►
Description
Unik: Unikernel Runtime for Kubernetes - Idit Levine, EMC
"UniK is an open-source tool written in Go for compiling applications into unikernels and deploying those unikernels across a variety of cloud providers, embedded devices (IoT), as well as a developer laptop or workstation. UniK utilizes a simple docker-like command-line interface, making developing on unikernels as easy as developing on containers. UniK’s convenient REST API makes integrating UniK with orchestration tools a breeze. To demonstrate the value of cluster management of unikernels, we implemented a UniK runtime for Kubernetes, making Kubernetes the first cluster manager to support unikernels. This integration allows UniK to take advantage of core Kubernetes features like horizontal scaling, automated rollouts and rollbacks, storage orchestration, self-healing, service discovery, load balancing and batch execution.
On top of it all, UniK offers a highly pluggable and scalable architecture, allowing developers to add support for new processor architectures, programming languages, unikernel compilers, and cloud providers with ease."
About Idit Levine
Idit Levine is the CTO for cloud management division at EMC and a member of its global CTO office. Her passion and expertise are focused on Management and Orchestration (M&O) over the entire stack and on microservice, cloud native apps and Platform as a Service. Idit’s fascination with the cloud sprouted when she joined DynamicOps (vCAC, now part of VMware) as one of its first employees. She subsequently took part in developing the new-generation public cloud of Verizon Terremark, and served as an acting-CTO at Intigua, a startup company that focuses on container and management technology.
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
Hi
everybody
thank
you
for
coming.
For
my
talk.
My
name
is
Edith
Levine
and
I'm
from
BMC
Delhi,
MC,
emc,
Dell,
Dell,
EMC
and
and
I'm
going
to
talk
to
you
a
little
bit
about
unique
kernel.
What
is
it
why
we
should
use
it,
and
then
we
kind
of
like
nicely
tied
it
to
cubed
I,
didn't
explain
what
what's
the
benefit
of
playing
with
the
two
together
so
but
stuck.
We
would
understand
what
is
Unicode
so
usually
when
you're
running
your
application.
A
Today,
specifically
if
you're
running
in
the
cloud
for
SAS
on
AWS,
what
happening
is
that
you
have
this
big
stack
that
you're
running
your
application
on
top
of
it?
The
first
thing
that
you
have
is
the
hardware
AWS
someone
behind
the
scene.
There
is
hardware,
they
have
an
IP
voice,
then
in
that
use
case,
and
then,
on
top
of
it,
you
put
in
your
VM
instance,
which
is
basically
an
operating
system,
and
in
this
operating
system
you
have
their
virtual
drivers.
You
have
the
OS
kernel.
You
have
the
your
de
OS
user
processes
right.
A
That's
ever
between
your
processes,
then
you
are
the
doctor
runtime
kind
of
doing
the
same,
but
it's
a
great
packaging
though,
and
then
you're
putting
inside
the
docker
container.
The
shell
library
is
the
language
runtime,
the
application
and
the
application
config.
All
of
this.
This
is
all
huge
stuff
that
you're
running
every
time
that
you're
trying
to
run
your
micro
services
application
and
when
I
looked
at
it,
I
wonder
what
actually
we
trying
to
do
why
we
need
all
this
big
stack
and
surprise.
Surprise.
Our
home
is
kind
of
like
very
simple.
A
We
try
to
run
a
single
application
with
a
single
user
on
a
single
server
that
all
we
try
to
do,
and
there
is
quite
a
lot
of
ice
out
of
redundant
same
distaff,
for
instance,
isolation.
It's
a
great
example.
You
have
an
isolation
on
the
hardware
which
is
a
physical
isolation.
Then
you
have
it
on
the
hypervisor
v
arm
isolation.
Then
you
have
it
between
the
processors.
Then
you
have
it
between
in
the
docker
and
then
you
have
it
on
the
application
on
the
user
right.
A
So,
quite
a
lot
of
redundancy
for
what
we're
trying
to
do
and
when
I
looked
at
the
kernel
and
try
to
understand
why
it's
so
complicated
understand
that
one
of
the
reason
it
so
complicated
is
because
it's
job
is
to
protect
right,
its
protect
application
for
application,
user
form,
user
and
application
form
user,
and
that
was
I'll
make
a
lot
of
sense.
The
56
right
there
was
like
a
business
use
case,
I'm
buying
this
huge
mainframe
right,
so
expensive.
I
cannot
afford
myself
on
my
application
on
it
by
myself.
