►
From YouTube: CEO interview for blog posts
Description
Valerie Silverthorne and Sid about multi-cloud computing
-- If refer to https://medium.com/gitlab-magazine/multi-cloud-maturity-model-2de185c01dd7 proposal first and OSS maintainability
A
Talk
to
see
the
idea
is,
is
to
look
at
like
hey.
How
frequently
do
people
deploy
and
what
programming
languages
are
they
using
and
can
we
find
correlations?
For
example,
I'd
find
I
expect
to
see
that
people
programming
in
C
deploy
less
often
than
people
programming
in
JavaScript,
not
sure.
But
if
that's
the
case,
we
you
can
write
article
about
that.
Basically,
programming
languages
are
popular
subjects.
A
A
A
B
A
In
a
perfect
world,
they
would
want
application
portability,
but
it's
too
expensive
to
kind
of
rewrite
applications
to
to
be
able
to
deploy
to
multiple
clouds,
and
it's
even
there's.
There's
pressure
marks,
whether
that's
even
worth
it.
What
is
clear
is
that
they
will
have
to
use
multiple
clouds,
so
at
least
their
processes
should
be
portable
should
be,
should
be
able
to
use
the
same
process
with
respective
of
where
you
go
to
is
that
answer.
The
question
feel
free
to
yeah.
B
A
So
multi
cloud
is:
it
has
multiple
layers
as
reference
in
the
article
and
I
think,
for
example,
data
portability
where
data
changes
can
happen
in
multiple
chatter.
Clouds
is
still
kind
of
a
pipe
dream,
I
think
there's
novel
I
think
there's
hey
if
you
grounded,
if
you
make
it
a
round
number
there's.
Zero
percent
of
customers
is
now
running,
something
like
cockroach
DB
across
multiple
clouds,
so
I
think,
like
data
portability,
is
still
very
far
away.
I
think
another
dimension
is
that
the
clouds
kind
of
all
looked
the
same.
A
A
You
could
even
see
a
future
where
there's
stuff
running
in
the
cell
phone
towers
right
now,
there's
not
a
lot
of
computing
cell
phone
towers,
but
that
might
change
in
the
future,
and
that
might
give
you
something
not
that
you're
going
to
do
a
lot
of
computer,
but
at
least
you're
running
on
kind
of
trusted
infrastructure
or
the
device
of
a
user
is
always
interested
to
use.
You
could
manipulate
that
the
cell
phone
tower
would
be
a
place
where
you
get
super
low
latency,
but
you
can
run
on
trusted
hardware
that
they
use.
B
A
The
interesting
thing
is
right
now
that
would
5g
is
that
it
seems
that
the
the
best
applications
or
the
most
speed
you
gain
is
is
with
direct
line-of-sight,
which
is
interesting,
which
would
mean
we
have
to
have
a
bunch
more
antennas,
but
also
one
of
the
interesting
things
is
fine.
She
just
doesn't
just
improve
download
speeds
and
the
upload
speeds,
but
it
also
greatly
reduces
latency
compared
to
4G
and.
B
A
Might
be
as
big
of
a
contribution
as
the
the
speeds,
the
the
capacity
that
it
improves
and
I'm
I'm
really
encouraged
that,
for
example,
cloud
gaming
is
now
becoming
a
thing
where
the
latency
on
kind
of
fiber
connections,
cable
connections,
is
getting
low
enough.
That
you
could
have
to
compute
gaming
being
offloaded
that
makes
5g
might
enable
that
for
mobile
and
for
sure,
if
I
he
might
replace
wired
connections.
Where
there's
going
to
be
a
few
up
links
every
from
the
cell
phone
tower
up.
A
B
So
when
it
comes,
though,
when
we
think
about
you
know
sort
of
all
of
this
going
on
and
then
we
think
about
the
process
of
you
know
software
development,
you
know
concurrent
DevOps,
like
do
you
kind
of
see.
