►
From YouTube: From Code to Cloud With Klotho
Description
Taking a cloud application from local development to production is difficult. It requires packaging the application and its dependencies, integrating against libraries and SDKs for the cloud of your choice, provisioning infrastructure on your chosen cloud provider, and doing so in a repeatable way.
It takes experience and knowledge to choose the right infrastructure for the developer’s use case, and changing requirements can further complicate matters. It’s time-consuming, error prone, and requires application changes for experts in infrastructure and cloud operations to provision infrastructure in a repeatable manner. Engineers and small teams often turn to managed services to simplify the process. Utilizing these services come with quite a few downsides, including vendor lock-in, nonstandard technologies, and black boxes that cannot be tweaked.
What if you could take advantage of all the cloud's features while focusing on application code during development? What if a system could determine what cloud resources would be required based on your code? As your requirements change, what if you could change these cloud resources dynamically or switch cloud providers entirely without modifying your code? When adding a load balancer, storage, or any other dependency to your application, this system would automatically determines what pieces are required. Klotho provides a unique approach to solving these problems while avoiding the common issues associated with typically managed service solutions.
A
A
A
And
thank
you
very
much
for
joining
us
on
the
beaming
ordelius
podcast
and
today
we
are
talking
about
Cloud.
So
so
let
me
introduce
a
bit
about
what
clother
is
and
my
guess
as
well.
What
if
you
could
take
advantage
of
all
clouds
feature
while
focusing
on
application
code
during
development?
What
if
a
system
could
determine
what
cloud
resources
would
be
required
based
on
your
code
as
your
requirement
change?
A
What,
if
you
could
change
those
Cloud
resources,
dynamically
or
switch
Cloud
providers
entirely
without
modifying
your
code
when
adding
a
load,
balancer
storage
or
any
other
dependency
to
your
application?
This
system
would
automatically
determine
what
pieces
are
required
and
closer
provides
a
unique
approach
to
solving
these
problems,
while
avoiding
the
common
issues
associated
with
typical
managed
service
solution,
and
today,
I'm
joined
by
the
co-founder
of
the
Clinton
projects,
Allah
Shivan
and
Aaron
Torres.
Thank
you
very
much
for
joining
me.
So
now,
I
go
to
the
island.
B
Sounds
good,
thank
you
for
having
us,
so
my
name
is
I
grew
up
on
Commodore
64s,
with
cassette
tapes.
I
was
a
software
engineer
for
about
five
six
years
then
joined
Microsoft
as
one
of
the
product
managers
on
visual
studio
and
then
joined
Riot
games
where
I
met
Aaron
to
work
on
or
was
leading
the
cloud
services
organization
responsible
for
running
and
operating
all
of
League
of
Legends
and
all
the
newer
games
that
try.
C
Absolutely
yeah
as
well,
echoing
that
thank
you
for
for
having
us
on.
My
name
is
Aaron.
I've
I
have
been
mostly
focused
on
back-end
systems
and
infrastructure
teams
for
for
the
past
10
years,
my
my
background
started
off
in
sort
of
high
performance,
Computing
working
on
super
computers
as
a
system
administrator
and
researcher,
moving
into
basically
being
on
a
team
that
had
a
breakup
monolith
into
microservices
and
sort
of
gold.
C
The
whole
route
of
first
containerizing
everything
and
doing
all
that,
for
the
first
time
to
Riot,
where
I
met,
Allah
and
I
actually
was,
was
one
of
the
management
team
that
saw
usko
from
three
different
Technologies
over
time
right
and
migrating
multiple
teams
in
game
teams.
During
that
time,
as
well.
A
Yes,
thank
you
very
much
for
the
reduction
introduction
as
well
and
I,
see
like
in
this
day
we
all
Focus.
We
all
talk
about
the
shift
left
approach
where
you
can't
file
a
ticket
for
the
devops
team
to
give
me
the
infrastructure
and
then
run
my
application
now
and
when
I
look
at
the
cloud
through,
it's
actually
solve
this
problem
because
I
have
this.
My
application
I
want
to
run
them
on
the
cloud,
but
clothes
to
understand.
A
My
requirements
go
to
understand
my
language,
where
I
write,
I
write
the
where
the
code
is
written
from
and
it's
eventually
deployed
to
the
production
server
in
the
cloud.
So
it's
really
cool
and
it's
today's
your
show.
You
tell
us
my
audience
about
what
the
cool
features
of
the
cloth
is
and
if
you
want
to
share
screen
or
have
some
demos
prep
going
on
I
will
that's
will
be
a
fantastic
I
will
ask
question
as
we
go.
