►
From YouTube: Keynote: The Future of Continuous Delivery - Panel
Description
For more Continuous Delivery Foundation content, check out our blog: https://cd.foundation/blog/
Keynote: The Future of Continuous Delivery - Sarah Novotny, Microsoft; Hong Wang, Akuity; Isaac Mosquera, AWS & Tim Jacomb, Kainos; Moderated by Fatih Degirmenci, Continuous Delivery Foundation
A
B
I'd
be
happy
to
I'm
a
long
time
open
source
nerd.
I
lead
open
source
strategy
for
microsoft
and
I'm
reporting
to
the
azure
office
of
the
cto,
but
for
fun.
I
get
to
play
with
lots
of
open
source
communities,
including
many
here
today
are
here
this
week
as
well
as
many
that
will
be
back
here
in
austin
in
three
weeks.
C
E
A
Thanks
all
so,
in
order
for
us
to
start
talking
about
future,
maybe
we
should
perhaps
talk
about
today,
state
of
affairs
and,
as
we
have
been
talking
about
conductor
foundation,
released
the
state
of
cd
report
today,
and
there
are
interesting
findings
in
it
and
overall,
the
report
looks
pretty
good.
We
are
improving,
but
I
think
we
can
improve
more.
So
what
are
your
thoughts
on
the
challenges
organizations
are
facing
and
how
we
can
address
them?
B
Can
I
can
start,
I
think
frequently,
the
largest
challenge
is
adoption
and
the
cultural
change
necessary
for
this,
because
tools
are
very
important,
but
when
you
have
engineers
or
leaders
who
have
grown
up
in
the
industry
equating
deployments
with
faults
and
risk,
it's
a
it's
a
big
change
to
lift
and
move
through
it.
It
doesn't
it
isn't
as
easy
as
which
tool
should
we
fix
which
tool
fits
our
needs,
which
tool
should
we
choose.
C
Well,
yeah,
in
my
opinion,
I
think
nowadays
it's
the
infrastructure
is
getting
way
complicated.
We
are
talking
about
vm
in
the
private
data
center
we
are
talking
about.
Kubernetes
is
a
buzzword
of
everywhere,
and
we
also
talking
about
the
serverless
and
you
just
get
it
like
overly
complicated.
Then
there
are
so
many
tools
addressing
different
problems
and,
in
the
end
like
how
you
can
deliver
the
unified
experience
for
your
developers.
D
Yeah
well
first,
I
agree
with
that's
what
the
first
two
speakers
said.
D
But
but
to
add
on
that,
I
think
there's
also,
I
think,
compliance
and
regulation
and
how
that
was
applied
in
the
world
of
vm
and
and
infrastructure
and
when
it
was
physical
and
trying
to
take
that
into
the
virtual
world
that
we're
in
now
when
we,
you
know,
you
mentioned
serverless
kubernetes.
All
of
that
mixing
kind
of
these
rules
and
regulations
that
were
created
20
years
ago
into
today's
world
is
very
hard
and
left
up
to
a
lot
of
interpretation,
and
you
see
a
lot
of,
I
think
you
know,
sir.
D
You
were
talking
about
the
the
people
in
the
organizations
a
lot
of
like,
I
think,
conflict
and
discussion
around
how
to
move
forward
in
this
new,
more
virtual
world
versus
the
physical
world
that
we
kind
of
used
to
live
in
and
that
that
slows
down
progress
as
well.
E
Yeah,
I
think
I
think
sarah
put
it
very
well,
but
for
me
it
really
comes
down
to
people
and
processes.
It's
about
those
rules
that
were
probably
put
there
for
a
reason,
but
it's
about
putting
other
processes
in
place
and
automation
and
protection
and
safety
in
there
and
making
it
a
very
safe
culture.
E
That's
improved
over
time,
and
I
think
it's
also
really
important
to
get
to
have
buy-in
from
the
top
levels
to
have
it
sort
of
management,
understanding
and
helping
removing
barriers,
because
a
lot
a
lot
of
these
barriers
for
people,
it's
kind
of
their
job,
to
do
these
sort
of
things.