Is
there
a
sweet
spot
where
you
think
this
is
just
gonna
all
and
let's
throw
in
maybe
a
I
or
machine
learning
in
there
to
where
this
is
just
all
gonna
converge,
I'm,
just
I'm
cute
I'm,
making
you
put
your
future
hat
on
today,
I'm!
Sorry,
that's.
B
It
just
sort
of
as
companies
as
we're
with
the
backdrop
of
gee
and
multi-cloud
and
sort
of
the
changes
going
on
in
with
some
of
the
the
hardware
how
I
guess
I'm
what
I'm
wondering
is.
What
do
you
see
this
meaning
for
sort
of
software
development?
Concurrent
DevOps
today
like?
How
is
there
a
is
there?
Are
there
gonna
be?
Are
they're
going
to
have
to
be
some
sort
of
sweeping
changes
as
we
go
forward
to
sort
of
accommodate
some
of
these
breakthrough
technologies.
B
A
Know
I
think
there's,
there's
gonna
be
more
moving
parts
instead
of
how
you
used
to
talk,
you
used
to
have
the
ploy
software
to
almost
a
single-serving
and
now
you're
deploying
it
to
probably
multiple
servers
at
scale
up
and
down
automatically
in
a
cloud
from
there.
It's
gonna
go
it's
not
a
single
Club
of
multiple
clouds
from
there
it's
gonna
go
well,
it's
not
just
those
hyper
clouds,
it's
also
the
CD
ends
and
cell
phone
towers
and
everywhere
else.
B
A
A
You're
doing
much
more,
like
everything
is
trending
to
kind
of
integrated
open
source
where
you
in
deploying
application
and
then
specify
all
the
dependencies
as
well.
I
wonder
whether
the
managed
services,
like
a
database
service
like
RDS,
whether
that
will
be
controlled
by
open
source
software,
and
that
will
be
something
like
cross
fling,
which
we
recently
had
an
announcement
at
and
our
supporting
it
lab
or
that
the
manages
managed
services
will
be
replaced
by
kind
of
kubernetes
operators.
That
do
the
same
thing,
but
then,
just
by
using
the
basic
primitives.
A
B
In
the
context
of
I
know
that
we
just
moved
to
Google
cloud
platform
any
and
we
just
we,
you
know
we
did
an
interesting
blog
post
about
it.
Curious,
though,
from
your
perspective,
any
thoughts
for
gitlab
customers
or
people.
Reading
this,
when
they're
kind
of
going
through
this
sort
of
how
do
we
choose?
How
do
we
move?
How
do
we?
How
do
we
sort
of
manage
our
sort
of
our
presence
in
the
cloud
anything
you
walked
away
from
that
experience
with.
A
B
B
A
A
I
think,
right
now,
it's
workflow
portability
makes
a
lot
of
sense
that
all
your
teams
are
working
in
the
same
way
that
you
can
measure
results
across
the
company
and
that
the
people
can
join
other
teams
without
a
hassle,
I
think
application
portability.
The
verdict
is:
do
it
as
much
as
it's
needed.
So
if
you
have
an
application
that
is
very
essential
and
expensive,
you
can
get
in
a
better
negotiation
position
by
making
by
proving
that
it's
portable,
that
might
be
worth
the
effort,
but
for
a
lot
of
applications.
A
It
might
not
be
worth
the
effort
to
build
on
somewhere
else
and
in
order
to
kind
of
make
the
best
decisions
for
the
future.
When
you
have
the
option,
build
something
on
an
open
source
technology
instead
of
a
closed
source
technology,
for
example,
instead
of
using
Amazon
Kinesis,
you
use
Amazon's
Kafka
offering
so
that
you
can.
If
you
end
up
having
to
move
it
to
another
cloud,
you
don't
have
to
refactor
that
code.
A
You
had
a
great
question
like
in
there
I
think
about
get
labboard.
That's
how
I
interpreted
it
it's
interesting
to
see
what
we
do
get
live.
We
have
workflow
portability.
We
we
are
getting
close
to
application
portability.