A
I
need
to
share
it
and
therefore
we
need
somehow
to
make
sure
that
we
will
have
to
protect
the
guys
that
sharing
it
with
me.
So
my
application
will
not
go
and
do
it
is
memory,
for
instance,
right,
and
that
was
make
a
lot
of
sense
back
then.
But
the
question
is:
is
that
make
sense
today
and
it's
probably
not
because
that
causes
a
lot
of
problems?
For
instance,
there
is
a
needless
from
a
permission
check
right.
You
always
need
to
see.
You
is
always
the
guys
running.
A
You
know
what
is
free
is
permission
and
so
on
and
that's
load
on
the
actual
application.
If
you're
looking
at
the
kernel
itself,
when
I'm
running
right
now
on
my
laptop,
you
will
discover
that
actually
I'm
running
and
a
lot
of
drivers
that
doesn't
make
any
sense,
for
instance,
I'm
right
and
I'm
right
now
running
the
floppy
driver,
no
one
using
floppy
anymore.
They
written
the
right
are
still
running
on
my
machine.
A
If
you're
looking
at
the
cloud
when
I'm
running
on
AWS
you're
running
us,
you
a
UPS,
you
ask
me
sorry,
even
though
you
cannot
directly
go
to
the
machine,
you're
still
running
this
driver
right,
so
a
lot
of
fun,
necessary
component
running
on
your
machine
and
when
you're
trying
to
operate
your
machine
and
you're
doing
something
like
a
Yama
update
or
get
up
you're
getting
a
lot
of
junk
to
your
machine.
Again,
you
don't
know
what
you
really
need.
A
What
you
donated
and
I
will
tell
you
that
you
don't
need
that
all
of
this
still
going
to
run
on
your
storage
right
in
terms
of
security.
If
you
have
a
lot
of
stuff
that
you
don't
need
on
your
machine,
this
is
a
very,
very
vulnerable,
be
a
vulnerability
point
right.
So
the
larger
the
doubt
the
surface
of
attacked
is
huge
right,
because
I
can
think
about
where
I
will
do.
A
Maybe
tell
it
right,
like
the
last
day,
like
the
last
big
DDoS
and
so
on,
because
there's
a
lot
of
variability
on
the
stack
and
micro-services
of
attack.
They
were
sharing
a
lot
of
stuff
we're
sharing
kernel
and
memory
and
file
system
and
hardware,
and
the
only
thing
that
we're
really
protecting
us
here
is
see
group,
and
this
is
like
putting
a
lock
on
a
very
big
house
with
a
lot
of
windows
and
doors.
It's
not
really.
You
know
people
are
still
going
to
penetrate
that
right.
A
If
we're
looking
at
the
Linux
kernel
language,
there
is
quite
a
lot
and
that's
the
language
that
one
need
to
know
in
order
to
maintain
that
so
it's
going
on
and
all
right
and
I
will
argue
that
there's
not
a
lot
of
people
in
the
Ecosystem.
Don't
know
all
of
this
and
understand
and
not
to
reason
about
with
all
this
language
in
terms
of
in
in
size
lines
of
code,
it's
just
a
partner
that
I
choose.
My
definition
is
that
a
small
application
is
10
K
line
of
code
and
medium
application.
A
It's
like
100
K
and,
if
you're
looking
at
a
huge
one,
that
a
team
is
maintaining
and
making
sure
that
it
said
you
know
working
it's
1
million
line
of
code.
This
is
only
the
kernel.
Only
the
kernel
of
the
operating
system
that
you're
running
20
to
millions
line
of
code
right.
If
you're
looking
at
the
DBM
destroy
the
last
tight
I,
just
it
was
490
million
lines
of
code.
This
is
crazy.
Right
I
mean
you
have
a
lot
a
lot
of
day
of
code
there.
A
It's
very
hard
to
reason
about
so
I
tell
you
that
there
is
quite
a
lot
of
inefficient
of.
Are
we
running
stuff
today
and
the
question
is:
how
did
we
get
there
right?
Well,
how
come
so
inefficient?
And
the
answer
is
that
it's
a
pure
volution
right,
we
started
with
the
male
friend,
go
all
the
way
to
the
N
word
in
IOT
devices,
and
all
this
way
what
was
part
of
what
basically
was
walking
out
with
us.
It
was
a
Linux
UNIX.
Actually,
so
what
can
Linux
run
on
anything?
What
can
run
on
Linux?
A
Anything
I
can
take
the
last
destroy
today
off
I,
don't
know
Debian
this
job
put
it
on
paint
ten
years
ago.
