►
Description
Richard Morrell talks to Ross Turk about the latest greatest news around Ceph and how Red Hat Cloud marches on storage, why Ceph and Gluster are complimentary. We talk Firefly and a look forward to Giant and we discuss why Inktank was such a no brainer addition to Red Hat and why it adds to the strength of Red Hat Open Hybrid Cloud.
B
It
has,
it
was
even
busy
before
that,
because
the
startups
are
always
busy,
but
it
was
kind
of
strange
to
realize
that
it
gets
busier
than
a
start-up
when
you
get
acquired
by
a
large
company
like
red
hat.
So
there's
lots
of
excitement
a
lot
of
work
to
do
a
lot
of
a
lot
of
work
to
be
done,
but
it's
been
busy
but.
A
B
We
were
always
a
very
open
source
e
company.
You
know
we
always
not
only
released
an
open
source
product,
but
we
cared
about
the
methodology.
We
wanted
to
run
a
community
that
was
not
just
us,
so
I
think
it's
the
rrr
worldview
and
redheads
world.
We
were
compatible
from
the
start
and
you
know
it
we
couldn't
have.
We
couldn't
have
come
up
with
a
better
company,
but.
A
B
A
B
Who
are
looking
to
acquire
storage
should
look
at
red
head
for
the
same
reason.
You
look
at
Red
Hat
for
linux,
and
it's
because
redhead
understands
how
to
take
the
best
technology
from
open
source
communities
and
deliver
it
to
enterprises
and
right
now,
open
source
storage
is
is
emerging
and
it's
a
very
exciting
thing
and
red
head
is
the
right
company
to
talk
to
about
it
because
they
can
take
all
of
the
pieces
and
bring
them
together
into
a
reliable
customer
experience.
But.
A
Storage
has
always
been
one
of
those
emotional
purchases.
It's
always
been
the
EMC
clarion,
which
took
up
55
60
of
your
budget,
and
now
it's
heterogeneous.
It's
made
of
you
know
redundant
technology
really
cheat
to
acquire
technology.
Do
you
think
thats
F
really
has
blazed
the
trail,
especially
with
Apache
CloudStack
and
with
OpenStack
I.
B
Think
so
OpenStack
was
was
kind
of
act,
one
for
force
F,
and
it
was
a
good
match,
because
the
the
type
of
person
who's
deploying
OpenStack
is
also
the
how
the
person
who
will
feel
at
home
with
technologies
like
Seth.
So
it
was
a
really
good
first
match
if
you
were
to
try
to
take
steph
and
give
it
to
a
standard,
netapp
admin
there's
too
much
of
a
gap.
It's
just
a
completely
different
way
of
looking
at
things,
so
I
think
it's.
B
What
we're
seeing
is
a
pivot
in
the
storage
industry
between
you
know
the
old
way
in
the
new
way
and
SEF
and
gluster
and
technologies
like
that,
are
definitely
emerging
and
are
good
for
good
for
particular
use
cases
now.
But
I
think
that
storage
buyers
who
are
looking
towards
the
future
kind
of
understand
that
this
is
this
is
the
way
things
are
going
to
go
as
far
as
software
based
storage
that's
distributed
on
commodity
hardware
in.
A
B
Think
that
was
part
of
it
I
think
a
lot
of
our
early
community
came
from
that.
It
was
the
first
really
big
win
in
in
SEF,
and
you
know,
but
that
was
that
was
not.
That
was
not
the
beginning
of
the
story
that
was,
that
was
a
few
steps
in
because
it
took
it,
took
a
you
needed
a
community
in
order
for
linda's
to
feel
like.
That
was
something
that
needed
to
happen.
So
I
think
that
was
almost.
That
was
the
finish
line
and
the
starting
line
at
the
same
time
talk.
B
File
systems
are
what
we're
all
used
to.
Of
course,
block
devices
like
steps,
virtual
block
device-
are
it's
so
fundamentally.