It's
how
it's
been
done
for
so
long,
and
it's
about
just
changing
that
job.
So
it's
not
about
removing
it,
but
it's
about
changing
it
into
what
it
can
do
now
in
the
future.
B
A
Thank
you.
I
think
you'll
mention
sarah
when
you
started
like
tools,
are
important
process
and
culture
more
important,
but
if
we,
if
we
go
back
to
technology
and
think
about
ci
cd
ecosystem
in
countries
their
foundation,
we
have
eight
projects
and
most
of
those
projects,
work
well
with
each
other.
But
the
users
have
to
integrate
those
technologies
to
each
other
and
that
could
that
be
considered
an
an
issue.
E
Yeah,
so
I
think
there's
so
many
like
ci
cd,
tooling
out
there
and
bringing
interoperability
and
will
allow
us
to
like
chain
these
together
in
a
better
way.
Historically,
all
the
city,
cicd
tours
have
been
very
different.
It'll,
be
interesting
to
see
what
interoperability
can
do
out
there
there's
quite
a
few
things
going
on
like
there's
the
cloud
events
to
help
change
those
events
through
and
help
these
tools
work
together.
E
Open
telemetry
is
a
great
example
of
a
success
in
this
area,
with
interoperability
with
so
many
vendors
able
to
work
together
and
instead
of
having
to
implement,
integrations
for
six
different.
Like
apm
platforms,
you
can
just
write
it
once
and
then
you
don't
have
to
write
another
one
and
then
recently
with
the
open
feature
management
there's
another
very
similar
one,
so
you're
very
interested
to
see
how
this
goes,
but
just
not
sure
currently.
C
Yeah
so
my
wheels
a
little
bit
different,
I
feel
the
direction
is
we
want
to
unify
the
infrastructure
to
some
extent,
so
I'm
from
the
kubernetes
perspective
more
that's
why
I'm
more
believer
of
the
kubernetes
will
be
the
linux
for
operating
system.
Basically,
when
you're
talking
about
multiple
clusters
or
clusters,
you're
talking
about
kubernetes
in
the
end,
so
as
aks,
eks
and
gke,
everything
is
so
readily
available
nowadays,
so
I'm
truly
believing
with
that
direction.
So
in
that
sense
I
feel
a
lot
of
infrastructure,
a
lot
of
the
companies.
C
When
you
are
thinking
about
problem,
you
are
thinking
more
a
little
bit.
You
can
be
more
thinking
about.
Bottom
up
is
what
is
your
future?
Are
you
betting
on
the
kubernetes
as
the
defective
solution?
Then
maybe
you
are.
You
are
picking
the
tools
towards
in
that
direction,
because
I'm
talking
about
more
about
kubernetes
native
more,
in
my
opinion,
is
it's
not
about
cloud
native.
It's
actually
kubernetes
native
means.
You
can
still
running
your
kubernetes
cluster
at
edge,
even
in
your
private
data
center.
There
are
so
many
distributions
like
certified,
confirmed
kubernetes
descriptions
there.
C
D
Well,
I
think
you
know,
I
think
we
do
see
a
lot
of
adoption
of
kubernetes,
but
I
think
what
makes
it
challenging
is
that
when
you
look
at
the
clouds
and
the
rate
that
they're
they're
innovating,
the
the
pace
is
moving
so
fast
that
you
know,
we
see
organizations
that
decide
to
go
all
in
on
serverless
or
lambda
and
it's
it's
hard
to
kind
of
unify.
D
So
I
think
interoperability
is
going
to
be
important
and
we
see
that
I
think
in
in
you
know
like
I'll,
give
you
an
example
here
with
lambda,
which
is
you
know,
we
we
support
the
packaging
runtime,
the
docker.
You
know
the
docker
packaging
with
inside
of
lambda
right,
so
the
that
the
fact
that
the
industry
has
coalesced
around
some
of
these
standards
allow
for
that
interoperability
to
happen.