B
Sounds
good
what
we,
what
I'll
show
with
the
demo
is
something
very
powerful
that
demonstrates
how
developers
can
write
applications
without
necessarily
knowing
what
the
underlying
infrastructure
is
going
to
be,
but
it
gives
them
all
the
ways
to
write
applications
just
like
they
would
write
them
locally.
It's
actually
local
applications
and
then
they
can
get
them
deployed
with
clothing
or
have
them
be
ready
for
the
cloud
with
clodo.
B
But
if
you
have
a
devops
or
infra
team,
they
can
also
configure
and
control
what
the
infrastructure
will
look
like
without
having
to
ask
the
Developers
for
work.
So
it
creates
the
separation
for
the
two
groups
to
work
and
do
what
they're
best
at
and
so
I'll
show
that
to
in
the
in
the
demo.
So
maybe
we
can
jump
right
in
I'll
present.
My
screen.
B
That
yeah,
absolutely
here
we
go,
we
can
do
another
one
and
we'll
make
it
look
nicer,
Okay.
So
what
I'm
going
to
show?
You
is
an
application
that
we
start
with,
and
you
can
think
about
this
as
the
starter
repo,
and
this
is
where
everybody
hides
their
complexity.
Right,
just
get
your
tool
clone
this
repo.
Look,
how
simple
it
is!
B
Don't
look
into
that
special
folder
with
the
70
spec
files
that
you
have
to
learn
and
every
time
you
want
to
change
or
add
something
it
takes
you
hours
or
you
need
to
read
five
blog
posts
or
a
book,
and
so
I'm
going
to
show
you
that
there's
no
tricks
in
here
that
we're
not
hiding
any
complexity,
and
so
this
is
the
TS
config.
This
is
a
typescript
application.
We
support
typescript
JavaScript,
we're
working
on
python
now
and
then
golang,
Java
and
c-sharp
as
the
first
steps.
B
This
is
the
TS
config
for
the
typescript
configuration.
This
is
the
package.json
for
the
node
dependencies.
There's
only
application
Level
dependencies,
we
don't
introduce
any
sdks
and
then
there's
the
rest
is
just
source
code.
There's
nothing
else
other
than
Source
no
specs.
Your
code
is
the
spec,
so
now
we
want
to
make
this
application
a
cloud
application.
B
So
first,
let
me
show
you
what
this
does
I'm
going
to
run
the
application,
it's
just
a
local
one,
so
you
can
run
it
locally
debug
it
locally,
there's
nothing
special
about
it
and
let's
open
another
terminal
and
just
hit
the
local
endpoint
with
this
URL.
What's
with
this
URL,
or
this
is
the
result,
one
let's
actually
post
to
it
first,
and
so
the
endpoint
we
have
in
this
API
is
a
recognized
one.
We
give
it
a
URL
that
you
can
see
here
it's
an
image.
B
It
tries
to
detect
all
the
different
words
in
it,
basically
an
OCR
app
and
then
return
that
to
the
user
in
an
event-driven,
Manner,
and
everything
here
is
in
process,
meaning
there's
no
external
Services
we're
not
calling
out
to
AWS
it's
all
local,
and
so
we
hit
that
API.
It's
telling
us
it's
recognizing
the
text,
I'll
get
this
ID
and
then
ask
the
API
the
other
endpoint.
To
give
me
the
result,
if
it
found
one-
and
we
can
see
that
it's
found
something
that
it
didn't
quite
recognize
on.
B
B
Normally
this
would
do.
This
would
take
a
lot
of
effort.
You
need
different
tools.
You
need
to
know
how
to
specify
kind
of
the
specs
or
use
some
starter
application,
that
they
have
here's,
how
we
do
it
in
this
world.
First
of
all,
we
want
to
make
this
a
web
accessible
API.
So
this
is
a
an
Express
app.
It
doesn't
have
to
be
expressed.
We
work
on
interfaces
rather
than
specific
kind
of
libraries,
and
now
we
can
just
add
the
exposed
capability.
B
B
We're
done
so.
Let
me
show
you
what
that
means,
I'm
going
to
compile
the
typescript
and
then
run
clodo
and
just
say
clotho.
This
application,
and
just
give
it
a
name
in
this
case,
just
my
app
so
I'm,
going
to
run
clodo
it's
going
to
do
some
interesting
analysis
on
the
code
and
produce
this
folder
called
compiled.
B
What's
in
here
is
everything
you
need
to
run,
deploy
and
operate
this
application
without
us,
there's
no
sidecars,
no
binaries,
no
server,
no
private
Cloud,
nothing!
