►
Description
Want to view more sessions and keep the conversations going? Join us for KubeCon + CloudNativeCon North America in Seattle, December 11 - 13, 2018 (http://bit.ly/KCCNCNA18) or in Shanghai, November 14-15 (http://bit.ly/kccncchina18)
Keynote: Software's Community - Dave Zolotusky, Software Engineer, Spotify
We talk about community software and about how companies are adopting it and using it, but we often ignore the community around that software in those conversations. As Spotify has ramped up its use of Kubernetes and other cloud native software, we have also been working on involving ourselves more in the community around that software. Dave will talk about how Spotify is changing its approach and culture around infrastructure to be more open and to be more involved in the infrastructure software community.
About Dave
Dave Zolotusky is a software engineer on Spotify's Infrastructure and Operations team in Stockholm, Sweden. His team builds and maintains the core platform that the rest of Spotify runs on. Before joining Spotify, Dave worked at Amazon Web Services, VMware, and Microsoft.
Want to view more sessions and keep the conversations going? Join us for KubeCon + CloudNativeCon North America in Seattle, December 11 - 13, 2018 (http://bit.ly/KCCNCNA18) or in Shanghai, November 14-15 (http://bit.ly/kccncchina18)
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
So
initially
I
was
going
to
come
up
here
and
talk
to
you
about.
Spotify
is
infrastructure
transition
to
kubernetes
into
cloud
native
apps,
and
then
I
realized
that
I
was
going
to
say
about
80%
of
the
same
thing.
Sara
did
and
used
the
exact
same
architecture
slides
so
instead
I
want
to
talk
about
a
different
transition
that
we're
trying
to
lead
at
Spotify
and
that's
transitioning
to
being
a
really
strong
active
member
of
this
community
and
to
try
to
push
all
the
rest
of
you
guys
to
do
the
same
thing.
A
So
I
think
we
all
agree
that
free
and
open-source
software
is
a
really
amazing
thing,
but
at
least
in
my
opinion,
most
of
us
are
pretty
bad
at
it
and
I
say
that,
because
I
think
the
greatest
thing
about
free
and
open-source
software
isn't
necessarily
contributing
upstream
and
the
code
itself,
or
necessarily
even
adopting
all
the
technology.
It's
really
engaging
with
this
community
and
I,
don't
think.
Spotify
is
necessarily
any
better,
and
it's
probably
worse
than
many
of
the
companies
here,
but
we're
trying
really
hard
to
get
better
and
to
actively
commit
to
getting
better.
A
So,
first
a
little
bit
about
Spotify.
We
don't
really
do
open
software,
we're
not
red
hat
or
a
core
OS
type
company
whose
kind
of
core
business
is
in
open
source
software.
We
do
highly
personalized
streaming
music.
We
have
very
autonomous
teams,
hundreds
of
them
across
the
company,
thousands
of
services
and
thousands
of
engineers
and
hundreds
of
people
in
a
technology
platform
team
that
is
supposed
to
support
all
of
that
and
create
the
tools
and
systems
and
structures
that
everyone
in
the
company
uses.
A
So
we
have
very
loved
rough
alignment
across
the
org,
but
no
real
authority.
There's
on
someone
that
can
just
kind
of
put
their
foot
down
and
say
you
must
all
go
I,
don't
know
blog
or
tweet
or
go
to
conferences.
It's
really
up
to
every
one
of
those
individual
teams
to
decide
what
they
do
and
how
active
they
are
about
it.
But
my
team
and
infrastructure
and
operations
at
Spotify
in
those
couple
hundred
people
on
technology
platform
is
trying
really
hard
to
kind
of
be
a
trendsetter
across
the
company
and
engaging
one
thing.
A
We've
been
doing.
A
lot
of
right
now
is
moving
to
cloud
and
cloud
native
technology.
