►
From YouTube: CNCF Storage WG Meeting - 2019-02-27
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
A
B
A
Okay,
so
the
first
question:
how
would
you
identify
yourself
so
here
are
the
categories
and
then
are
you
aware
of
the
CNCs
storage
landscape
paper
so
have
a
link
there,
yes
or
no,
which
CNCs?
What
project
are
you
using
or
contributing
to
you
so
this,
since
this
is
the
more
with
the
surveys
more
for
the
user
site?
So
you
have
this,
you
know
whether
you're
using
it
or
not
here
so
we
have.
A
A
Okay
and
what
attributes
some
store
systems
are
the
most
important
for
you
when
the
in
decisions
we
like
which
system
to
choose.
So
this
is
new
addy.
This
one
I
think
this
is
a
Quinton's
feedback,
so
I
don't
know
many
people,
I
choose
everything.
Basically,
just
saying
there
are
some
trade
off
that
you
have
to
between
your
choices
right.
So
then,
if
you
have
any
problems
making
those
decisions,
I
can.
A
A
A
A
A
So
is
it
ok,
we
can
we
can.
We
can
go
through
the
rest
of
them
and
then
you,
then
you,
let
me
know
if
you
think
it's.
This
is
ok,
so
basically
the
the
in
the
white
paper.
We
first
talked
about.
Ok,
so
those
are
the
the
main
of
5
Sorge
attributes
and
then,
when
you're
making
decisions,
you
know,
do
you
have
to
decide
what
is
most
important
for
you
right?
A
So
if
you
let's
say
I
have
some
example
here,
which
is
something
that
is
a
system
that
is
strongly
consistent,
but
maybe
that
cannot
scale
horizontally.
You
know
that
maybe
some
system
has
very
high
durability,
but
then
maybe
performance
is
not
good,
so
things
like
that,
so
some
of
the
other,
like
the
thin
protection,
those
those
are
on
top
of
that
there
are.
Those
are
the
other
services
that
you
add
on
top
so
putting
them
together
then
impact
your
decision
and
the
impact
of
the
all
the
basic
attributes.
B
A
A
A
B
A
A
Maybe
somebody
will
provided
some
database
service
on
top
of
her,
like
AWS,
a
Google
cloud,
and
then
you
know
customer
were
just
to
use
that
once
you
actually.
So
that
would
be
this
layer
on
top.
So
we
have
this
one
there
anything
else
we
could
add,
or
this
is
I,
think
either
it's
a
hardware.
Actually,
the
hardware
storage,
the
machine
were
software
on
commodity
servers
or
public
cloud
Pryda.
A
B
A
C
A
A
Okay
block,
okay,
so
blog,
then
you
specify
what
blocks
through
system.
Then
you
give
a
name,
then.
Actually,
if
you
give
a
name,
then
if
it's
well
known
that
we
probably
already
know
or
the
other
choose
about
that
particular
street
system,
so
we
still
highlighting
those
two
questions.
Why
is
the
data
protection?
A
Basically,
those
are
the
you
know.
We
basically
just
listed
raid
and
eurasia
coding,
those
type
of
technologies,
and
in
the
next
question
we
have
other
order
binds
to
services
like
replication
snapshots,
own
encryption
or
other
so
just
to.
If
some
people
may
not
know
all
those
details
about
like
the
the
first
one
I'm,
not
sure
so
so
that
we
have
other
or
don't
know.
If
they
Dre,
don't
know
the
details,
just
they
could
just
sing,
don't
know.
A
A
D
A
A
A
Okay,
what
the
protocol
is
it's
a
kind
of
important
I
can't
I
can't
like
IBM.
That
could
be
its
ass
cause.
Yes,
that's
covered
at
least
that's
one,
one
more
protocol
there,
I
scuzzy,
just.
C
A
A
A
C
That
was
actually
general
comment
that
I
gave
Jing
was
we've
got
users
of
the
stuff
and
they
have
knowledge
and
a
set
of
interests.
C
A
B
A
A
So
I
don't
know,
do
we
need
to
add
a
question
for
that,
but
then
do
we
need
to
add
a
question
that
do
you
manage
your
own
storage
I
think
it
depends
on
what
some
customers
do
some
kinds
of
you?
Don't
do
you
need
to
say
to
you
manage
your
own
storage?
Do
we
need
to
milk,
add
a
question
somewhere
on
top
or
I.
Don't.
B
E
A
F
E
F
C
Had
a
lot
of
similar
comments,
I
think
as
soon
as
somebody
answers
you
know,
DBS
thing
they
use,
then
we
kind
of
know
what
the
answers
to
a
whole
bunch
of
other
questions
are.