It
everything
you
need
is
here!
You
don't
actually
need
this
at
this
stage,
and
one
of
the
things
that's
nice
in
here
is
this
image,
which
shows
you
what
we
created
based
on
your
annotations
and
what
that
means
is
that
there's
infrastructure
as
code
and
there's
a
slightly
modified
version
of
your
application
that
is
ready
to
run
in
the
cloud
and,
like
I,
mentioned
this
image,
that's
it!
B
We
can
now
start
adding
and
making
this
application
even
more
kind
of
of
a
cloud
application,
so
we
talked
about
the
results
or
the
you
know.
The
we've
seen
the
results
of
the
API
when
we
called
it
and
we
want
to
store
those
results
because
right
now,
they're,
probably
in
memory,
let's
see
that
let's
go
to
the
OCR
router,
we
see
that
it
sends
a
request
to
this
module.
Let's
let
me
make
this
smaller
and
what's
once
it
gets
a
an
event
or
a
request
to
recognize
this
text.
B
It
recognizes
it
and
then
saves
it
into
a
data
store.
Now,
what's
in
this
data
store,
what
is
this
status
map?
Well,
let's
go
here.
We
can
see
that
it's
being
imported
from
this
module,
so
let's
go
in
here
and
it's
just
an
es6
map,
a
normal
map,
so
this
is
in
memory.
This
is
nice
when
we
work
locally,
but
once
you
restart
the
process,
the
data
goes
away,
which
is
not
something
we
want
definitely
not
in
a
production
environment,
so
we
want
to
persist
it.
B
B
B
Now
we're
getting
the
application
is
growing.
We
have
more
users,
we
have
more.
You
know
we
want
to
scale
the
application
differently,
because
the
OCR
part
needs
more
CPU.
The
API
part
is
more
throughput.
So
how
do
we
do
that?
This
is
kind
of
the
microservices
approach
or
needing
to
split
a
serverless
method
into
multiple
ones.
So,
let's
do
that.
B
The
way
to
do
it
in
today's
world
is
painful.
You
have
to
split
this
up
into
multiple
actual,
you
know,
services
or
functions
or
lambdas,
and
then
you
have
to
put
the
API
Gateway
definitions
to
route
to
each
different
service.
You
have
to
Define
how
they
package.
The
events
understand
the
Lambda
Handler.
Let
me
show
you
what
this
looks
like
in
our
world.
So
this
is
the
API
router.
The
only
thing
I'm
going
to
do
is
add
another
capability,
clotho
capability
called
exec
unit.
B
This
defines
the
boundaries
of
the
services,
and
so
we'll
call
this
API
routes.
The
names
again
are
arbitrary
they're
there
to
help
us
understand
what
this
is
and
then
let's
go
to
the
OCR
one
and
just
add
another
execution
unit.
That's
a
different
one.
Let's
call
it
OCR
routes,
that's
it.
Let's
recompile.
B
Sends
it
to
the
request,
emitter
and
what
this
is
is
just
an
OCR
module
that
we're
importing.
So
it's
just
a
normal
module,
and
you
see
that
there's
it's
an
emitter,
just
a
plain,
node.js
emitter.
We
can
see
it
here
from
the
events
object
and
then
it
has
a
Handler.
So
when
it
sees
the
event
or
the
the
request,
recognized
it'll
take
the
ID
that
it's
being
passed,
recognize
the
text
and
then
save
it
into
that
map
and
that's
it.
B
So,
let's
split
this
into
its
own
execution
unit.
Now
and
we'll
do
it
like
this,
we'll
just
add
another
execution
unit
and
let's
call
this
OCR
logic
and
then
go
into
this
emitter,
which
right
now
is
just
a
normal
emitter,
and
we
want
to
make
it
a
pub
sub
emitter,
meaning
it'll,
be
able
to
get
events
and
push
them
into
a
queuing
like
system
and
we'll
call
this
the
request
or
recognize
Pub
sub.
B
And
once
we
do,
we
get
this
topology,
so
we
just
built
a
cloud
native
application,
non-trivial
topology.
This
would
have
taken
at
least
days
to
build
in
any
of
the
tools
that
we
have
today.
We
just
built
it
in
less
than
10
minutes
and
now
I'm
going
to
show
you
something
that
no
other
system
has
right
now.
All
of
these
are
lambdas
in
our
world.
In
the
clothing
world,
this
is
not
microservices
or
serverless.
We
call
this
adaptive
architectures
and
the
reason
is
everything
in
this
diagram
is
actually
interchangeable.
