►
Description
The promise of hybrid multi-cloud applications has been something the industry has been touting for as long as we have had cloud computing. The reality though is sorely disappointing, with only a few super technical teams coming even close. This talk poses the challenge of identifying why it is so complex and what we can do about it. Attendees will see
– a view of why Multi-cloud deployments are interesting
– background on the challenges we are seeing across companies in going multi-cloud
– strategies to put in place to make it easier to get started
Check out all recordings from the Destination: Automation 2021 event: https://bit.ly/3AwzL9X
A
Well,
good
morning,
from
the
other
side
of
the
globe,
to
those
of
you
in
california
and
in
the
us,
my
name
is
sean
o'mara,
I'm
from
marantis
field
cto,
and
today
I'm
going
to
be
talking
a
little
bit
more
about.
What's
holding
us
back
from
true
hybrid
multi-cloud
applications,
we
have
a
specific
interest
in
dealing
with
multi-cloud
and
we've
been
working
with
a
lot
of
customers
who
are
attempting
this
multi-cloud
journey.
A
So
let
me
kick
off
straight
away.
You
know!
If
we
look
at
the
community,
if
we
look
at
everything
that's
going
on,
there
are
millions
of
different
terms
being
used
and
different
bodies
who
are
deciding
their
own
terms,
but
for
simplicity,
and
for
the
purposes
of
today's
conversation,
I
just
try
to
simplify
things
for
the
discussion.
A
So
we
all
know
what
a
public
cloud
is
and
for
the
purposes
here,
we're
talking
about
a
single
public
cloud
provider
offering
the
range
of
services
and
an
example
of
this
would
be
an
aws
or
an
azure
or
google
anybody
who's
offering
infrastructure
as
a
service
or
platform
as
a
service
services
from
a
public
cloud
that
you
can
subscribe
to
now.
The
next
two
terms
which
we
see
bandied
about
a
lot
is
this
line
between
hybrid
cloud
or
multi-cloud.
A
I
mean
what
is
the
difference
ultimately
to
me,
they're,
basically,
the
same
thing,
the
difference
being
that
hybrid
cloud
generally
refers
to
a
combination
of
public
private
and
some
kind
of
on-premise
cloud
solution
where
you're
spreading
workloads
across
those
two
or
three
different
platform
types,
whereas
multi-cloud
is
just
two
or
more
cloud
providers
and
I
tend
to
go
well.
Does
that
make
a
difference,
whether
it's
hybrid
or
multi,
if
I
also
have
private
cloud
solutions
within
that,
so
for
today's
conversation,
hybrid
multi-cloud,
we'll
use
them
interchangeably.
A
So
also
for
the
purpose
of
the
conversation,
I
just
wanted
to
lay
out
a
couple
of
very,
very
high
level
application
patterns
that
we
need
to
be
aware
of
when
we're
talking
about
multi-cloud,
and
this
is
what
we're
seeing
in
customers
just
for
the
more
technical
out
there
don't
shoot
me.
I
realize
that
these
are
just
illustrative
diagrams
and
not
actual
applications.
You
would
probably
not
do
very
well
with
using
these
as
your
actual
applications.
A
Sorry
about
that
first
point
I
want
to
raise
here
is
we're
seeing
two
core
application
patterns,
one
where
people
are
deploying
to
the
multi-cloud,
deploying
distinct
components
of
a
application
stack
to
a
different
cloud
provider.
So
I
know
it'd
be
crazy
to
have
the
load
balancer
in
a
different
cloud
to
the
web
services,
but
use
that
as
an
example,
we
have
our
web
services
in
one
cloud
potentially
close
to
our
consumers
and
our
application
services
may
be
in
a
public
or
a
private
cloud
service.
A
So,
let's
align
things
that
way,
whereas
the
second
option,
which
is
the
stretch
option,
is
where
we
start
seeing
much
more
complex,
multi-cloud
solutions
where
we
start
having
direct
network
connectivity
between
these
multi-cloud
components
where,
as
an
example,
we
have
web
services
that
have
that
are
on
the
same
networks
within
the
same
kubernetes
infrastructure,
where
those
pipe
networks
are
stretched
across
the
multi-card,
and
this
is
a
much
more
complex
method
to
orchestrate.
A
We
talking
to
a
lot
of
customers
across
the
world-
and
I
talk
specifically
here
in
europe-
we're
seeing
a
lot
of
customers
who
are
standing
up
and
going.
We
cannot
use
a
single
cloud
provider.