So
we
should
not
ask
them,
and
so
the
leave
then
use
SIF
or
they
didn't
choose
whatever
something
else.
Then
a
whole
bunch
of
questions
is
very
relevant.
You.
C
Agree
and,
and
and
that's
why
I'm
saying
this
kind
of
is
looking
more
and
more
like
it
needs
to
be
a
proper
interactive
questionnaire.
Where,
depending
on
the
answers
to
your
questions,
you
get
asked
other
questions
because
you're
totally
right,
I
mean
when
you,
when
you
ask
them
which
of
these
do
you
use
and
they
say:
I
use,
EBS
and
I
use,
SEF,
on-prem
or
whatever,
because
I'm
a
hybrid
user.
C
Then
the
answers
to
the
EBS
questions
are
different
to
the
answers
to
the
set
questions
and
you
need
my
to
say.
What's
the
same
question
twice
once
for
EBS
and
once
was
saved
except
that
for
EBS,
we'll
really
know
what
the
answers
are
for
the
most
case,
because
they're
defined
by
the
safe
product
or
sorry
the
EBS
product,
so
yeah
we
end
up
making
a
very
we
either
I
see.
C
Chris
is
saying
yeah
we
need
a
branching
questionnaire
and,
and
the
other
overarching
kind
of
comment
I
made
was
if
we
give
people
more
than
you
know,
a
small
number
of
questions
to
answer,
particularly
if
a
percentage
of
them
are
like
questions,
we
shouldn't
need
to
ask
them.
They
will
just
give
up
and
not
give
us
a
list.
B
B
A
E
C
Out
to
people,
are
we
gonna
email
it
to
them,
or
are
we
gonna,
you
know
announcer,
but
the
community
meeting
or
whatever
was
that
your
question
Steve?
Yes,
okay,
so
so
I,
don't
know
the
precise
answer
to
that
I
believe
it
was
just
mailing
lists.
I
can
find
out
precisely
how
they
did
it.
I
think
it's
a
different
question
is,
is
how
do
we
actually
expose
this
to
to
users?
Is
it
that's.
C
Is
it
a
piece
of
software
that
we
develop
or
is
it
something
else
spreadsheet
they
fill
in
or
whatever
that's
a
different
question
and
it's
looking
increasingly
to
me
I
think
if
we
had
a
simple
set
of
questions
that
we
asked
everybody
and
they
just
went
and
clicked
through
multiple
choice,
questions
you
know,
Survey
Monkey
would
be
a
reasonably
decent
answer.
It's
not
clear
to
me
that
we
can
actually
formulate
those
branching
question
and
effectively
using
Survey
Monkey,
but
I
I'm,
not
an
expert
on
that.
So
maybe
we
can
figure
that
out
later.
Okay,
yeah.
A
C
It's
actually
multiple
serving
ok
one
other
thing,
while
I'm
babbling
on
I
just
wanted
to
point
out
that
I
think
the
idea
that
you
had
Steve
of
having
multiple
rounds
of
surveys
where
you
asked
you
know
one
set
of
Christians
and
then,
depending
on
who
answers
what
you
send
out.
Another
set
of
surveys
I
think
the
problem
with
that
is
that
you
end
up
with
a
potential
like
survey
fatigue
problems
where
people
say
I
I've
already
done
the
bloody
the
storage
wrote
in
group
survey
I'm
not
doing
another
one,
always
in
the
worst
case.
B
A
B
A
A
So
can
we
just
ask
some
general
questions
first
and
then
we'll
see
if
they're
you
know,
if
it's
too
much,
you
know
we
can
say
if
you,
if
you
are
your
row,
is
something
then,
for
example
like
this
question
right
and
then
what
you
say
if
you
are
and
this
this
one
requires
something.
We
can
add
this
and
we'll
see
how
that
how
that
goes,
because
if
they
are
not,
they
can
just
keep
this
question
rather
than
designing
a
complete,
even
survey
for
a
different
row,
it's
probably
not
necessary.
C
Optional
questions,
then
at
least
we
can,
you
know,
get
the
basic
answers
from
the
people
who
who
not
prepare
to
spend
much
time
and
get
more
information
from
people.
You
are
little
bit
like
those
freeform
questions
at
the
end
of
some
of
these
things
like
is
there
any
other
stuff
want
to
tell
us,
and
then
some
people
will
write
tons
of
stuff
there
and
other
people
will
leave
it
blank,
but
maybe
we
want
to
do
it
in
a
slightly
more
structural.
