
You
know,
by
by
ourselves,
I
mean
these
communities,
so
scuzz
there's
a
lot
of
work.
That's
already
done
out
there,
so
we
want
to
be
able
to
collaborate
with
all
these
different
kinds
of
organizations
to
do
the
work,
like
also
like
kubernetes,
sorry
practice,
certification
kind
of
program.
B
The
the
governance
of
this
project
is
very
similar
to
a
lot
of
these
open-source
projects,
especially
Linux
Foundation
kind
of
projects.
The
value
have
a
board,
and
then
you
have
like
the
technical
steering
committee,
the
user
advisory
committee
and
then
also
in
our
accommodation,
the
the
under
the
technical
steering
committee.
All
the
different
projects
will
sit
there,
but
we
will
have
an
architecture
kind
of
workgroup
to
make
sure
that
all
these
projects
they
are
that
are
contributed
into
this.
Our
foundation
is
looking
works
together
with
other
projects.
B
Okay,
so
that's
the
the
purpose
of
the
architecture
workgroup
and
the
QA
workgroup
is
to
you
know,
make
sure
that
all
the
software
I
mean
there
are.
There
are
like
so
certain
kinds
of
QA
quality
that
we
ensure
over
there,
and
we
also
have
like
the
regional
committees
to
build
up
the
committees
in
these
different
regions.
So.
B
For
for
events,
we
do
a
lot
of
events
all
over
the
world.
We
just
did
the
one
in
their
own
open
source
summit
and
then
here
and
then
the
actually
the
next
one
that
we're
going
to
be
doing
is
the
kubernetes
forum
in
Seoul
and
also
not
singing
we're,
not
gonna
go
singing.
It's
too
far
for
a
lot
of
people
to
go
so,
but
but
the
the
key
thing
is
that
in
on
December
17
we're
gonna
do
is
sort
of
forum.
B
It's
called
like
a
mini
mini
sonic
and
I
think
in
Tokyo,
and
then
we
all
have
like
interested
parties.
They
are,
you
know,
interested
in
joining
the
sort
of
foundation
to
or
to
choose
the
starting
foundation
member
to
come
and
then
meet
up
and
discuss
the
plans,
and
you
know,
governance
for
the
project.
B
So
if
you
guys
are
interested
in
this,
come
talk
to
talk
to
us
after
this
meaning
yeah.
So
so
we
so
we
welcome
you
guys
to
join
us,
for
you
know
as
whether
you're
developer
you're
a
vendor
or
end-user
out
there.
There
are
all
kinds
of
benefits
to
joining
this
project.
Okay,
there's
data
out
there.
So.
C
A
It's
really
a
birds
of
a
feather
session,
so
somewhat
a
unique
setting.
This
is
not
just
a
technical
session
where
we
are
presenting,
so
we
are
looking
for
feedback
in
different
areas.
So
when
you
look
at
the
data
management
platform,
there
are
different
challenges,
so
we
are
kind
of
looking
at.
How
do
you
deal
with
things
like
multi
cloud?
Where
do
you
see
the
challenges?
Where
do
you
see
the
opportunities?
What
would
be
the
typical
use
cases?
A
You
know
integration
edge,
use
cases.
These
are
all
different
areas
that
we
are
looking
at
and
figuring
out.
The
best
way
to
not
only
prioritize
the
community
work,
but
also
look
for
ideal
use
cases
that
actually
help
the
solve
the
real
customer
challenges.
So
any
ideas
feel
free
to.
You
know
pick
up
a
mic
and
you
know
let's
brainstorm,
to
see
what
what
do
you
see
are
the
challenges
when
it
comes
to
data
management,
not
just
a
Cubana
days,
CSI
driver,
that's
pretty
narrow
area.
We
don't
need
to
solve
that
problem,
but
emerging
problems.
A
A
A
B
So
the
solar
certification,
we
expect
the
the
work
group
to
reform
America,
forming
the
work
group
right
now,
and
we
expect
the
definition
I
mean
the
graph
for
the
definition
by
D
by
me
next
year
and
then
the
actual,
because
we
are
talking
to
some
universities,
also
like
how
we
gonna
do
this
certification
and
stuff
like
that.
So
we
are
talking
to
some
Japanese
University,
some
US
code,
universities
and
even
you
UK
universities
to
to
do
that
kind
of
certification
yeah.
A
Yeah,
so
the
question
is
what
kind
of
user
experience
are
you
thinking
about
when
you're
talking
about
ingestion,
engineering
and
analysis
in
the
data
pipeline
flow?
So
we
are
essentially
looking
at
so
everything
is
going
to
be
from
a
deployment
perspective.
Obviously,
this
software
stack
has
to
be
deployable
very
easily
right,
so
obvi.
What
is
the
best
way
to
enable
the
developer
ecosystem?
So
it
has
to
be
written
and
go.