Of
course,
we
kind
of
forklift
that
all
of
our
data
center
software
into
cloud
and
now
we're
really
figuring
out
what
the
next
steps
are,
we're
working
on
our
gauging
of
the
community,
and
we
have
a
whole
lot
of
kind
of
problems
inside
our
company
that
I'm
sure
very
similar
to
what
you
guys
have
of
aligning
with
on
techno
and
moving
much
faster
and
scaling.
A
But
the
big
question
for
us
was
really
kind
of
how
do
we
get
started
with
this?
How
do
we
get
started
engaging
with
people?
How
do
we
get
started
being
a
part
of
this,
and
we
had
a
lot
of
people
in
the
company
that
had
done
very
interesting
things
internally,
but
at
any
given
moment
felt
like
I,
haven't,
really
done
anything
that
exciting?
A
A
Maybe
there's
someone
in
another
office
or
somewhere
down
the
street,
that
a
similar
thing
still
within
Spotify,
but
is
cooler
so
like
that
person
should
talk
instead
of
me
and
then
we
really
didn't
know
who
to
engage
with
I,
mean
I,
say,
engage
with
the
community,
but
who
or
what
is
kind
of
the
community.
So,
for
example,
when
we
were
starting
to
work
with
kubernetes,
we
really
wanted
to
talk
to
some
customers
that
had
done
some
of
the
things
that
we
were
doing
or
that
were
kind
of
experts
on
it.
A
So
we
could
ask
them
how
they
did
it,
how
they
sell
some
listening
problems,
but
we
didn't
know
who
to
ask:
we
didn't
know:
can
I
just
pick
up
the
phone
and
call
some
random
company.
That
I
know
is
using
kubernetes
and
have
them
actually
answer
the
phone
and
talk
to
us
and
of
course
the
same
thing
for
everybody's
other
thing
is
whether
it's
gr,
PC
or
sto,
or
monitoring
tools
like
Prometheus
or
anything
else.
A
Of
course,
we
have
our
existing
providers,
our
cloud
provider,
our
existing
on-prem
infrastucture
providers,
all
of
that,
but
they're
still
not
necessarily
the
right
people
to
be
able
to
connect
us
to
the
best
users
of
all
of
these
things.
So
the
place
where
I
started
is
really
just
kind
of
just
sharing.
A
That
would
love
to
talk
to
us,
but
even
before
this,
there
were
really
a
couple
of
things
that
convinced
me
that
we
absolutely
have
to
do
this
and
to
me
it
really
fell
into
these
kind
of
three
obvious
buckets
of
getting
support,
getting
guidance
and
really
just
figuring
out.
What's
out
there
and
I
have
some
kind
of
some
stories.
That
really
convinced
me
on
how
great
this
is.
A
The
first
thing
for
me
in
this
community
was
starting
to
play
with
with
kubernetes
mid
last
year
with
one
of
our
customer
teens,
and
they
really
wanted
to
figure
out
a
way
to
get
an
ingress
controller
that
can
do
ingress.
Two
pods
running
in
two
different
namespaces
and
most
ingress
controllers
can't
do
this,
but
we
did
find
that
voyager
can
do
this
where
at
least
voyager
was
documented
field
to
do
this,
but
we
could
make
it
work
to
save
our
lives.
And
of
course
we
did.
The
normal
developer
thing
like
go
on.
A
Google
look
around
go
on
Stack
Overflow,
look
around
dig
around
github
issues
on
the
open
source
project,
and
none
of
that
really
solves
our
problem.
So
we
do
this
very
unnatural
thing
of
jumping
on
their
slack
Channel
and
just
asking
people
and
within
about
five
or
ten
minutes
we
walked
through
on
slack
in
real
time.
What
the
problem
was
was
someone
for
maps
code
and
they
just
walked
us
through
it
and
it
turned
out.
A
It
was
completely
our
fault
and
we
had
our
back
set
up
wrong
and
that's
why
nothing
was
working,
but
we
have
no
support
channel
with
them.
We
have
no
prior
relationship
with
them.