It
will
work
right,
it's
just
going
to
work.
I
will
not
need
to
do
anything,
so
we
make
a
decision
as
a
community.
Another
decision
was
that
it's
more
important
to
us
to
support
compatibility
versus
efficiency
right
and
it
works
right,
we're
running
it.
You
are
running
it
in
your
infrastructure,
but
you
know
what
maybe
it's
time
right
now
to
make
it
right
and
then
or
fully
to
make
it
fast
and
that's
great.
A
But
the
question
is
how
ok
we'll
make
it
better?
Can
we
rewrite
the
operating
system?
That's
probably
a
lot
of
work.
What
about
running
going
from
scratch
again
probably
will
be
very
compliment
a
problematic,
or
is
it
and
III
encourage
you
to
think
differently
this?
This
is
like
a
lioness,
the
guy
who
actually
wrote
the
kernel
of
the
operators,
the
Linux
kernel
and
basically,
what
is.
A
This
is
like
an
argument
between
micro
micro,
kernel
and
monolithic
kernel
back
on
the
day,
and
what
I
want
you
to
take
is
the
last
thing
that
he
say,
which
is
basically
that
I
wrote
this
operating
system.
This
kernel,
basically
without
any
knowledge
in
a
year,
which
means
that
it's
not
a
hard.
We
can
do
that
and
that's
exactly
what
the
community
of
the
unicorn
all
did.
They
basically
rewrote
am
or
am
operate
the
system,
so
that,
let's
understand,
is
work
right.
A
So
today,
when
you're
running
your
application,
you're
running
kernel,
you
have
all
the
OS
libraries
and,
on
top
of
it,
you
put
in
your
application,
no
matter
what
you're
running,
which
application
you're
running.
You
always
have
all
this
stack,
but
the
unique
kernel
is
doing
they
say.
Whatever
we
will
look
at
it
differently
will
look
at
the
application.
We
will
see
only
what
this
application
you
need
and
we
will
be,
can
choose
what
we
need
and
that's
exactly
what
they
did.
So
it
would
look
something
like
that
right.
A
It's
only
going
to
take
part
of
the
kernel,
it's
going
only
to
take
part
of
the
other
libraries
and
basically
package
them
together,
so
that
this
is
more
kind
of
like
the
technical
thing.
So
what
is
eunuch
canal?
So,
in
order
to
make
something
like
this,
we
need
to
make
some
design
choice
right
and
now.
A
Design
choice
was
we're
going
to
run
only
one
single
process,
because
if
you
wanna
run
single
process
and
one
user,
which
is
what
we're
doing
in
my
in
micro
services
architecture
today,
then
we
can
be
very,
very
smart
about
reducing
a
lot
of
the
complexity,
so,
for
instance,
so,
first
of
all,
it's
all
based
on
library,
OS.
All
the
drivers
are
library
OS.
It's
kotel
only
what
its
name
it's
a
single
process.
So
if
your
application
is
14,
you
cannot
use
it,
but
it
is
supporting
multi
threads
right
and
by
the
way.
A
So
single
address
space
again,
because
all
this
multi
address
space,
usually
it's
for
support,
protect
right,
I,
need
to
make
sure
that
if
I'm
running
my
process
and
you're
running
your
process
on
the
same
machine,
so
I
want
to
really
make
sure
that,
if
my
that
my
application
will
not
go
to
your
memory
space-
and
you
know,
mess
it
up
right
because
then
your
application
will
be
killed.
No
reach
our
memory,
isolation,
No
context.
Switching
again,
all
those
rings
that
usually
you
have
you
have
the
kernel
mode
versus
their.
A
The
user
mode
is
not
necessary
because
everything
is
going
to
run
on
the
kernel
log
and
the
reason
is
because
those
those
actually
a
rings
it's
again
for
protect
for
protect,
because
if
I'm
running
and
I
want
to
do
something
nasty
to
the
Machine
and
my
application
is
trying
to
do,
nestea
I
need
to
make
sure
that
your
process
is
not
getting
hurt
right.
So
that's
basically
the
ID.
But
again,
if
I'm
the
only
processors
running
on
the
machine,
guess
what,
if
I
will
kill
the
Machine,
it
will
died
and
application
will
not
work.
A
But
you
know
what
so
what
I
mean
I
need
to
fix
that,
but
I
didn't
influence
anybody
else.
I
didn't
take
an
eye
professor
down,
didn't
take
that
out.