B
Yeah
I
was
having
the
same
thoughts
and
I
started,
searching
for
data
on
recommended
numbers
of
questions,
and
it
looks
like
there's,
there's
people
who
say
that
suggest
that
you
tell
them
knotti,
not
just
the
number
of
questions,
but
your
request
says
this
will
take
10
to
15
minutes
and
a
lot
of
people
will
will
enter
into
the
path
if
you
promise
them
that,
but
beyond
that,
you're
not
going
to
get
much
participation
for
a
non-paid
survey.
So
maybe
what
we
should
do
is
rather
than
questions
we
should
test
this
on
ourselves.
B
B
A
A
B
A
Already
ice
or
we
can
cut
I'm
finished,
but
I
was
just
thinking
the
cluster.
We
don't
really
have
a
cluster
admin.
Do
we
need
to
add
that
one?
We
only
have
you
see
the
vendor,
the
contributor
distributor
user?
Do
we
need
to
add
a
admin
yeah
probably
should
or
is
the
operator
will
be
operating
will
be
there
honestly.
B
A
So
basically,
we
just
remove
this
one.
Just
let
them
decide
if
they
can't
everyone.
So
let's
say,
for
example,
you
know
we
make
this
one
optional
question
right
optional.
This
is
a
more
detailed
question.
If
they
really
need
this,
then
then
they
can
answer
this
one
right,
they
can
say,
don't
know.
So
we
put
this
one
at
the
end.
It's
ok
or
yeah.
B
E
A
B
A
B
A
Right
so,
okay,
so
maybe
we
should.
We
have
something
similar
just
give
some
examples.
What
is
actually
I,
don't
know
what
is
the
one
on
there?
What's
the
name
of
the
one
arm,
it
obvious
is
that
EFS
something
like
that
or
something
like
that,
and
then
a
lot
of
stuff
and
I
sell
something
something
all
right.
Something
like
that
we
actually
yeah
I,
do
have
the
prodigal
questions
earlier.
My
dad
also
removed
that
I.
Think
after
let's
see
facts
about
they.
If
this
specified
this
one,
then
we
would
know
right.
A
A
So
I
thought
the
addicts
who
say
you
know
he
he
was
actually
wanted
to
keep
those
separate
like
that.
910,
don't
have
a
question
that
he's
thought.
Maybe
with
this
we
could
have
some
ideas
for
the
next
round
of
we
could
have
some
use
case.
You
know
look
at
what
type
of
use
case
we
want
to
look
into
next.
I.
Think
that's
what
he
was
saying
so
know
a
little
bit
more
details
are
then
or
just
a
very
high-level
questions.
So
I
think
that's
what
he
was
saying.
A
C
Just
to
be
clear,
I'm
not
saying
we
shouldn't
ask
them
questions
I'm,
just
saying
every
question
we
ask
mm-hmm,
we
should
be
clear:
why
are
we
asking
and
what
we're
going
to
do
with
the
information
they
forget
it?
Otherwise
we're
gonna
end
up
asking
them
all
the
questions
we
can
think
of,
asking
them
and
they're
not
going
to
answer
them
right
right.
A
So
if
they,
because
in
the
last
you
know
buy
paper,
Ryan
would
presented,
it
was
said.
Okay,
how
do
I
people
know
the
next
step?
We
want
to
do
some
use
case
right.
So
basically,
look
at
some
specific
use.
Cases,
though,
goes
through
this
survey
and
then,
after
they
answer
this,
we
can
look
at
SRI
and
then
decide
what
use
cases
we
want
to
use
and
then
maybe
do
some
more
deep
analysis
of
those
use
cases.
A
A
A
So
didn't
list
I
know
some
of
those
actually
it's
up
here
in
the
key
values,
but
actually
they
are
also
databases
and
then
the
oxygen,
no
I,
don't
this
oxygen
section
should
move
up
from
a
bit.
This
is
kind
of
a
touristy
and
we
have
this.
What
contain
oxygen
system?
Are
you
using
and
what
type
of
workloads
are
currently
running?
So
this
you
know
this
is
you
know
what
we
discussed
last
time
so
steep,
steep,
logoff
I
think
he
gave
some
suggestions.
A
You
know
we
just
went
to
the
the
docker
hub
site
and
and
just
see
what
are
the
most
used
ones
so
take
two
a
few
just
give
it
give
them.
As
an
example,
saying:
okay
tell
us
what
type
of
workloads
are
you
running
not
on
containers
on
containers
and
what
are
you
planning
to
run
in
the
next
three
to
six
months
and.