We
can't
potentially
even
be
reliant
fully
on
an
american-only
cloud
provider,
because
there
are
also
some
privacy
challenges
and
privacy
issues.
We've
seen
significant
failures
in
public
cloud
providers
over
the
last
year.
A
A
A
core
challenge
we
see
is
privacy,
regulation
and
data
sovereignty
regulation,
especially
here
in
europe.
The
u.s
has
similar
rules,
but
where
do
I
store
my
data
that
it
is
contained
within
the
data
sovereignty
rules
that
it
applies?
The
specific
data
privacy
rules
for
that
region
and
as
an
example
with
the
end
of
privacy
shield
last
year,
we're
not
supposed
to
store
european
data
on
american-based
servers.
A
A
We,
the
world,
is
moving
rapidly
forward
towards
edge
and
accepting
more
capabilities
around
edge.
The
reason
for
that
is,
move
the
workload
closer
to
the
user,
reduce
the
latency,
the
cost
of
storage.
I
have
different
types
of
things
I
need
to
store.
Sometimes
storing
in
cloud
is
very,
very
expensive.
I
may
want
to
store
that
locally.
I
may
have
a
better
solution
to
do
that
or
I
might
get
a
better
deal
somewhere
else.
A
So
if
I
can
pick
my
location
where
I
both
run
my
workload
as
well
as
store
my
data,
it's
a
great
deal
of
value
to
me
both
in
cost
saving
but
also
insecurity
of
that
information
and
securing
that
from
a
risk
perspective
that
leads
to
the
discussion
of
price
and
efficiency.
Of
course,
you
know
being
able
to
leverage
multiple
different
public
cloud
providers,
I'm
not
locked
in.
A
I
can
also
start
to
pick
the
cloud
provider
that
has
the
best
service
for
my
needs.
If,
let's
use
the
example,
azure
has
a
better
ai
service
and
google
has
a
better
analytics
service,
I
can
start
to
pick
which
of
those
I
prefer,
but
still
run
my
own
applications
and
workloads
effectively
across
those
environments.
A
We
start
to
talk
about
being
locked
in
and
if
I
get
locked
into
a
single
cloud
provider,
I
no
longer
have
any
leverage
over
them
from
the
perspective
of
competition,
so
I'm
stuck
there
and
if
I
go
into
a
single
public
cloud
provider-
and
it's
very
very
hard
for
me
to
move
out
without
having
a
proper
multi-cloud
solution,
I'm
stuck
and
I
they
can
put
up
the
price-
there's
not
much.
I
can
do
about
that
because
the
cost
of
moving
out
is
probably
a
lot
more
than
just
staying
and
living
with
that.
A
A
so
of
use,
cases
and
examples,
just
very,
very
brief,
high
level
ideas
which
we've
actually
seen
people
starting
to
implement
in
the
usage
of
multi-cloud.
So
the
very
first
of
those
and
I've
just
mentioned
is
cost
optimization
by
moving
the
workload
to
the
most
appropriate
place.
For
me,
a
good
example
is
big
data
storage.
I
may
wish
to
do
my
big
data
storage
more
cheaply
on
existing
data
center
hardware
that
I
have
inside
my
company,
but
the
web
workloads
which
are
much
more
dynamic
and
they
can
grow
and
shrink.
A
I'm
going
to
run
out
in
public
cloud.
That
immediately
is
a
multi-cloud
use
case.
I'm
running
some
workload
on-prem
or
in
my
private
cloud,
some
using
web
services
to
run
them
close
to
where
my
users
are
so,
I
potentially
could
be
using
different
services
in
different
regions
of
the
world,
and
I'm
doing
that
primarily
to
reduce
my
overall
cost
and
be
able
to
manage
and
have
a
better
understanding
of
my
cost
and
to
use
the
burst
ability
in
things
like
public
clouds.
A
It
also
aligns
to
this
preventing
of
lock-in
idea
where
I
can
start
to
run
components
wherever
is
cheapest
to
run
them
at
that
time
and,
as
you
know,
there's
more
and
more
options
for
that
coming
up
in
the
public
cloud
world
regionality,
putting
your
workloads
where
they
need
to
be-
and
I
spoke
about
edge
and
age-
is
a
big
driver
of
this
we're
seeing
more
and
more
edge
use
cases
we're
seeing
more
and
more
capabilities
that
need
that
faster,
more
local
processing.
A
A
great
example
is
your
backdrop
in
your
zoom
session.
You
know
that's
having
to
be
processed
and
in
a
server,
that's
extremely
far
away
from
you
and
is
highly
latent.