They
have
no
idea
who
any
of
us
are,
but
they
just
happily
I-
am
with
us
and
fix
this
whole
thing
and
to
me
coming
from
a
place
where
I
had
worked
a
lot
of
enterprise
solutions
and
work
through
support
channels,
and
all
of
this
this
was
amazing.
A
So
then
I
really
kind
of
I
really
wanted
to
do
more.
There
were
a
whole
bunch
of
places
where,
as
opposed
to
this
example,
we
just
really
didn't
know
what
to
do
or
what
exact
technology
to
use.
So
we
just
needed
guidance.
The
most
recent
example
is
for
me
was
compliance.
Things
like
we
have
a
lot
of
services
that
are
under
IP
general
control
compliance
requirements,
and
we
really
want
to
get
them
running
in
kubernetes,
but
I
couldn't
easily
find
anyone
out
there.
A
That
was
doing
similar
things
to
what
we
were
doing
running
them
in
kubernetes
and
running
them,
with
ittc
compliance,
so
I
just
effectively
kind
of
emailed.
Whoever
I
could
find
whether
it
was
a
CN,
CF,
end
user
community
or
some
other
Cobra
DS
users
just
to
feel
out
what's
out
there
and,
on
the
one
hand,
I
don't
have
a
great
answer
like
somebody
or
their
emailed
me
and
said:
here's
our
magical
tool.
That
does
everything
you
need
we're
still
in
a
place
where
frequently
what
to
do.
A
Even
know,
what's
out
there
like
in
the
monitoring
space,
of
course
we're
familiar
with
things
like
Prometheus,
but
we
have
a
whole
bunch
of
interesting
monetary
requirements
that
span
our
kind
of
scale
and
with
the
way
that
our
work
is
structured,
where
it
didn't
feel
like
just
deploying
Prometheus
the
way
that
it
kind
of
exists,
as
the
open-source
project
was
going
to
work
for
us
and
I
really
wanted
to
find
someone
that
kind
of
Anu
and
I
flow
from
a
theist.
That
would
walk
me
through
it.
A
But
again,
I
didn't
feel
like
I
could,
just
if
email,
one
of
the
maintainer
Zoar,
someone
that
really
runs
prometheus,
so
I
tried
to
feel
out
who
I
could
talk
to
in
an
email
some
of
my
connections.
They
kind
of
immediately
responded
and
just
connected
me
to
Julius
and
I,
had
this
moment
where
it's
like
well,
but
he
does
like
important
Prometheus
things
with
real
people
that
are
using
it.
A
Why
would
he
talk
to
me
and,
of
course,
we
exchanged
a
whole
bunch
of
very
detailed
emails,
but
he
gave
us
a
lot
of
great
tips
on
what
we
couldn't
could
do
and
what
he
would
recommend
and
all
from
just
kind
of
reaching
out
and
trying
to
go
get
somewhere
at
that
point.
I
was
really
hooked
at
that
point,
I
said
we
absolutely
have
to
be
a
part
of
this.
We
have
to
use
this
across
the
company.
A
A
We
do
a
regular
off-site
among
our
platform
team
and
I
just
picked
a
basically
just
looked
up
the
CNC
F
ambassadors
Paige
found
Paul
Bert
from
Chora
less
on
that
list,
because
he
was
in
New
York
near
where
we're
having
the
the
off-site
and
I
just
emailed
him
and
said:
hey
I'd
love
to
have
someone
that
could
talk
about
CNC
F
to
our
platform,
and
he
was
excited.
He
was
glad
to
do
it.
A
A
We're
all
facing
very
similar
problems,
we're
all
building
debugging,
deploying
operating
containerized
micro
services
at
some
scale
sure
we
have
different
scale
requirements.
We
have
slightly
different
software,
but
at
the
end
it's
very
similar
problems
and
chances
are
whatever
problem
is
keeping
you
up
at
night
right
now,
where
you're
doing
the
most
work
on
there's
lots
of
people
here,
but
at
least
someone
here
working
on
similar
things,
and
we
just
have
to
find
those
people
and
reach
out
to
them
and
work
on
those
things
together.