Files
and
directories
are
a
complicated
problem,
especially
when
you
distribute
them.
So
that's
the
piece
of
stuff
that
is
the
most
complex.
The
block
interface
on
top
of
Seth
was
a
far
thinner
piece
of
technology
and
allowed
building
on
top
of
Ray
doses,
object
store.
That
was
a
really
simple
thing
to
build
and
I
say
simple,
of
course,
and
nothing's
ever
simple,
but
it
was.
B
It
was
a
relatively
thin
part
of
the
architecture
and
allowing
you
to
store
virtual
disks
inside
of
Rados
gave
us
the
ability
to
be
a
good
fit
for
virtualization,
which
was
something
that
a
lot
of
people
really
really
really
needed
right
now,
so
I
think
files
and
directories
are
always
going
to
be
required
for
legacy.
Apps
they're
always
going
to
be
required
for
things
like
desktop
sync
and
share,
but
if
you're
running
virtualized
environments,
what
you
need
is
block
storage.
B
A
B
Sf
is
definitely
built
for
that.
It's
it's
it's
built
to
be
able
to
adapt
very
quickly
so
that,
instead
of
your
setup
and
teardown
of
storage
requiring
a
screwdriver-
and
you
know
physical
access
to
a
data
center
that
you
can
do
that
all
virtually
that's
the
whole
point
of
storage
virtualization,
and
that's
that's
what
we
that's.
What
we
are
trying
to
bring
to
the
next
generation
of
storage.
Is
that
flexibility
so
similar
to
OpenStack
but
virtualizing.
You
know
it
storage,
in
addition
to
compute
and
everything
else,
I'm.
B
B
A
B
So
we
recently
released
SEF
Firefly.
It
was
a
little
bit
ago.
The
major
features
in
Firefly
were
erasure
coding
in
cash.
Steering,
erasure
coding
is
a
different
kind
of
storage,
back-end
force
F.
That
allows
you
essentially
to
fit
more
stuff
in
less
space
and
preserve
the
same
amount
of
durability,
and
it
does
that
by
using
parody
codes,
instead
of
full
replicas
of
the
odd
that
is
paired
with
the
cash
steering
which
allows
you
to
define
a
pool
as
a
cash
for
another
pool.
B
So
you
can
essentially
have
a
cheap,
slow
pool
and
an
expensive,
fast
pool.
You
can
scale
them
both
independently
and
they
work
together
to
move
data
from
the
hot
tier
to
the
cold
tier
depending
on
on
access.
So
it
allows
you
to
have
what
we,
what
we
call
the
multi
temperature
data
management
and
be
able
to
sort
of
very
flexibly
manage
that
price
performance
trade-off,
but.
A
B
B
You
know
and
be
able
to
pull
that
information
report
it
to
their
management
in
a
logical
insane
way,
just
because
all
the
information
exists
doesn't
mean
that
the
industry's
figured
out
how
to
do
anything
with
this,
so
I
think
there's
a
lot
of
growth
that
needs
to
happen
within
you
know,
within
the
storage
admin
community,
to
figure
out
how
to
how
to
pull
that
data
out
of
setup
and
use
it
most
effectively
in
business
communication.
But.
A
B
Actually,
mail
is
still
one
of
the
things
out
there
that
doesn't
have
a
good
storage
solution.
Now
we're
waiting
for
somebody
to
build
mail
on
top
of
Rados,
because
we
really
need
it,
but
yeah
I
think
that
then,
in
the
future,
storage
is
going
to
be
it's
going
to
be
more
about
DevOps
than
it
is
about.
You
know,
forklifts
and
web
browsers
back.
A
B
Absolutely
and
I
think
even
we're
seeing
now
that
to
order
a
start
to
operate
a
safe
cluster
having
developer
expertise
is
really
good
because
you
can,
you
can
build,
you
know
all
the
things
you
need
to
automate
yourself,
cluster
growing
and
shrinking
et
cetera.