A
It
has
to
be
containerized
easy
to
deploy
and
all
that
good
stuff
right,
but
from
a
user
experience
perspective,
what
we
want
to
see
is
everything
is
actually
exposed
through
REST
API,
both
southbound
plugins,
as
well
as
the
not
born
plugins
right,
that's
a
key
fundamental
design
principle,
and
then
we
can
take
those
api's
able
to
stitch
the
UI
so
that
there
is
a
initial
ramp
up
curve.
If
you
just
have
REST
API,
so
it's
very
difficult
to
get
traction
in
the
real
deployment.
A
You
know
use
cases,
so
we
do
need
to
have
a
UI
that
can
actually
bring
together
all
the
different
parts
of
the
building
blocks
like,
for
example,
how
do
you
discover
the
storage
back-end?
How
do
you
create
a
profile
that
essentially
says
this?
Is
the
performance,
oriented
storage,
but
behind
the
profile
you
have
three
different
storage
backends.
A
It
needs
to
be
very
easy
to
use
UI
and
then
how
do
you
actually
create
workflows
that
can
actually
do
the
data
lifecycle
management,
so
everything
is
exposed
to
risk
to
UI
and
then
GUI
is
just
a
mechanism
to
get
some
jump.
Starting
on
real
deployment
is
what
we
are
thinking.
Some
of
this
plumbing
is
already
there
like
Steven
talked
about,
we
do
have
you
know
basic
building
blocks
with
rest.
A
Api
is
exposed
to
integrate,
with
VMware
and
humanities,
for
example
with
few
backends
the,
and
there
is
a
UI
as
well,
and
we
continue
to
refine,
but
in
general
the
goal
is
everything
is
exposed.
All
the
data
visas
are
actually
exposed
through
REST
API,
and
you
have
a
GUI
that
can
actually
bring
out
all
those
features
in
a
very
easy
to
n1.
You
know,
easy-to-use
fashion
did
I
answer
your
question.
D
A
Right
right,
so
bunch
of
these
use
cases
are
still
evolving.
So
if
you
look
at
machine
learning
and
deep
learning,
for
example,
we
have
to
understand
the
use
case
to
an
extent
where
it
and
then
translate
that
into
common.
You
know
data
services
framework
and
make
sure
it
works.
So
there
is
a
big,
lengthy
curve
for
certain
use
cases
like
autonomous
driving,
for
example
right.
A
So
you
collect
the
data
from
a
car
and
then
that
data
gets
ingested
into
the
storage
sitting
in
a
cloud
or
on
Prem
you
know,
environment
and
then
you
need
to
be
able
to.
You
know,
do
some
sort
of
filtering
and
analysis
and
data
reduction
process,
and
then
you
need
to
expose
through
machine
learning
and
deep
learning
frameworks.
So
bunch
of
these
use
cases
are
still
evolving,
but
the
plan
is
to
kind
of
make
sure
we
understand.
A
The
use
case
pipeline
very
well,
translate
this
into
data
management
pipeline
and
then
make
sure
we
test
the
theory
with
users.
Stephen
was
talking
about.
We
have
few
end
customers
who
are
actually
giving
the
feedback,
and
the
goal
is
to
kind
of
address
the
critical
use
cases
and
use
them
as
our
proxy.
Then
we
implement
and
make
sure
they
consume
it.
A
That's
what
we
are
thinking
as
opposed
to
you
know:
DevOps,
developer
type
of
mentality,
where
we
judge
develops
certain
features
and
see
what
happens
so
we
are
kind
of
trying
to
bring
in
the
user
end
users.
Their
prioritize
use
cases,
make
sure
we
implement
the
workflows
correctly
so
that
it
gets
consumed
through
proof
of
concepts
and
all
that.
C
A
You
know
these
databases
have
different
ways
of
actually
extracting
so
the
goal
is
we
want
to
be
able
to
tap
into
anything
that
is
open
source
that
can
actually
does
that
job
for
us.
So
Stephen
talked
about
Prometheus,
for
example.
It's
a
very
good
monitoring
tool.
Telemetry
you
can
actually
collect,
you
can
actually
store
and
you
can
analyze
through
that
framework.
A
So
our
goal
is
to
actually
tap
into
the
existing
open
source
projects
that
are
very
popular,
have
been
very
successful
or
very
mature
and
use
them
as
the
core
infrastructure
building
blocks
and
then
be
able
to
actually
do
the
data
management
pipeline.
Things
like
if
you
want
to
extract
the
data
from
Viroqua.
A
Splunk
different
databases
logs,
you
know
unstructured
data
sitting
in
some
object
store.
We
need
to
be
able
to
pull
that
and
make
that
happen.
Whether
we
are
going
to
do
that.
You
know
like
things
like
mainframe,
ingesting
the
data
from
mainframe,
it's
just
a
matter
of
time,
but
the
goal
is
to
prove
out.
The
framework
is
flexible
enough
to
do
that,
whether
it
happens
as
a
first
use
case
or
the
next
use
case
really
depends
on
which
one's
customers
really
care
for.