A
Rather
than
working
on
them
on
our
own
every
time
and
not
isolating
ourselves,
and
even
if
you
are
that
one
person
that
has
some
problem
that
no
one
else
is
working
on,
you
might
just
become
a
thought
leader
in
that
area.
As
soon
as
the
next
group
of
people
hit,
that
problem
and
I
realize
that
you've
already
solved
it
and
been
fairly
public
about
it.
A
A
We
are
hundreds
of
millions
of
users,
tens
of
thousands
of
data
pipelines
per
day,
multiple
mobile
apps
thousands
of
services,
but
at
the
end
of
the
day,
it's
still
we're
figuring
out
how
high
we
can
scale
a
cluster.
What
we
can
do
for
Federation
across
multiple
clusters,
how
we
convert
our
old
configuration
management,
a
system
that
works
cloud
native
all
of
the
same
kind
of
problems.
Everyone
else
talks
about,
and
we
have
a
long
history
of
kind
of
attempting
to
engage
with
the
community
of
open
sourcing.
A
A
The
CNC
F
is
the
first
foundation,
we've
really
joined
and
we've
been
a
fairly
large
enough
name
and
software
for
years
and
to
me
that's
because
we're
all
mostly
engineers
and
for
us,
the
software
thing
is
really
easy,
but
talking
to
people
is
really
hard
and
we
need
to
fix
that
because
if
we
talk
to
people
we
can
make
the
building
software
parts
significantly
easier
and
the
some
of
the
numbers
that
we
just
saw
for
Cloud
Foundry,
actually,
we've
seen
very
similar
things.
Internally.
A
We've
done
a
few
surveys
internally
over
the
past
few
years
and
be
very
often
hear
from
a
lot
of
individual
engineers
that
they
really
want
to
contribute
more
or
they
think
Spotify
doesn't
contribute
enough
to
the
ecosystem.
But
we
and
I
don't
know
how
unique
this
is,
but
we're
in
a
place
where
most
of
our
management
at
all
levels
is
extremely
supportive
of
us
being
active
in
the
community
and
us
joining
and
us
reaching
out
and
us
speaking
or
blogging,
or
just
talking
to
more
people
learning
more.
A
We
have
a
core
kind
of
cultural
value
of
learning
from
others
in
the
industry.
So
a
lot
of
that
actually
giving
back
and
contributing
is
on
us
individuals,
the
same
individuals
that
are
answering
these
surveys
and
saying
that
Spotify
is
not
giving
back
enough
and
it's
easy
for
the
leaders
on
the
team
to
get
that
survey
and
then
allocate
money
to
donate
to
some
foundation
or
some
group.
But
it's
much
harder
for
the
leadership
to
kind
of
get
all
the
individuals
to
suddenly
be
more
active
and
learn
more
from
others.
A
But
that's
really
where
we
are
now
and
I
think
we
can
all
do
better.
I
know
spot
if
I
can
do
better,
but
I
think
we
can
all
do
better
and
I
use
this
analogy
of
raising
your
hand.
But
we
can
call
it
sitting
at
the
table
or
speaking
up
or
we
want
to
call
it
because
to
me,
I
think
all
the
same
people
that
answer
in
those
surveys
that
they
want
Spotify
to
give
back
more,
are
the
same
exact
people.
That
really
would
be
the
ones
giving
back
but
they're,
afraid
they're
nervous.
A
They
say
that
same
thing.
If
I
haven't
done
anything
I
mean
we
have
people
that
have
built
our
own
internal
monitoring
system
that
monitors
the
incredible
scale
that
I
just
listed
to
you
that
don't
really
feel
like
there's
much
to
talk
or
write
about
there
where
they've
built
us
amazing
system
and
I.
Think,
even
if
it's
not
the
greatest
monitoring
system
in
the
world,
it's
a
really
interesting
thing
for
people
to
hear
about
so
to
them
and
kind
of
anyone.