You
know
or
sit
down
because
no
one
else
running
in
this
sauce
and
the
beauty
of
it
that
this
is
a
really
beautiful
infrastructure
because
to
unicameral
you
cannot
SSH
right,
because
we
don't
want
to
put
something
that
you
don't
need
there
and
SSA
it's
something
that
it's
very
vulnerable.
A
So
how
do
you
do
that?
Right?
Okay,
so
you
have
your
application
you're,
taking
your
application,
you're
taking
application
constraint,
you're,
taking
application
dependency,
he's
taking
your
language
runtime,
and
your
drive
and
Driver
basically
put
all
of
them
in
kind
of
like
a
magic
tool
and
then
creating
a
unique
arm,
and
you
cannot
can
run
its
bootable
image,
so
it
can
run
on
bare
metal.
But
do
we
really
want
to
take
an
application
and
56
Meg
and
dedicate
a
full
Austria
meta?
Probably
not
that's!
A
Why
we're
going
to
use
an
AI
provider
right,
because
that
just
makes
sense
again
the
beauty
of
what
I
per
visor
is
that
it's
given
us
a
solution
on
the
physic
physical
isolation
right.
So,
if
you're
looking
at
so
that's
very,
very
secure,
so
you're
getting
the
security
of
VM
because
it
is
VM
right
again
the
security
of
the
VM
you're.
A
So
what
did
we?
Do?
We
just
remove
what
we
don't
need.
That's
basically
it
we
just
stood
the
stack
said.
All
of
this
is
not
necessary,
let's
remove
it
and
if
you
look
at
the
stock
right,
the
stock
of
we'll
be
different
right
now,
much
much
more
slim,
much
more
down
right
and
if
you
wanna,
if
someone
will
say
hey,
we
can
run
it
on
bare
metal
and
then
anyway,
we're
making
it
small.
A
So
not
exactly
exact
here
is
that
it's
the
physical
isolation-
and
this
is
key
and
I
provider-
is
the
one
who
knows
actually
today
how
to
do
a
physical
and
basically
leveraging
the
CPU
to
do
a
physical
Hardware
isolation,
which
is
very,
very
important.
So
again
summarize
the
unique
kernel
there
is
no
other
user
there's
no
a
multi-user
support.
A
You
can
put
it
on
your
application
level,
but
it's
not
on
the
OS
level,
no
permission
check,
which
is
great,
because
now
you
can
actually
utilize
one
percent
of
your
of
your
advert
to
your
application,
so
performance
of
ever
strong
isolation,
an
aim
in
the
individual
harddrive
only
we
mean
that
you're
sharing
all
only
hardware
and
minimal
if
you're
looking
at
size
minimum
over
the
system.
Today
that
you
will
take
instance
or
whatever,
it's
probably
one
gig,
at
least
when
we're
talking
about
única
and
we're
talking
about
casing
size.
A
Basically,
your
application
size
is
the
size
of
your
unicorn.
They
should
the
boot
time
is
very
short
because
there
is
not
a
lot
to
put
basically
what
we
did.
We
kind
of
like
wrapping
it
with
a
little
bit
of
memory
management,
a
little
bit
of
scheduling.
So
it's
really
it's
like
two
thousand
line.
You
know
it's
nothing
on
your
application,
which
it
should
short
boot
time,
it's
very
interesting
to
something
like,
for
instance,
several
s
or
lambda
function
as
a
service.
A
Whatever
it's
been
called
because
today
they
are
a
reusing
container
with
unicorn,
you
donated
it's
fast,
it's
very
very
fast
when
it's
boring,
so
you
can
actually
dedicate
machine
full
function.
The
the
surface
of
attacked
is
very
tiny,
but
it's
also
custom,
which
is
very,
very
important,
which
means
that
for
one
application
that
said
that
I
have
an
application
and
somehow
managed
to
penetrate
to
it.
It's
not
going
to
work
for
me
on
a
different
unit.
Kernel
right
because
there
were
snipers,
are
different.
A
So
it's
much
harder
for
me
to
guess
what
will
be
there
and
what
not
and,
as
I
said
before
their
most
important,
it's
a
it's
a
really
immutable
infrastructure.
It's
perfect
fit
for
what
we're
running
today.
That's
why
it's
starting
to
be
successful.
Okay,
so
our
there
was
no
question
no
question
for
now
good.
So
there
is
two
type
of
unicameral
today
in
the
ecosystem.
One
of
them
is
what
called
language-specific.