It's
going
to
make
your
session
terrible,
which
means
spending
our
lives
for
last
year
and
zoom
sessions,
anything
that
make
them
better.
That's
what
we
want
and
then
the
other
part
of
regionality,
of
course,
is
the
governance
component
and
I've
spoken
about
data
sovereignty.
A
Disaster
mitigation
is
a
big
one,
and
we've
been
talking
to
a
lot
of
our
customers
who
have
gone
all
in
on
public
cloud,
have
either
suffered
from
failures
or
have
governance
reasons
within
their
business.
To
have
that
backup
some
way
and
that
backup
just
the
way
the
rules
are
structured,
has
to
be
in
a
different
provider.
A
A
We
need
to
put
the
right
tools
in
place
and
the
right
process
in
place
and
the
whole
multi-cloud
model
of
application
deployment
definitely
helps
us,
handle
that
and
then
the
last
use
case,
which
we're
seeing
quite
a
bit
of
is
the
scaling
or
bursting
use
case,
and
the
primary
one
here
is
scaling
and
bursting
when
we're
talking
about
regional.
Sorry,
excuse
me
regional,
seasonal
expansion
of
workloads,
retailers
during
black
friday
or
the
christmas
season.
You
know
when
they
need
more
web
front
ends.
A
They
need
more
capability
to
handle
the
load,
which
would
be
crazy
for
them
to
provision
for
the
entire
year
where
they
wouldn't
be
using
it,
and
that
gives
us
that
flexibility
and
that
cost-effective
resource
management
we're
also
seeing
a
lot
of
enterprises
who
have
invested
heavily
in
on-premise
solutions
that
work
well
for
them.
You
know
that
investment
is
maturing
they're
happy
with
it,
but
they
need
that
little
extra
capacity
and
it
doesn't
make
sense
to
build
their
local
capacity.
A
A
You
know
a
real
multi-cloud
solution
has
a
higher
initial
effort
for
a
number
of
reasons.
It
takes
a
lot
of
planning.
We've
got
to
take
into
account
that
we're
dealing
with
different
providers
we're
dealing
with
different
systems,
different
apis,
different
costing
models,
different
levels
of
complexity,
we're
dealing
with
different
skill
levels
in
the
organization,
so
we
need
to
think
with
more
detail
up
front.
A
Complexity
is
something
we
we
have
to
be
aware
of,
and
we
have
to
find
ways
to
reduce
that
complexity.
That
complexity
comes
in
needing
specific
tools
with
data
replication
across
dissimilar
systems.
It
comes
in
having
to
manage
all
of
these
different.
You
know
whether
it's
gcp
ec2
or
an
eks
and
an
aks.
You
know
what
is
it
that
I
need
to
manage
and
how
do
I
get
some
level
of
consistency
across
all
of
those?
A
The
next
one
is
obviously
networking
and
dealing
with
all
the
latency
challenges.
It's
cool
that
we
can
stretch
a
cluster
across
two
clouds.
But
what
happens
when
the
latency
between
the
two
clouds
starts
to
shoot
up,
because
an
internet
connection
is
overloaded
and
that
becomes
a
problem
because
you're
going
to
get
weird
performance
in
your
applications.
A
We
need
the
right
tools
and
the
right
systems
in
place
to
handle
and
manage
that
communication
effectively
localizing
the
data,
but
still
providing
that
correct,
cross-site
communication
and
then
also
I'm
suddenly
stretching
across
a
very
broad
range
of
systems.
How
do
I
abstract
all
the
networking
complexity?
Different
ip
address
ranges
whether
it's
ipv4
ipv6,
whether
it's
different
security
models
for
egress
and
ingress?
A
A
Now
they're,
giving
us
databases
and
they're,
giving
us
analytics
tools
and
extra
security
tools.
And,
of
course,
every
cloud
has
a
different
set
of
tools,
different
way
of
working
with
those
tools.
If
I
build
applications
that
are
reliable
on
something
that
amazon
offers
versus
something
that
is
my
on-prem
database,
it's
not
exactly
portable
and
it's
not
exactly
easy
to
move
it
around
if
I'm
certainly
tied
into
one
of
those
services.
A
So
you
have
to
examine
the
use
of
those
add-on
services
very,
very
carefully
and
think
about
the
rules
you're,
putting
in
place
to
guide
your
developers
to
make
the
right
decisions
notice.
They
guide
your
developers
not
dictate
to
your
developers,
because
they're
also
very,
is
very
rarely
successful.
A
So
how
do
we
make
this
all
work?
So?