B
It
doesn't
have
to
be
any
of
those
pieces,
and
let
me
show
you
an
example
of
that
right
now,
it's
lambdas.
So
when
you
run
clothing,
what
you
get
on
the
other
side
isn't
just
the
application
and
the
image
and
the
infras
code,
but
you
also
get
a
configuration
file.
We
did
not
write
this
or
you
didn't
write.
B
This
clodo
did-
and
it's
based
on
the
defaults
that
it
needs
to
run
this
application,
but
you
can
override
anything
and,
even
more
importantly,
the
configuration
for
the
infra
is
done
using
the
same
infra
as
code
framework
that
you're
already
using
so
we're
not
inventing
a
spec.
So
let's
take
this
configuration
out
into
the
main
folder
and
we
can
delete
everything
that
we
don't
care
about,
that
we
don't
want
to
change.
B
The
defaults
are
fine,
so
I'm
going
to
delete
these
and
I'm
going
to
delete
everything
in
here
I'm
going
to
delete
the
parameters
because
we're
good
with
the
defaults
and
what
I'm
leaving
is
the
different
execution
units
which
you're
seeing
here
that
are
lambdas,
and
now
we
decide
that
lambdas
aren't
good
for
the
use
case.
Now,
because
it's
been
six
months
and
now
we
realize
that
it's
not
working
for
the
use
case.
B
B
So
they
don't
need
to
go
and
ask
developers
to
change
anything
in
their
application
and
that's
why
why
we've
designed
this
this
way
it
is
purposefully
built,
so
each
team
and
each
function
can
do
what
it's
best
and
good
at
and
if
you
are
in
a
place
where
you
are
a
like
true
devops,
like
I,
do
all
the
things
developer,
then
you
don't
even
have
to
look
at
this
image.
It
just
works.
B
We
haven't
built
out
those
pieces,
yet
we
have
parts
of
gcp
ready,
but
we'll
show
that
on
a
different
day,
and
so
we
can
do
follow
me
up,
and
in
this
case
it
would
actually
go
ahead
and
build
out
the
plan
to
deploy
so
I'll.
Stop
it
here
and
maybe
unshare
my
screen.
So
we
can
jump
back
and
have
a
chat.
Does
this
make
sense.
A
This
is
super
cool
I.
Think
everything
that
you
showed
is
everything
you've
been
working
on
the
cloud
and
the
application
development
they
can
see
how
easy
and
how
exciting
this
was,
because
I
want
to
write
the
application
on
the
cloud,
but
I
need
to
learn
how
to
Fork
how
to
set
up
forget
how
to
set
up
the
load
balancer
and
when
we
set
up
the
load.
Balancer
I
want
to
set
up
the
mutual
TLS
between
the
application
routing
to
the
traffic,
so
this
is
insanely,
powerful,
I,
I,
hope
everybody
watching
or
listening
to
this
podcast.
A
A
This
is
I
really
like
about
it,
because
in
the
previous
conventions
we
used
substrings,
we
use
hard-coded
string,
we
use
hard-coded
configuration
hard-coded
secrets
and
that
need
to
be
changed
and
let's
say
if
the
beam
is
some
of
the
employees
leaving
the
company,
and
this
is
their
responsibility
and
New
World
hires
up.
Who
has
this
responsibility
now
to
stick
these
things
because
it
just
might
be
the
ramification
of
the
certifications
or
the
password
or
the
configuration
settings
the
database
strings.
This
is
insanely
popular.
A
You
just
have
I
just
see
only
you
tell
the
application
that
I
need
AWS
for
that
you
have
annotations
and
then
application
can
figure.
It
out
do
I
use
Lambda
4K
and
you
can
switch
between
them.
This
is
extremely
powerful
for
me
at
least
I
really
like
the
demo
in
20
minutes.
I.
Think
you
ground
it
up.
Everything
everything's
summarizes
wonderfully
well
during
the
demo
like
what
it
all
talk
about
it.
So
wonderful,
wonderful,
absolutely
wonderful
and.
C
We
haven't
shown
that
so
yeah
go
ahead.
Aaron
yeah
I
want
to
say
a
couple
of
other
things
that
we
we've
kind
of
glossed
over,
but
are
there
because
we
have
the
graph
and
we
know
the
dependencies
of
everything
we
have
some
certain
advantages
that
you
might
not
normally
have.
For
example,
we
can
be
very
strict
on
like
IM
policies,
where
we
know
what
resources
talk
to
one
another
and
therefore
we
can
set
up
policies
to
only
allow
them
to
talk
to
one
another.