A
You
probably
all
learn
about.
Your
aim,
unicameral
system
that
got
bought
by
acquired
by
docker
they
running
a
unique
and
I
call
me
Roger
s
I'll,
it's
very
good.
Unique
are
not
the
only
thing
with
it
is
that
you
have
to
write
your
code
in
your
comma,
because
that's
the
way
it
works.
You
know,
comma,
not
a
lot
of
people
using
this
language
today,
so
that
limitation
include
OS
is
another
great
I.
Am
a
Munich
colonel,
it's
a
simple
space,
unicorn
and
again.
A
The
other
option
is
to
go
with
something
called
Posse
compliance,
so
Bostic
is
like
they
operate,
the
API
of
the
operating
system
and
what
you
can
do
is
basically,
if
your
application,
it
can
run
on
pass
a
companion
to
run
on
this
unique
honor.
A
great
example
is
ROM
prong.
This
is
probably
the
best
one
or
sweets
kind
of
única,
not
really,
but
it's
good
for
Java.
So
again
we
run
on
unique
and
I'll
go
a
Ruby
Python,
whatever
just
name
it
like
C++
Java,
just
name
it.
A
We
can
run
it
on
unicorn,
oh,
so
this
is
a
performance
benchmark,
just
an
example.
So
why
sweet?
So
it's
unique
and
that
God
brought
by
a
company
called
Claudius
in
Israel
and
they
were
all
about
journey
kernel
and
about
a
code.
It
is
until
doctor
came
and
then
they
kind
of
like
freaked
out.
So
what
did
they
do?
They
pivot
right
and
their
pivot
said
we're
going
to
take
a
Sandra.
A
We
will
rewrite
it
to
run
on
unique
kernel,
we
will
call
it
Scala
DB
and
we
will
get
better
performance
and
that's
exactly
what
they
did
and
actually
it's
actually
10
times
better.
So
this
is
like
a
huge
put
in
a
performance.
A
oh.
We
went
time,
okay,
so
this
is
an
example
for
cool
project
in
unicorn.
Oh
one
of
them
is,
is
the
guy's
former
address.
They
basically
wanted
to
show
you
our
securities.
So
what
did
they
do?
A
They
created
any
kernel,
the
pinata
they
put
some
Bing
code
inside
and
they
said
to
yoga
I
try
to
attack
it,
and
this
is
there
I
think
for
over
two
years,
no
one
managed
to
actually
press
right
through
it.
They
managed
to
take
it
down,
and
then
it's
went
up
again
again,
smarter.
So,
in
the
end
of
the
day,
it's
a
very,
very
secure
in
terms
of
security,
so
I
mean
you
can
go.
Thank
you.
What
if
I
would
go
here?
Basically,
you
can
go
and
look
at
it
and
try
to
take
it
down.
A
So
that's
that
the
other
unique
kernel,
it's
a
it's
air
link,
I
think
what's
cool
about.
It
is
what
they
put
in
your
statistics
to
show
up
quick,
it's
actually
boring
so
300
second,
is
how
much
you
taking
you
taking
an
instance
up
in
the
Amazon
ec2
15.
S.
Second,
is
how
much
you
power
the
lock
screen
from
your
android
and
0.2
second
ago.
I
click
this
button
and
then
I
did
the
unicron's
already
up
and
running
right.
A
So
just
give
you
an
example
of
how
quick
you
can
you
can
you
can
actually
spin
up
your
Nakano?
Ok.
So
what
did
we
do?
Ok,
so
I
looked
at
it
and
I
said
this
is
very
wait
right.
This
is
sounds
like
this
is
the
right
of
you,
texture
for
the
cloud
and,
of
course,
Internet
of
Things,
because
in
Internet
of
Things
there
is
three
parameter
that
very
interesting.
The
tuna
kernel
is
is
very
good
word.
One
of
them
is
very
slim
when
it
taking
stuff
with
you
actually
think
video.
A
You
want
it
to
be
small,
so
you
cannot.
You
know
you
don't
have
an
endless
storage.
Therefore,
it's
really
good
if
we,
if
the
application
is
not
taking
this
storage,
for
your
potassium,
is
not
taking
the
storage.
The
second
thing,
it's
very
good
in
performance
and
the
last
thing
it's
very
secure.
All
these
these
things
make
it
very,
very
good
for
Internet
of
thing.
So
again,
the
question
is:
what
do
we
do
when
we're
lucky
that
we
understand
that
that
was
very,
very
you
know.