Well,
we
have
application
portability.
Get
lab
can
run
on
any
cloud
disaster.
Recovery
portability
is
harder
like
we
can
fail
over
to
another
cloud,
but
right
now
that
takes
kind
of
too
much
time
to
to
be
a
super
feasible
option.
We
do
have
workload
portability
as
in
RC.
A
B
Okay,
you
talked
about,
and
this
is
one
I'm,
not
I'm,
gonna
be
really
looking
to
you
for
guidance.
Here.
Tell
me
what,
where
you
were
headed
with
this
whole
sort
of
mr
instead
of
issue
first,
and
what
what's
the
message
again,
this
may
or
may
not
be
a
blog.
It
was
in
that
list,
so
so
yeah.
What
were
you
thinking?
Yeah.
A
A
Interpreting
the
description
of
the
issue,
lended
itself,
so
multiple
interpretations
of
the
issue
and
or
the
subject
at
hand,
and
also
just
by
its
nature,
if
you
describe
a
problem,
there's
all
kinds
of
ways
to
solve
it,
and
what
we
found
is
that
there
was
a
lot
of
time
spent
on
submitting
different
ideas.
So
you
create
an
issue
like
this
is
a
problem.
There
were
all
kinds
of
ideas
submitted
and
it
was
almost
a
route
to
say:
okay.
A
Well,
this
is
the
small
change
we're
gonna,
make
and
and
thanks
everyone,
because
there
was
so
many
great
ideas
that
if
you
almost
felt
like
you
had
to
do
too
much,
but
our
values
are
like
iteration,
so
making
the
smallest
possible
step
and
also
we
have
a
concept
of
a
DRI,
a
directly
responsible
individual.
So
just
have
someone
make
that
call,
and
it's
not
it's
rude,
not
to
ask
for
suggestions,
but
it's
not
rude
to
ignore
suggestions
in
the
end
want
us
to
do
the
work
and
they
can
decide.
A
But
it's
it
didn't
a
lot
of
things
got
kind
of
stuck
like.
Yes,
there
was
a
problem.
Yes,
there
are
lots
of
opinions
about
both
what
the
problem
was
and
what
the
solutions
were,
but
it
it
didn't
quickly
result
in
a
change.
So
now
we're
suggesting
look
if
you
want
to
make
a
change
that
get
proposed
to
change.
Don't
say
this
is
the
problem.
This
is
a
possible
solution,
make
the
make
the
change
propose
to
change
in
the
in
the
handbook.
That
makes
it
much
easier
for
people
to
review.
A
It
makes
it
much
quicker
and
much
more
constrained.
People
are
not
going
to
come
up
with
500
ideas
to
also
do
something
in
the
same
realm,
because
it's
much
more
ok
is
this:
is
this
better
or
worse
than
it
it
used
to
be?
And
how
can
we
make
this
specific
thing
slightly
better,
but
it
constrains
that
the
scope
and
if
you
constrain
the
scope,
you
can
go
faster
and
that's
what
we're
trying
to
do
do
many
iterations
quickly
in
order
to
have
many
iterations,
you
need
to
make
them
very
small.
B
A
B
B
Let's
talk
about
this
I
you're
you've
brought
up
the
idea
that
early
adopters
and
and
sort
of
late
late
adopters
are
obviously
in
different
places
and
need
different
things.
So
can
you
talk
about,
maybe
start
it
back
up
slightly
and
how
would
you
describe
get
lab
now
me
an
early
adopter?
We
late
adopter,
we
in
the
middle
or
on
that.
Where
are
we
and
then
then
we
can
maybe
that's
a
good
place
to
start.
A
Yeah
I
get
love
us
now
in
in
its
adoption
curve.
It's
at
the
early
majority.
It's
no
longer
an
early
adopter,
okay
thing.
If
people
are
not
reaching
out
to
get
lab
for
the
sake
of
trying
something
new
people
are
reaching
out
to
get
lab
because
it's
it
makes
a
lot
of
sense
and
they
they're
getting
the
results
with
it.