C
So
you
actually
have
a
the
lowest
trust
possible
in
the
network
as
well,
which
means
if
you're
security-minded,
you
don't
have
to
think
about
that
at
all.
Right
and
that
alone,
just
making
sure
the
ion
policies
are
managed,
and
you
don't
have
to
recreate
them
or
think
about
them
has
been
a
huge
Time
Saver.
A
B
So
we
built
out
so
we
have
two
versions
of
clodo
the
earlier
one
and
the
newer
one,
the
earlier
one
supported
both
AWS
and
gcp,
and
we're
now
working
towards
porting,
the
gcp
portions,
into
the
new
compiler.
So
we
have
both
AWS
and
gcp,
with
gcp
being
Early
Access
or
in
development,
with
the
newer
version
and
we're
doing
Azure
after
that.
B
So
right
now
python
is
in
development.
We
already
have
versions
out
with
the
early
users
and
adapters
that
are
using
it.
We're
expanding
the
capabilities
that
I
showed
in
typescript
and
JavaScript
to
be
equivalent
or
parity
in
in
Python,
so
types
for
JavaScript,
premature
in
production,
with
multiple
users
and
customers
working
on
non-trivial
loads-
and
you
know,
production
systems.
Python
is
coming
probably
in
the
next
few
weeks
or
months,
and
then
we're
going
to
do
golang
next,
because
golang
is
used
heavily
in
the
distributed
systems
or
microservices
world
and
then
likely
several
months
out.
B
We're
going
to
do
c-sharp
Java,
because
there's
more
there's
more
work
to
be
done
there
and
we
don't
want
to
spread
ourselves
thin
too
much
with
too
many
languages,
because
we
do
work
on
a
like
language.
Specific
Cloud
understands
each
and
every
language
differently,
and
that
adds
a
level
of
analysis
that
we
have
to
do
that.
Isn't
quick
to
build
out.
A
Yes,
absolutely
absolutely
and
another
question
that
it
came
with
your
mind
like
do
you
think
how
clotho
or
in
future,
not
today,
if
how
cloth
is
replacing
the
devops
or
the
off
State,
because
if
people
are
watching
this
podcast,
we
know
we
need
to
have
the
infrastructure
already
built
up,
but
now
you're
just
compiling
your
code
into
the
terminal
and
every
diagram.
It
shows
how
the
architectural
view
is.
Look
like
and
plus
is
plus
is
adding
resources
into
the
cloud.
A
B
We're
not
replacing
anyone,
so
the
answer
is
no
we're
not
replacing
anyone.
We
see
it
as
a
no
Ops
to
full
up
Spectrum,
if
you're
small
enough,
if
you're
a
smaller,
startup
or
individual
or
a
smaller
company
that
doesn't
need
in
our
world
that
much
operational
management,
because
your
application
is
smaller
you're
using
more
managed
services
using
clodo
like
lambdas
You,
probably
you're
you're.
Doing
the
devops
work
yourself
when
the
Apple
application
grows
to
a
really
really
large
size,
because
we
ran
League
of
Legends.
We
understand
you
know.
B
Devops
are
your
best
friends,
specifically
the
platform
and
infra
team
are
your
best
friends
and
you
want
to
not
necessarily
have
as
many
people
working
only
on
that.
So
what
we
expect
to
see
is
in
larger
companies
where
the
applications
are
more
complex,
is
you'll
need
less
people
to
work
on
the
same
kind
of
low
level
things
or
the
same
people
working
on
harder,
more
interesting
things.
That
clothing
now
enables
you,
whether
it's
building
more
interesting
systems
or
upgrading
existing
systems.
B
That's
how
we
see
the
world,
so
you
may
need
less
people
to
do
the
same
things
today,
but
you
free
them
up
to
do
more
interesting
things
or
if
you're,
a
smaller
company
or
kind
of
you,
have
a
few
use
cases.
Then
you
wouldn't
need
as
many
folks
or
be
able
to
not
need
like
you
can
do
the
the
all
the
info
work
yourself
or
you
know,
using
clodo.
You
will
get
to
do
the
devopsy
things
yourself,
which
means
just
little
work
and
that's
it.
A
Yes,
absolutely
absolutely
and
I
like
to
see
it
in
a
way
like
I
need
to
focus
on
my
application
code,
but
that
application
code,
the
logic
the
business
logic,
need
to
run
on
the
infrastructure.
So
I
do
not
spend
two
three
months
in
in
building
up
the
infrastructure,
but
I
will
focusing
on
what
my
business
was
over.
A
The
business
logic
was
and
go
to
make
it
very
easy
for
me
to
build
up
the
infrastructure,
so
I
can
focus
entirely
on
the
core
and
the
business
logic
and
forget
about
where
this
application
is
running
behind
the
scene.