That's
exactly
the
same
scenario.
That
container
was
before
doctor.
A
Make
it
easy
to
use
it's
just
very
hard
right.
You
need
to
compile
type
of
a
drivers.
You
need
to
make
all
the
mess
that
I
just
described
it's
very
very
hard.
So
what
we
wanted
to
do
is
make
it
easy
for
you
to
do
it,
so
we
will
do
all
the
hard
work
for
you
and
that's
exactly
what
unique
is
is
about.
So
it's
a
go.
It's
reading
it
go.
It's
open
source
projects
that
will
give
you
the
link
very
soon,
and
it's
very
simple.
A
So
how
does
it
work
you're,
taking
you're,
basically
doing
unique
bills,
you're
building
an
image
like
docker,
so
you're,
giving
the
application
the
code
of
the
application
like,
as
you
see
here
example
after
instance,
the
folder
you
tell
me
which
unique
kernel
type
you
want
so,
for
instance,
maybe
you
want
to
run
into
name
or
address.
Maybe
you
I'm,
going
to
run
into
run,
run,
make
choice.
This
is
your
choice
and
then
which
language
your
application
is
reading
in.
A
So,
as
I
said,
you
will
go
to
our
website
web
base
agita,
but
basically
we're
supporting
I
mean
almost
everything
that
it's
out
there,
except
of
Ruby
and
not
for
good
reason.
Just
because
it
didn't
do
that
and
then
in
terms
of
provider
which
provider.
What
do
you
want
to
run
it?
So,
do
you
want
to
write
an
AW
ask
the
owner
to
locally
on
your
virtual
machine?
A
Do
you
want
to
run
it
and
Google
just
tell
us
where
we
will
do
that
for
you
and
then
the
last
one
give
me
a
name
of
the
image
so
now
I
will
be
building
it
with
making
the
magic
and,
in
the
end
of
the
day,
are
you
getting
some
image
and
now
what
you
can
do
it
just
you
know,
run
so
run
your
image.
So
it's
that
simple.
A
This
is,
for
my
opinion,
the
most
important
slide
in
my
talk,
because
I
think
that
what's
special
about
unique
is
the
way
we
actually
build
it
and
surprise,
surprise.
We
were
all
very
inspired
by
the
kuva
Nate's
architecture,
because
we
feel
that
this
is
the
best
which
is
very,
very
kind
of
like
not
tied
together.
Right,
everything
is
an
is
a
is
a
interface
and
that's
exactly
what
that
this
is
our
main
interface
in
unique.
So
one
of
them
is
the
unicorn
type.
We
don't
know
which
one
is
the
best.
A
We
have
our
guest,
we
have
our
favorite,
but
but
this
is
a
very,
very
new
community,
so
we
don't
know
you
know
we'll
pick
up.
Therefore,
we
didn't
want
to
make
a
choice.
So,
for
instance,
we
supporting
out-of-the-box
include
OS
include
OS
is
actually
a
it's.
Actually,
a
pull
request
form
include
OS
itself,
so
very,
very
simple.
To
add
it.
It's
basically
creating
a
docker
content
and
everything
it's
very,
very
easy.
A
Marah
gos
run
for
an
anniversary
in
terms
of
fire
of
cloud
provider.
So
we
didn't
want
to
choose
that
as
well.
Right,
I,
don't
know
what
you
want
to
run
so,
right
now,
out
of
the
box,
we
were
running
on
a
OpenStack.
That's
a
pull
request
for
my
project.
You
actually
that's
using
it
VirtualBox.
If
you
want
it,
local
AWS
photo
an
ATM,
see
it's
not
it's
it's
we
aware
and
you
DC
that's
supposed
to
scale
vCenter
human
for
debugging,
so
we
are
debugging
Unicare.
Now
that's
a
we
building.
A
What
we're
building
and
right
now
we
automate
only
the
QM,
you
only
the
QM,
you
debugger,
but
actually
we
can
do
it
for
anything.
So
it's
not
true
that
you
cannot
be
buggy
any
camera,
we're
doing
it
all
the
time
and
KVM
again
if
you
want
to
run,
but
the
beauty
of
it
is
that
today,
I
will
announce
that
we
are
also
supporting
Google
cloud.
So
this
is
new.
A
Because
it's
a
little
bit
different,
the
docker
hub,
because
there
is
no
layers
on
the
filesystem
right,
it's
basically
image.
So
this
is
an
example.