A
C
Mm-Hmm
the
reason
I'm
asking
is
I'm
wondering
if
one
way
of
simplifying
this
dramatically
would
be
just
to
limit
it
to
end-users
in
the
first
round,
mm-hmm
interested.
If
you
are
someone
consuming
cloud,
and
so
we
remove
essentially
all
but
one
of
the
options
in
the
first
question,
and
then
we
can
basically
delete
that
question
and
then
any
question
in
the
questionnaire
that
does
not
pertain
to
an
end
user.
C
We
remove,
and
the
benefit
of
that
is
one.
The
questionnaire
gets
much
simpler.
It
becomes
very
easy
to
decide
what
to
put
in
and
what
not
to,
and
then,
if
we
wanted
to
do
another
questionnaire
later
for,
like
CNC
a
project
contributors,
we
can
do
that
and
they
won't.
There
will
be
very
minor
overlap
that
the
audiences
will
be
completely
different.
So
we
want
to
have
the
questionnaire
fatigue
problem.
All.
A
B
B
A
A
C
A
We
can
move
this
up.
So
if
we
are
going
to
say,
I
have
only
10
single
questions.
We,
maybe
we
don't
need
those
titles
anymore.
It's
just
a.
We
initially
have
those
sections,
because
there
are
many
many
questions
for
each
section.
So
if
we
just
have
10
questions
that
would
just
get
rid
of
those
sections,
I
think
yeah.
We
can
move
this.
Do
it
this
one?
We
still
should
have
urn
all
right.
That.
D
A
A
B
A
C
I
want
one
suggestion
we
could
potentially
get
rid
of
question
seven
and
merge
it
into
the
sections
you
have
below.
Just
just
a
thought.
Okay,
you
have
a
block
section.
The
first
question
can
be
if
you're
using
block
complete
the
section
and
then
you've
got
an
answer
to
the
questions
of
an
automatically
okay.
A
C
A
So
there's
option:
okay
Lister,
how
Yahoo
everything
else
in
optional,
then
the
supporting
will
be
optional
to
you.
I
think
it's
maybe
not
that
important
and
then
attributes
one.
We
need
this
one
right.
We
need
to
know
they
what
is
important
for
them,
but
they
probably
will
answer
everything
kazumi.
A
C
A
A
C
A
C
Well,
I,
don't
know
what
other
people
think
if
we
said
something
along
the
lines
of
we're
only
interested
in
your
answers
here.
If
you
use
cloud
native
storage,
so
you
know
if
you're
implicitly,
if
you're
a
storage
vendor
and
you
use
stuff,
then
with
you
as
a
user,
not
as
a
vendor,
and
if
your
project
distributors
same
thing
and
if
you're
a
cloud
to
buy
the
same
thing
and
all
of
the
yeah
same
your
project.
So
basically.
A
D
C
It's
actually
very
important,
so
if
we
can
make
it
simple
to
answer
and
not
optional
and
I'd,
be
in
favor
of
that,
because
I
think
this
is,
you
know
one
of
the
main
things
we
want.
My
is
okay.
What
do
you
not?
What
like
about
cloud
native
storage,
so
I,
don't
know
to
what
extent
we
can
make
that
a
multiple
choice
with
another
category
yeah.
So
what
do.
A
D
A
A
A
C
A
That's
a
good
idea:
okay,
I
think
that's
a
good
idea,
yeah
yeah!
So
basically
they
cannot
achieve.
They
cannot
achieve
this.
So
yeah,
it's
kind
of
here,
but
I
think
what
I
have
is
probably
a
little
bit
more
thing.
Could
you
read?
It
said:
that's
basically
what
that
means
you
know.
Have
you
run
into
any
problems
trying
to
meet
those
requirements?
That's
one.
C
C
A
C
A
A
C
Pretty
straightforward
thing:
they
have
a
thing
and
we
can
go
there
and
we
can
ask
panel
of
end
users
questions
okay.
Then
he
tells
that
to
the
mailing
list,
it's
pretty
straightforward,
I
think
we
should
do
it.
Okay
question
is
just
win
when
we
want
to
do
it
and
I
guess:
okay,
once
we
find
lights,
maybe
in
a
few
weeks
time,
once
we
finalize
this
questionnaire,
we
can
maybe
use
the
other
opportunity
to
one
what
users
at
the
questionnaire
and
maybe
even
discuss
some
of
these
questions.
You
know
asking
these
specific
questions
and
okay.