So
I
really
like
about
this
idea
as
well
and
also
like
you
know
like
this-
is
a
question
which
came
up
to
my
mind
when
I
first
demos
on
my
cluster,
when
I
first
tried
out
Cloud.
A
So
now
you
see
the
kubernetes
is
very,
very
vast
and
everybody
is
using
it
and
we
know
eks,
gke
or
AKs.
These
are
very
complex
setup
to
build
up,
and
these
Cloud
providers
give
the
managed
services
do
in
the
future.
We
see
by
using
the
cloud
through
CLI.
We
can
deploy
our
apps
on
top
of
ETS
GK.
Your
AKs
is
this:
are
you
looking
at
this
roadmap
as
well,
or
do
you
have
plan
for
this.
C
Yeah
so
a
couple
couple
things
there:
we're
planning
on
being
a
kubecon
later
this
year,
so
we're
prepping
a
few
things
to
talk
about
while
we're
there
related
to
this,
but
we
have
early
proof
of
concepts
of
spinning
up
eks
and
targeting
fargate
very
similar
to
to
the
parking
Tempo
with
ECS
yeah.
It's
definitely
on
our
radar.
It's
a
big
space,
almost
big
enough
that
it's
like
another
Cloud
by
itself.
But
yes,
it's
definitely
something
we're
investigating
and
thinking
happily
about.
B
C
B
A
Yeah,
that's
absolutely
that's
really
exciting
news
for
me,
because
I
want
to
have
this
capability
built
in
today,
and
there
are
some
special
comments
or
capabilities
that
you
need
to
tell
clothes
on
how
to
create
ictps
servers
when
building
you
in
front
of
the
cloud.
Can
you
liberate
to
my
audience
on
this
topic,
because
when
you
set
up
the
attribute
or
annotation,
on
top
of
it,
you're
telling
you're
telling
actually
collotho
I
need
a
https
servers,
but
behind
the
scene?
What
is
happening?
A
C
Yeah,
so
the
the
main
reason
that
we
won't
bloomy
first
is
one
we
had
a
lot
of
familiarity
with
typescript
in
the
original
compiler
was
written
in
typescript,
which
made
it
easy
to
adopt,
but
also
we
really
appreciate
the
we.
We
really
appreciate
how
fast
we
can
move
with
it
being
a
coding
environment
right
like
thinking
of
it
as
like,
especially
in
terms
of
like
importing
code
completion,
just
thinking
of
it
in
terms
of
normal
language
constructs.
It's
been
really
nice
for
us
to
go
fast
early
on,
I
would
say
like
long
term.
C
C
There's
a
good
question
and
also
to
that
point
like
AWS
releases,
many
many
new
Services
every
year
as
well
right.
So
so,
right
now
in
terms
of
roadmap.
For
this,
it's
largely
going
to
be
based
on
demand
users,
but
what
we?
What
we
support
today
for
compute
is
Lambda
and
fargate
which
we're
both
in
the
demo.
We
have
a
few
other
things
related
to
storage.
We
have
support
for
dynamodb.
C
We
have
support
for
RDS,
with
postgres,
using
using
orms
and
also
initial
Reddit
support
as
well,
so
that
the
plan
is
that
we'll
continue
to
expand
that,
but
keep
in
mind
that,
as
we
add
more
languages,
it
becomes
a
big
Matrix
of
like
language
capability
cloud,
and
so
so
we
have
to
be
somewhat
strategic
on
like
where
we
prioritize
those
things
on
a
roadmap.
Although
the
intent
is
to
support
as
many
as
possible.
B
B
Most
applications
wind
up
using
maybe
20
or
30,
of
what
exists
in
AWS.
If
you
think
about
ec2s,
if
you
think
about
lambdas
fargate,
kubernetes
app
Runner,
those
probably
cover
80
of
the
use
cases
same
thing
with
storage,
you
can
think
about
five
or
six
ones
that
wind
up
so
carrying
most
of
the
applications.
B
If
there's
one
that
we
don't
support,
you're,
not
blocked,
because
the
worst
case
scenario
in
our
world
is
that
you
continue
doing
exactly
what
you're
doing
today,
you
can
have
your
own
infras
code
file
that
you
import
all
the
clothing
things
into,
and
then
you
can
spin
up
your
own,
let's
say
memdb
or
your
own
mongodb.
If
we
don't
support
it
yet,
and
then
you
just
add
the
connection
strings,
just
like
you
would
do
today,
there's
you're
not
waiting
on
us
to
add
support.