A
And
this
was
based
on
the
open
source
infrastructure,
white
paper
and
the
link
is
broken,
but
I
hope
we
can
find
it
somewhere
else,
and
it
basically
says
that
open
source
projects
have
a
a
cycle
to
them
and
in
the
beginning
you
have
early
adopters
those
tend
to
adopt
new
technology
for
almost
for
the
sake
of
adopting
new
technology,
and
they
are
very
passionate.
They
have,
they
are
very
intelligent
and
they
are
very
I.
A
Want
to
say
self-managed,
but
they
they
can
figure
out
stuff
on
their
own
much
better
than
others.
So
if
I
went
to
this
phase
and
greatly
expanded,
then
when
you
get
to
early
majority,
people
are
much
more
pragmatic,
so
they
will
more
quickly
ask
a
question
without
like
trying
to
dive
into
the
source
code
to
find
the
answer
themselves.
They
are
less
likely
to
answer
all
the
people's
question
have
a
job
to
do.
They
just
need
to
get
the
job
done.
They're
not
here,
for
hanging
out
or
for
the
social
community.
A
We
don't
have
enough
people
to
kind
of
deal
with
every
single
question,
but
we
do
have
people
that
are
helping
to
triage
issues
and
we
got
em
our
coaches
that
are
trying
to
help
Mercer
crest
over
the
finish
line.
If
there
wasn't
a
company
around
gilt
lab
the
outside
that
we
that
these
activities
wouldn't
be
done
and
that
the
project
would
kind
of
get
stale
and.
B
B
B
When
you're
moving
towards
the
late
adopter
stage,
I
guess
you
know
what
you
talked
to
that
for
the
later
or
adopted
you
know,
we've
got
people
staff
who
could
help
push
things
along.
What
else
do
you
need
in
that
stage,
when
you
you've
become,
you
know
like
a
Xerox
or
a
Linux,
or
you
know
it's
more
of
a
household
name?
What
are
the
other
pieces
that
need
to
be
there
yeah.
A
So
I
think
we're
still
in
the
early
majority
stage,
but
that's
later
than
the
early
adopter
stage
and
I
think
you
need
more
people
who
can
answer.
Questions
and
people
are
prepared
to,
for
example,
pay
money
like
so
a
lot
of
people
pay
us
money
because
they
want
to
support
contract,
you
need
better
documentation
and
the
document
documentation.
It
needs
to
be
more
comprehensive.
You
need
better
marketing
materials
not
only
to
convince
people,
but
also
to
allow
people
to
convince
their
boss
or
their
teammates.
How
to
do
something
you
need
these.
A
You
know
merge,
request,
coaches
to
get
stuff
over
the
finish
line
you
need
more
maintains
people
are
able
to
do
code
review
things
like
that.
You
need
more
compliance
by
people.
People
need
to
have
like
federal
compliance
for
the
product
and
things
like
that
to
be
able
to
adopt
in
so
those
are
areas
in
which
we're
making
investments.
A
When,
when
we're
not
answering
every
question,
so
like
there's
an
IRC
channel
about
gillip
and
has
a
used
to
have
a
lot
of
questions-
and
we
didn't
did
that
was
staffed
by
volunteers,
who
did
a
great
job,
didn't
have
the
capacity
to
answer
everything?
We're
not
answering
any
every
issue
about
get
laughs
like
they're
still
issues
on
shatrak?
Oh,
we
just
can't
keep
up
with
demand
and
we're.
Not
we
don't
have
people
dedicated
to
like
answering
get
live
things
on
Stack
Overflow,
for
example.
So,
there's
a
lot
with
that.
We're
not
doing.
A
And
on
one
hand,
I'd
like
to
do
more,
on
the
other
hand,
like
as
a
company,
you
have
to
make
these
trade-offs.
Like
are
you?
How
is
that
worth?