You
know,
so
it's
very
simple:
it's
pull
push
open
source
you
can
go
play
with
it
contribute
to
it,
which
is
even
better,
and
what
you
see
here
is
that
I'm
pulling
minecraft
and
running
it
on
your
Nick
Cannon.
So
that's
what
this
slide
is
about,
and
and
ok,
so
they
like
that.
The
other
thing
that
we
did
is
basically
wanted
to.
A
We
understand
that
the
everybody
will
will
like
the
doctor.
Api
really
really
feel
comfortable
with
that,
and
they
were
already
script
at
running
with
with
the
docker
API.
So
we
didn't
want
you
to
need
to
learn
new
API
and
therefore
what
we
did
we
basically
create.
We
teach
unique
how
to
speak.
Docker
API
that
simple.
Now
there
is,
of
course
limitation,
because
the
reason
we
created
unique
API
is
because
unique
it'll
be
different,
write,
unique,
unique
and
need
to
know
where
it's
running
on
it's
not
infrastructure.
Independence.
Docker
is
right.
A
You
just
need
to
run
it
on
on
OS,
so
you
kept
the
sagar
phone.
It
is
that
you
will
not
be
able
to
build
with
the
doctor
api
because,
as
a
set,
it
just
doesn't
make
any
sense,
but
you
will
be
able
to
run
and
PS
and
so
on,
and
that's
what
you
see
here
in
the
slide,
and
today
we
also
going
to
announce
that
we
support
core
binaries
so
unique
and
are
very
great
and
unique.
A
Is
very
great
to
make
Unicom,
but
it's
like
doctor
making
a
container
it's
not
going
to
make
sure
that
it
will
stay
alive.
It's
not
going
to
make
sure
that
then
you
know
the
the
health.
The
monitor
at
all
of
this
is
not
something
that
the
unicorn
you
know
unique
is
going
to
do
for
you.
So
that's
what
kuba
net!
That's
why
we
decided
to
leverage
kubernetes.
So
basically,
the
idea
with
this
is
right
before
you
could
around
rockets
or
docker
container
in
the
pod.
A
Now
you
can
run
your
Nakano
in
fun,
so
I
will
show
them
a
real
quick.
So
so
what
did
we
do
there?
Because
it's
kind
of
like
interesting
right,
unique
and
it's
basically
a
VM.
So
how
does
it
work
so
the
way
to
work
is
basically
I'm
trying
to
show
it
it's.
Basically,
the
idea
is
that
it's
still
talking
to
the
couplet,
but
the
couplet
is
talking
to
your
neck
and
making
us
to
run
the
VM.
A
So
that
means
you
can
actually
run
a
container
side
by
side
next
to
eunuch
Colonel
in
your
keuken
and
your
in
your
in
your
kubernetes
and
the
beauty
of
it.
Is
that
think
about
it
when
you're
running
on
couplet,
you
still
need
the
node,
so
if,
for
instance,
you
over
located,
you
need
to
put
more
nodes
in
uniquely,
not
right
when
your
application
is
running
you're,
creating
unique
application
running
you
created
kubernetes,
oh,
so,
basically,
there
is
no
node
notion
here.
So
I
will
show
you
again
a
demo.
No
actually
any
question
by
the
way.
A
Welcome
why
sorry
yeah
we
took
the
last
one?
Yes,
but
it's
open
source
like
go.
Look.
What
we
did
I
can
show
you
if
you
want
okay,
so
what
we're
going
to
do
is
we're
going
to
open
a
command
line.
So
first,
let
me
show
you
what
you
think
it
is
that
big
enough.
Can
you
see
it
okay?
So
what
I'm
going
to
do
is
I'm
SSH
to
a
machine
that
running
your
neck?
What
we're
going
to
do
it
is
unique.
Ps
and
you'll
discover
that
there
is
no
unicorn
running.
A
We
can
do
unique
images,
that's
the
one
that
I
built
before
so
I
just
go
and
run
something
unique
crying,
and
they
just
to
know
that,
as
you
see
it,
just
I
put
instance,
name
will
be
did
because
it's
my
name
and
nodejs
application
just
for
York
of
it,
so
I'm
clicking
it
and
then
it's
up.
So
this
is
specifically
running
on
AWS.
If
we
will
do
yes
again,
I
will
see
I'm
getting
an
IP
address
and
I
can
just
go
and
run.
A
B
A
Bigger
because
it's
always
finish
yeah,
because
it's
always
free,
so
we
work
in
rightness,
hi
sweet.