You
can
import
the
AWS
SDK
or
the
Azure
SDK.
B
A
And
another
question:
okay,
that's
I!
Think
people
love
to
see
like
we
have
infrastructure
of
their
job.
We
deploy
the
code
on
top
of
it,
but
when
it's
running
in
the
production
environment,
I
need
to
have
observability
in
that
app
I
want
to
see
how
many
requests
going
on
are
successful.
How
many
discounts
requests
are
failing
what
the
response
rate
was,
how
much
time
you're
taking
when
requesting
from
goes
for
the
service
a
to
service
B,
all
these
services
are
communicating
in
a
healthy
setup.
A
C
Yeah,
the
minimum
you
get
is
is
what
is
currently
there
like
the
AWS
tools
that
are
there
so
like
we,
we
frequently
will
jump
into
the
AWS
accounts,
console
or
use
the
AWS
CLI
to
help
debug
with
that
said,
part
of
what
we
think
makes
it
makes
our
story
good
for
observability.
Is
that
the
stuff
that's
going
into
the
cloud?
You
have
full
visibility
on
there's
no
black
box
right,
the
stuff
that's
going
up
is
standard
infrastructures,
code
and
regular
code.
So
there's
no
tools
you're
restricted
from
using.
C
As
an
aside,
though,
just
one
thing
that
that's
in
the
back
of
our
minds
also
is
so
you
know:
we've
talked
about
how
we
have
these
interfaces,
the
the
kind
of
levels
of
abstractions
that
can
vary.
It
can
go
everything
from
like
hey.
This
is
a
map,
that's
very
portable
to
all
clouds
too
hey.
This
is
a
dynamodb
library,
that's
clearly
only
AWS.
Eventually,
we
plan
to
support
third-party
plugins
as
well
for
specific
use
cases.
So
someone
wants
to
use
the
specific
library
with
specific
technology
that
should
be
supported.
A
Yes,
absolutely
and
one
last
question
on
the
development
part
of
it,
and
then
we
will
talk
about
how
to
learn
growth
or
do
we
have
repositories
or
how
we
actually
gets
our
hands
dirty
with
the
cloud
Zone
and
but
last
goes
to
the
development
side.
Is
that
like
I
need
to
tweak
some
environment
setting
because,
based
on
the
my
company
and
organization
policies
like
political
cloth,
does
provide
those
kind
of
requirements?
A
I
want
to
tweaks
and
set
up
like
I
want
I've
in
in
the
term,
I
can
say
cloth
or
like
if
you
construct
a
cloud
or
the
forget,
make
sure
like
this
is
already
built
up
on
the
far
gate
or
some
capability
that
is
changing
from
company
to
company
organization
to
organization?
Do
we
incorporate
this
requirement
in
the
classroom.
B
So,
there's
a
few
layers
to
this
one
because
of
the
way
we
construct
the.
What
what
we
out,
what
closer
outputs
is
standard
code?
Transpiled,
it's
not
actually
byte
code
or
you
know
special
closed.
You
know
close
the
result,
so
you
can
run
all
your
existing
tools
or
other
tools
that
exist
in
the
ecosystem
on
the
output,
whether
it's
Securities
scans,
whether
it's
you
know
Finding.
If
there's
keys
that
shouldn't
be
there,
you
can
use
all
the
tools
that
exist
exactly
the
way
you
do
them
today,
including
in
a
CI
CD
environment.
B
The
way
you
can
think
about
requirements
is
that
we
have
a
special
system
for
coming
for
how
you
describe
slos
things
like
I
have
a
video
that
is
10
minutes
long
I
want
it
to
render
in
one
minute.
So
every
10
minutes
of
video
should
take
one
minute
to
render
and
we
allow
you
to
specify
those
requirements
in
code.
B
Just
the
way
I
described
and
then
have
close
those
Optimizer
figure
out
how
to
change
Technologies
behind
the
scenes,
whether
it's
you
know
automatic
or
whether
somebody
has
to
press
a
button
or
the
info
or
platform
or
devops
teams,
have
control
over
that,
and
so
we'll
have
we'll
share
more
about
that
in
the
future.
But
for
now
we're
keeping
this
more
closed
hour
chest
but
expect
some
amazing,
like
technology
that
doesn't
exist
anywhere
else.
B
A
Absolutely
I
already
see
in
the
demos
today
that
some
of
the
things
that
glauthor
does
is
not
exist
anywhere
else,
so
we
can
expect
more
good
things
as
well.
So
let
me
talk
about
all
good
things,
around
engineering
concept
and
all
things.