D
I
know
a
lot
of
companies
that
where
kubernetes
is
the
right
choice,
but
I
think
the
more
we're
we're
able
to
allow
them
to
kind
of
move
their
workloads
around
and
their
compute
around
and
use
what's
best
for
them
is
what's
going
to
really
drive
cicd
forward
so
that
they
can
actually
continue
to
use.
You
know
the
cicd
tools
for
different
compute
platforms.
We
don't
even
know
what's
going
to
come
out
next
year,
the
year
after
that,
which
I'm
sure
there's
going
to
be
a
bunch
of
new
innovation
from
all
the
vendors.
D
B
You're
really
just
saying
interoperability
is
important,
but
what
we
end
up
doing
is
we
find
a
we
build
a
few
things
we
find
a
winner.
We
find
a
group
that
coalesces
around
it
becomes
a
de
facto
standard,
kubernetes
being
one
of
the
modern
de
facto
standards
and
then
over
time
it
gets
codified
into
what
we
used
to
start
with
in
standards.
B
B
It
needs
to
move
over
and
then
that
allows
the
proliferation
of
solutions
that
cover
many
more
of
the
vertical
stacks
of
things,
as
well
as
as
the
work
that
we
see
in
enterprises
moving
toward
the
continuous
delivery,
continuous
integration,
greater
reliability,
greater
mean
time
to
repair,
et
cetera.
A
B
I'm
happy
to
jump
in
and
say
that
every
customer
has
to
sit
and
look
at
what
their
risk
profile
is
for
change,
and
some
things
are
easier
to
change
than
others.
As
we
all
know,
in
tech
and
developing
a
plan
where
you
take
appropriate
risks
and
take
risks
which
you
can
support
is
really
the
best
way
to
look
at
this.
B
I
will
always
argue
for
open
source
solutions
or
solutions
that
have
an
open
source
parallel
or
a
standard,
or
a
de
facto
standard
that
may
exist
in
open
source
as
well,
but
we
have
to
balance
in
the
in
the
development
that
the
pace
at
which
we
can
do
these
things.
Maybe
it's
to
get
a
new
feature
out
quickly.
We
need
to
get
this
as
opposed
to
building
out
a
new
system.
It
ends
up
being
a
very
complex
yeah.
D
What's
the
what's
the
actual
cost
of
this
vendor
lock-in,
which
is
yes,
there's
a
cost
to
it,
but
that
cost
is
what
it
takes
to
actually
move
away
from
that
position
versus
its
benefits
if
the
benefits
far
outweigh
the
cost
of
moving
it,
meaning,
if
I
get
you
know,
a
year's
worth
of
developers,
time
and
benefit
versus
the
cost
to
move
from
this
solution
to
this
solution
is
only
going
to
cost
me
a
month,
then
it's
a
no-brainer
I
get
11
months
of
of
added
benefits
right
and
so,
and
the
reality
is
in
a
lot
of
the
cases,
because
actually
the
last
topic
that
we
were
just
talking
about
interoperability,
so
many
of
these
things
worked
so
well
together
that
the
cost
of
lock-in
was
not
what
it
was.
D
And
so
when
you
went
with
a
certain
vendor,
you
were,
you
were
locked
in
for
for
a
very
long
time,
but
now
there's
so
much
interoperability
and
then
going
back
to
just
kind
of
like
the
the
container
standards,
the
oci
standards
where
it's
it's,
the
packaging
can
easily
be
moved.
How
many,
how
many
services
are
out
there
like
that,
just
take
a
docker
file
and
then
create
a
container
out
of
it.
Everybody
has
that
right.
D
So
what
that
used
to
be,
you
know,
10
20,
10
years
ago,
used
to
be
very
difficult,
because
I
had
a
certain
packaging
standard.
You
had
a
different
one,
but
now
it's
all
the
same,
so
moving
that
is
is
becoming
fairly
trivial.
So
I
think
when
we
talk
about
vendor
lock,
and
I
think
we
have
to
really
dig
and
understand
what
is
the
cost,
and
what
am
I
getting
out
of
this
managed
service.
E
Yep,
so
I
think
I
just
add
on
to
it.