Else.
A
Be
that
first
one
to
reach
out
or
try
to
answer
someone's
question.
If
you're
say
at
a
conference
like
this,
and
you
hear
an
interesting
talk
and
someone
asks
something
that
you
know
you
can
answer
connect
with
them,
maybe
they'll
suddenly
change
the
way
you
work
on
this
problem
for
a
long
time,
because,
like
I
said
it
really
is
about
a
lot
more
than
just
contributing
upstream
or
working
in
the
open,
the
easy
things
that
you
can
do
by
just
interacting
with
github.com.
A
It's
really
about
attracting
with
all
the
people,
especially
all
the
people
in
this
room,
while
you're
here
and
I,
really
want
to
hammer
on
this
point
the
same
to
haven't
done
anything.
Yet
it's
really
just
imposter
syndrome,
like
I,
said
we
all
know
the
warts
in
our
systems.
We
know
the
mistakes
we've
made.
We
know
the
places
where
it
really
hasn't
worked.
A
We
know
the
moments
where
we
think
I
don't
belong
there.
I
don't
belong
on
that
stage.
That's
the
stage
Kelsey
Hightower
was
on
yesterday.
Who
am
I
to
share
a
stage
with
Liz
Rice,
but
you
probably
have
done
plenty
of
interesting
enough
things
for
people
to
want
to,
and
the
same
thing
goes
for
any
other
medium,
whether
it's
medium
or
Twitter
or
anything
else.
Everyone
does
this
and
there
are
millions
of
cartoons
making
the
same
point-
and
this
to
me
is
the
biggest
example.
This
is
lit
asses
email,
where
he
first
released
Linux
and
I.
A
A
On
the
other
hand,
if
you
listen
to
Red
Hat's,
command-line
heroes,
podcast
you'll,
hear
people
you'll
hear
Saranya
barik
say
that
he
was
literally
a
real-life
Luke
Skywalker
that
was
taking
down
the
Empire
with
this
one
message,
and
you
never
know
if
the
thing
you're
releasing
the
thing
that
you
think
is
it
probable
might
change
the
world
for
thousands
of
people
so
again
going
back
to
that
starting
line.
I
think
we
can
do
better.
I
think
we
need
to
do
better,
but
I
think
the
starting
line
that
I
used
initially
isn't
quite
the
right
edge.
A
On
the
right
analogy
to
me,
it's
more
of
a
diving
board
once
you
jump
in
you're
in,
and
you
get
all
you
get
so
deep
in
this
community
that
you
can't
give
it
up.
Like
I,
said:
I
got
hooked
and,
of
course,
being
Spotify
I
have
to
bring
this
back
to
music.
Last
week,
I
was
in
a
Spotify
conference
and
we
had
someone
interviewing
a
musician,
the
producer
actually
and
she
was
asked
she
was
talking
a
little
bit
about
how
she
got
some
of
the
most
interesting
gigs.
A
She
worked
on
some
of
the
most
interesting
meetings
that
she
that
she
had
some
of
the
most
interesting
performances.
She
did
really
with
some
of
her
idols
people
that
she
wasn't
sure
she
could
ever
work
with
or
that
ever
really
would
have
wanted
to
work
with
her
and
the
message
that
she
gave
us
at
that
point
was
really
just.
She
had
the
guts
to
ask.
She
had
a
person
that
she
idolized
and
she
sent
them
an
email
and
said:
hey
I
want
to
work
with
you
or
some
event.
A
That
was
happening,
and
she
said:
hey
I
really
want
to
be
a
part
of
this
event
and
more
often
than
not
just
that
one
moment
of
reaching
out
and
saying
she
wanted
to
be
a
part
of
something
got
her
into
it.
So
that's
what
I
want
to
leave
you
with
I
think
like
her,
we
just
need
to
have
the
guts
to
ask
and
to
ask
everyone
else
here.
Thank
you.