While
for
the
company
to
investing-
and
probably
it's
never
gonna-
be
that
you're
gonna
answer
every
last
question,
especially
not
at
the
size
of
popularity
of
get
line.
B
Is
this
one
of
that
I
hesitate
to
use
the
word
downside
because
it's
kind
of
strong,
but
is
this
one
of
the
the
sort
of
unique
challenges
of
building
a
business
off
of
open
source
that
you've
really
got
a
lot
of
opinions
and
a
lot
of
contribution?
B
A
I
think
I
think
it
is.
It
is
one
of
those
things
and
you
have
seen
a
lot
of
blog
posts
recently
by
open-source
maintainer
hsihu
were
just
struggling
with
burnout,
they're
things
like
that
and
how
they
deal
with
that,
and
there
was
a
good
one
from
the
maintainer
of
Redis
owner
is
Elias
anteras,
who
wrote
about
that
and
there's
a
couple
of
more
open-source
maintainer
and
some
of
them
have
been
very
successful
and
kind
of
Senate.
For
example.
Leanest
revolt
is
what
he
calls
lieutenants
that
helped
him
to
deal
with
the
great
influx.
B
B
A
The
first,
the
first
thing
is
for
like
maintain
years
to
be
able
to
dedicate
all
the
time
to
what
they're
what
they
want
to
dedicate
it
to
so
when
they
asked
they
meet
you,
my
co-founder
his
wife,
like
hey,
are
you
okay
with
me?
Hiring
Demetri
and
I
realized
that
and
I
start
up
with
a
lot
of
words
he's
like
no,
it's
great.
A
It's
been
a
lot
more
success
of
that,
for
example,
patreon
has
allowed
Evan
the
creator
of
few
to
quit
his
job
I
focus
on
that
there's,
an
initiative
by
the
Linux
Foundation
to
provide
fred
and
github
recently
announced
sponsors,
and
it's
also
an
initiative
called
get
coin
and
there's
a
few
others
to
all
to
kind
of
allow
for
open-source
contributors
to
to
get
paid.
So
that's
a
very
exciting
development.
A
There's
also
a
thing
called
like
paying
for
features
where,
if
you
want
to
feature
you
put
a
kind
of
a
bounty
on
it
and
then
hope
that
opens
or
then
people
contribute
that-
and
it's
just
like
that
are
called
things
like
bountysource.
We
doubled
in
that
price,
but
a
lot
of
the
work
is
in
kind
of
defining
what
something
is
and
then
assessing
the
contribution
and
then
reviewing
that
and
fixing
that
and
making
sure
it
gets
in.
A
A
A
If
you
do
it
in
an
open
court
model
like
us,
like
the
downside,
is
that
no
longer
all
the
code
of
your
project
be
open
source,
so
the
upside
is
that
you
might
be
able
to
create
a
lot
more
due
to
having
a
business
model
and
with
open
source.
It's
always
about
like
how
much
how
much
like
value
do
you
generate
and
how
much
value
do
you,
capture
and
I?
A
Think
for
keeping
a
completely
open
source
would
mean
you
have
a
very
low
capture
rate
and
therefore
you
need
to
create
an
enormous
amount
of
value
in
order
to
build
a
like
a
substantial
business
around
it
and
with
open
core
you
capture
a
bit
more
of
the
value,
but
you
can
that's.
That
would
enable
you
to
kind
of
do
a
lot
more
than
you
would
otherwise
be
able
to
do
on
a
smaller
and
therefore
kind
of
generate.
A
More
value,
therefore
generate
more
value
than
you
would
otherwise
be
able
to
do,
but
I
think
pure,
open
source
approaches
all
right,
most
appropriate.
If
it's
like
a
piece
of
infrastructure
that
every
single
person
is
gonna
need
like
it's
it's
becoming
us,
it
will
become
a
total
standard,
I
think
as
soon
as
it's
gets
a
bit
more
in
the
application
space
yeah.
What
needs
an
interface
and
everything
else,
it's
gonna
have
certain
market
share,
but
not
a
hundred
percent.