The
first
thing
is:
consistency
is
a
critical
component.
We
need
a
consistent
platform
across
all
of
these
environments.
We
need
to
have
the
operational
automation
set
up
correctly
so
that
I
have
consistent
security,
consistent
policies,
consistent
identity
management
across
all
of
these
systems
and
the
big
one
is.
My
applications
must
be
deployed
via
ci
cd
solutions,
no
manual
deployments,
in
fact,
no
way
to
do
a
manual
deployment
is
probably
a
really
good
idea.
A
We
have
to
understand-
and
you
know
kubernetes
is
a
good
example
here-
it's
more
than
just
a
container
orchestration
system.
There
are
many
different
layers
that
we
have
to
take
into
account,
and
I
mentioned
them.
Identity
are
back
service,
mesh
capabilities.
On
top
of
that,
how
am
I
linking
all
these
systems
together?
A
Ease
of
use
should
probably
have
been
at
the
top
of
the
list.
All
of
this
is
cool,
but
it's
so
hard
and
so
complex
for
me
to
deploy
applications
into
this
environment
as
a
developer.
I'm
never
going
to
bother
because
I'm
going
to
look
for
the
shortest
path
to
getting
my
applications
up
and
running,
and
that's
probably
going
to
be
directly
to
a
single
cloud
provider.
So
we
need
to
make
this
easy.
It
needs
to
be
simple.
A
A
A
What
cloud
am
I
choosing,
but
as
well
as
what
clce
tool
do
I
choose
as
a
developer,
what
security
tools
are
most
appropriate
for
the
type
of
application
and
then
the
last
bullet
that
I'll
point
out
here
is
obviously
centralized.
Monitoring
goes
with
centralized
control.
If
I
don't
have
a
view
of
all
of
this,
I'm
going
to
get
lost.
A
So,
in
summary,
I
see
my
time
is
running
down
here,
but
in
summary,
what
we're
talking
about
for
multi-cloud?
We
need
consistency,
we
need
connectivity
and
the
building
blocks
of
all
of
this
run.
Firstly,
we
need
a
clear
multi-card
management
tool,
something
that
can
stretch
across.
All
of
these
clouds
provide
us
the
correct
layer
of
abstraction,
so
we
don't
have
to
write
applications
to
deal
with
all
the
different
apis.
Don't
have
to
deal
with
all
of
that
in
our
ci
cd.
A
A
Having
a
life
cycle
management,
that's
going
to
completely
on
its
own
handle
an
operating
system
upgrade
without
taking
into
account
the
workload
on
top
of
that
is
going
to
be
a
nightmare
you're,
never
going
to
end
up
with
upgrades.
We
have
to
be
able
to
effectively
update
components
with
some
awareness
of
what
applications
are
doing,
or
at
least
some
feedback
into
the
application
orchestrators.
So
we
can
do
things
like
coordinator
draining
a
usable,
ci
cd.
I've
covered
the
cicd
solution.
A
Today
you
know
some
will
handle
multi-cloud,
but
not
really
very
well.
So
if
we
don't
have
to
have
them
address
the
multi-card,
we
can
address
a
management
api,
all
the
better
once
again
simplicity,
freedom
of
choice,
I've
spoken
about.
We
want
that
freedom
of
choice,
it's
the
whole
driver
and,
I
think,
what's
important,
is
you
know
we
have
to
have
the
right
partners.
There
are
a
lot
of
very
smart
agents
out
there
who
we
can
leverage
to
do
things
in
the
right
way.
Smart
tools,
smart
partners,
so
that
we
can
layer
up
an
effective
solution.
A
So,
in
summary,
what
are
you
doing?
We
would
love
to
hear
from
you
how
your
multi-cloud
journey
is
going
and
have
you
planned
anything
from
a
multi-card
perspective?
Are
you
doing
this
today?
What
what?
What
has
been
your
experience?
A
What
is
that?
What
challenges
have
you
come
across?
Where
have
you
done
this?
What
have
you
learned,
we'd
love,
to
learn
with
you,
we'd
love,
to
share
our
learnings
with
you,
you
we're
building
tools
to
do
this.
We're
really
focused
on
this
as
part
of
the
future
we'd
love
to
hear
how
we
can
help
you
what
we
can
do
to
make
this
easier.
A
Please
feel
free
to
reach
out
to
me,
as
I
said,
that
from
morantus
a
lot
more
information
on
our
website.
A
lot
more
information
on
khan's
website
love
to
hear
you
moving
forward
here
and
I'm
within
my
time.
Thank
you
very
much.