The
next
step
is
figuring
out
how
applications
can
be
built
to
talk
to
SEF
directly
and
take
advantage
of
all
the
things
that
are
in
the
SEF
object
store
like,
for
example,
object,
classes.
Allow
you
to
most
objects
towards
just
have
properties,
but
not
methods.
B
You
know
chef
has
methods
so
having
the
ability
to
build
your
application
on
top
of
SEF
as
a
platform
as
a
storage
platform,
instead
of
just
a
place
to
put
stuff
is
the
next
step
in
that
evolution
so
going
beyond
DevOps
into
you
know
into
engineering
and
architecture.
For,
for
you
know,
business
platforms,
I
think
is,
is
particularly
interesting.
I've.
B
We
do
there
are
enough
variables
and
SEF
that
you
know
having
to
having
to
have
a
community
that
isn't
on
the
latest.
Stuff
is
very
difficult
and
the
the
release
process
has
been
it's.
It's
been
a
compromise
over
a
long
period
of
time
and
that
we
release
every
three
weeks
generally,
but
then
every
three
months
there's
a
major
release
and
then,
of
course,
following
on
from
that,
the
enterprise
product
that
gets
released
by
redhead
is
yet
another
release
cycle.
B
So
there
are
multiple
release
cycles
that
you
can
sort
of
jump
on,
depending
on
your
level
of
risk
tolerance.
We
wanted
to
make
sure
that
their
releases
for
fast
enough
to
engage
a
community
and
fast
enough
to
compete
the
innovation
going
but
slow
enough
that
enterprises
could
have
a
trusted
thing
to
deploy,
and
that's
that's
tough
to
do
while
trying
to
keep
everybody
on
the
same
version.
So
it's
it's
a
big
compromise
and
it's
imperfect
but
I.
B
Think
generally,
you
have
the
the
the
leading
edge
users
who
are
always
on
the
latest
stuff
who
want
to
push
the
envelope.
And
then
you
have
people
who
end
up
using
ink
tanks
of
enterprise
who
want
something
stable
and
reliable
and
figuring
out
how
you
go
from
one
to
the
other
is
well.
That's,
that's
the
trick
right!
That's
what
Red
Hat
can
help
us
figure
out,
but.
A
I
think
you
came
to
red
hat
with
one
thing
that
most
startups
don't
have
pipeline
and
revenue.
You
were
a
company
that
already
had
bankable
revenue,
people
out
there
using
the
enterprise,
supported
version
and
I
think
combine
that
with
rail.
Seven
ready
for
the
cloud
combine
that
with
all
the
plans
that
Red
Hat
has
an
open,
hybrid
cloud
and
are
on
how
we
work
with
public
cloud
providers.
This
is
a
time
really
like
never
before
so.
I
think.
B
So
I
think
Seth
is
is
a
piece
of
a
big
puzzle.
You
know
when
we
were
at
intake.
Of
course.
Seth
was
our
whole
world
and
that's
what
we
looked
at
and
we
looked
at
it
as
a
standalone
business,
and
we
really
believed
it
was
the
future
of
storage
and
at
Red
Hat.
We
still
believe
that,
but
it's
a
piece
of
a
bigger
puzzle
that
includes
OpenStack.
It
includes
our
l7.
It
includes
all
of
the
management
you
know
technologies.
So
that's
that's
something
that
I
think
is
potential
that
we've
never
never
had
before.
A
B
That's
crazy,
that's
crazy!
I
was
at
via
linux
for
10
years
and
it
was.
It
was
a
rocky
ride,
but
the
whole
time
the
whole
time.
I
was
always
very
thankful
that
I
could
work
on
open
source
technology
and
that
there
was
that
there
was
money
behind
it.
So
I
mean
I.
I
I
don't
I,
don't
pinch
myself,
because
it's
a
lot
of
hard
work.
Yeah.