I
think
it
really
depends
how
much
of
these
platforms
you
use,
whether
you
dig
deep
into
extensions
and
plug-ins
and
use
these
features
that
people
have
built
for
you
or
with
whether
you
think
about
things
like
how
am
I
going
to
debug
this?
How
am
I
going
to
work
on
it?
E
A
Thank
you.
My
next
question
is
about
this
transformation.
Again
again,
some
industries,
they
fully
embraced
cloud
and
contains
delivery
and
devops,
but
some
industries
are
lagging
behind
and
those
industries,
some
of
their
workloads,
might
still
be
on
like
remote
age
sites,
and
they
may
still
need
to
be
containerized,
but
still
their
pipelines
will
need
to
operate
against
like
public
clouds,
local
data,
centers
or
remote
edge
sites
like
how
the
future
contains
the
rip
pipelines
would
look
like
to
satisfy
these
needs
operating
against
different
target
environments.
C
I
think
I
try
to
answer
it
once
is:
basically,
it's
kind
of
unified
have
a
proper
abstraction,
because
there's
a
lot
of
technical
companies
are
participating
in
kubernetes,
basically
trying
to
figure
out
oh
edge
system.
Even
it's
just
a
cell
tower.
They
can
run
a
single
node
kubernetes
cluster.
Then
they
can
package
the
software
the
same
way
they
package
it
in
the
data
center.
They
can
push
that
to
everywhere.
E
I
think
things
like
pool
based
delivery,
combined
with
like
events
so
rather
than
you
having
delivery
solutions,
pushing
everything
if
your
deployed
systems
will
pull
when
they're
notified
of
changes.
I
think
it
makes
those
sort
of
models
a
lot
easier,
so
things
like
git
ops
have
come
in
and
on
top
of
kubernetes
and
whatnot
or
and
then
these
applications
on
the
edge
can
get
notified
when
there's
new
ones,
and
then
they
can
update
themselves
and
if
they
go
offline
it
doesn't
matter.
B
Well,
this
goes
to
using
event
models
for
even
just
deciding
when
to
update
things.
You
know
you
have
you
put
something
out
on
a
on
a
queue
and
then
someone
subscribes
to
those
changes
and
then
or
something
so
just
subscribes
to
those
changes
and
pulls
their
choices.
It
makes
it
it
makes
it
a
lot
more
capable
to
be
individualized
in
a
way
that
I,
when
that
pull
happens,
it
requests
what
that
group
needs
or
what
that
location
needs
in
a
way
that
is
more
meaningful.
A
Yeah,
okay,
thank
you.
So
we
have
about
three
minutes
left
and
I
want
to
ask
a
final
question
to
all
of
you,
given
that
we
are
at
cdcon
and
we
have
our
community
with
us
here
in
person
and
virtually
what
would
be
your
one.
You
know
ask
from
our
community
that
could
have
changed
how
the
future
of
contingency
would
look
like
what
our
committee
would
be.
Focusing.
B
C
Oh
I'm
more
from
a
technical
perspective,
I'm
truly
believing
the
declarative
so
means
we
have
this
new
concept
called
iac,
aac
or
all
kind
of
like
as
code.
I
feel
the
keep
saying
that
kubernetes
is
a
good
destruction,
abstraction
it
doesn't
have
to
be
kubernetes,
but
in
end
the
declarative
manner
will
keep
a
lot
of
things
to
be
much
cleaner.
Basically,
that's
how
you
deploy
things
and
manage
things
in
the
end.
D
Isaac,
I
am
going
to
go
back
to
the
human
side,
which
is,
I
think,
the
best
thing
that
we
could
do
is
to
be
kind
to
each
other,
and
I
know
that
sounds
weird,
but
there's
so
much
change
happening
and
so
many
people
being
affected
by
this
technological
change
and
we
were
talking
about
people's
jobs
right
and
those
jobs
change,
and
it's
not
easy
for
a
lot
of
people
to
have
their
jobs
that
have
been
the
same
to
change.
They
think
it's
going
to
go
away.