Let's
talk
about
for
the
learning
prospectus
like
if
people
like
developer,
want
to
use
cloudflow
what
are
the
resources
available
for
them
in
terms
of
the
documentation
repositories?
Can
you
share
some
details
for
the
developer?
We
want
you
to
learn
and
Hands-On
with
the
collateral.
C
Definitely
so
we
we
just
launched
the
the
closed
beta
clotho
last
week,
so
the
website
is
up
to
date.
Now
we
have
a
bunch
of
documentation,
examples,
ways
of
doing
things,
including
things
like
hooking
up,
existing
AWS
resources
or
using
resources.
We
would
support
as
well
as
a
bunch
of
our
capabilities.
C
B
B
We
have
not
only
like
we
have
a
samples
repo
that
you
can
find
on
the
docs
page,
it's
on
GitHub
and
we're
going
to
be
open
sourcing
everything
that
cloth
like
basically
open
sourcing
all
of
clodo
within
the
next
few
weeks
on
GitHub.
We
also
have
specific
tutorials
for
things
like
nest.js
next
JS,
you
know
redis
Type,
R,
M,
sqlize,
how
to
you
know,
switch
lambdas
to
fargate
or
the
other
way
around
the
entire
like
getting
started,
but
also
everything
you
can
do.
A
B
B
And
yeah
just
send
us
a
message
saying
you
know
we
heard
the
podcast
we'd
love
to
meet
up
in
kubecon
or
just
tweet
at
us,
the
Emma's,
whatever
you
know,
the
DMS
are
open,
so
feel
free
to
reach
out
and
we
can
absolutely
meet
there.
There's
also
the
Discord
Channel,
which
is
linked
on
the
website
and
linked
now
here
join
us.
The
entire
team
is
there
and
we
we
would
love
to
talk
to
you.
We
always
do
yes,.
A
Absolutely
absolutely
wonderful
discussion
with
you
guys
and
before
lunch
you
got
the
final
question
like
if
we
can
see
the
diagram
in
the
closer
provide
in
the
future.
Can
we
just
click
on
the
diagram
and
it
can
give
me
to
the
resource
if
it's
created
for
the
development
part
like
I,
just
want
to
say
like
if,
if
in
the
diagram,
I
can
click
on
the
diagram?
Let's
say
in
the
forget
and
it's
hitting
me
under
the
console
of
the
data
layer,
you
can't
even
see
that,
because
that's
really
cool
for
me.
A
Great,
that's
a
wonderful
news
for
me
because
I
like
to
see
it
in
the
in
the
future
as
well.
So
it's
a
wonderful
podcast
that
you
and
we
will
follow
up
again
on
the
other
YouTube
channel,
where
you
can
do
another
another
stream
of
things
as
if
things
are
going
very
fast
in
the
clothes
and
I
think.
The
best
thing
you
can
do
is
to
join
the
Discord
Channel
and
go
to
the
website
and
start
exploring
the
project
and
the
repositories.
A
What
inside
there,
what
we
have
that
and
what
we
do,
and
we
have
a
wonderful
Victor
farsik
having
a
wonderful
30-minute
guide
on
what
the
cloth
is
and
some
of
the
things
he
mentions
on
this
in
his
YouTube
channel
is
already
been
supported.
As
of
today,
it
requested
some
more
languages.
Sport
and
now
we
have
small
languages
board
as
well,
so
go
join
this
Channel
and
have
a
wonderful
learning
experience
for
you
and
I'm
also
running
it
for
telliest
Discord
Channel.
You
can
join
up.
A
We
have
an
open
source,
microservices
catalog,
where
we
can
deploy
different
microservices
in
and
make
track
of
where
these
micro
services
are
residing.
What
technology
we
are
using,
so
it
keep
all
the
tracks
of
the
micro
services
and
you
can
start
contributing
to
the
GitHub
repository,
slash,
utilia,
slash
utilities.
So
all
the
go
explore
these
channels
as
well,
and
thank
you
very
much.
Allah
Shivan
and
Aaron
Torres
for
a
wonderful
discussion
on
clothing,
I
learned
about
this
tool
and
really
like
it
about
this
tool
as
well.
It's
helped
me
this.
A
Me
and
I
will
start
exploring
it
and
will
get
to
you
in
the
feedback
how
I
feel
about
it,
and
if
people
tell
me
about
your
experience,
I
will
let
you
know
as
well.
So
it's
been
a
wonderful
day
for
me
and
thank
you
very
much
for
joining
the
podcast
podcast
stay
safe,
stay,
healthy,
bye,
bye.
Everyone
thank
you
for
having
us.