The
other
part
is,
as
I
said,
it's
kind
of
unique
kind
of
it's
kind
of
like
a
carrot,
the
way
we
build
it
and
we're
going
to
change
it.
Actually,
we
have
already
pull
request,
for
that
is
that
we
kind
of
like
not
guessing
the
side.
We
just
said
we're
going
to
take
these
sides
and
we're
going
to
build
this.
This
is
the
way
I
switch
worked.
A
So
this
is
not
really
working,
so
it's
always
going
to
be
that
big,
but
we
already
have
a
pull
request
for
it,
so
it's
going
to
be
dramatically
shrink.
So
this
is
the
website
right.
It's
running
on
AWS.
Now
the
beauty
of
it
and
I
want
you
to
understand
it's
actually
running
on
the
Zen
of
day
ws.
We
didn't
put
any
layer,
there's
no
hypervisor
between
it,
so
like
they
actually
instances
running
via
Karen
on
the
Zen.
A
Okay,
and
as
you
see,
this
is
a
website
56
mega
guy
again
I'm,
not
sure
that
there
is
a
lot
of
website
in
that
small
running
there.
Now
what
we
can
do,
we
can
actually
go
in
a
I,
don't
know
this
is
a
demo
that
usually
I'm
doing
run
another
unique
now
unique,
also
support
log.
So,
if
you're
doing
unique
clocks,
you
can
get
all
the
logs
of
all
the
stuff
which
is
kind
of
cool
right.
A
A
Or
something
what
we
missed
up
the
address
so
we'll
show
you
that
in
a
minute,
because
I
probably
would
yeah
no,
but
that
should
be
fine,
I
think
he's
just
weak
or
you
know
it's
good
I,
don't
know
what
happened!
Okay,
so
right
now!
This
is
your
new
kernel
running
Minecraft
right
now.
The
beauty
of
it
is
the
right
now
I
can
go
to
this
address,
just
any
browser
and
when
I'm
taking
it
I
can
do
nine
nine
six,
seven
port,
slash
logs
and
you
will
see
all
the
logs
of
the
yeren
account.
A
So
I
don't
know
if
you
can
see
it
so
big,
but
it's
back
if
I
run
it.
So
it's
basically
the
ability,
unique
Adam,
started
running
the
operating
system.
I
during
the
game
zombie
write
me
and
so
on.
So
basically
I'm
getting
all
the
log
which
is
kind
of
cool
right.
So
that's
that
let's
remove
it
so
now.
The
next
thing
that
we're
going
to
do
is
we
show
you
the
kubernetes,
so
okay,
so
I'm
going
to
SSH
to
a
new
machine
that
I
am.
A
A
Run
so
basically
I'm
just
running
unicameral.
You
can
see
this
so
I
run
two
of
them
okay,
so
we
will
go
right
now,
for
instance,
to
the
kubernetes
dashboard
and
we
refresh
it
and
you'll
see
that
I
have
three
instance
of
pods
running
now.
What
I
can
do
is
do
this,
I
will
come
here
and
I
will
watch
unique
PS.
A
A
A
A
Instance
and
I'm
going
to
choose,
one
of
them
doesn't
really
matter
which
and
then
I
will
do
force,
because
it's
running
and
what
you
will
say
right
now
is
I
will
delete
it
there
and
you
see
how
Cove
is
I
think
right,
not
that
magic,
so
I
delete
it
here.
If
it
would
go
here,
you
will
see
now
we
have
only
doing
in
a
second
tack.
It's
return
rate
right,
second,
okay,
so
that's
basically
the
crew
Venus.
We
have
really
live
okay,
so
I
will
go
because
I
need
that
I
have
five
minutes.
A
I
will
go
over
quick
to
the
presentation,
because
I
want
you
to
see
a
little
bit
more
stuff.
So,
first
of
all,
this
is
very
interesting
to
me.
We
understood
that
the
unicorn
level
is
perfect,
fit
for
an
internet
of
things
we
wanted
to
make
it
instructs.
Take
it
off
talking
about
it,
actually
make
it
done.
So
what
you
see
here,
it's
a
laptop
we're,
basically
putting
a
unique
kernel
on
top
of
Raspberry
Pi
and
guess
what
black
screen.
We
cannot
see
anything
okay,
we
can't
even
debug
it
right.
A
So
what
we
did
we
connected
another
Raspberry
Pi
with
a
gdb
and
start
to
debug
it
and
basically,
usually
in
a
conference-
and
you
can
go
online
and
see
what
I'm
doing
is
this
demo?
Basically
you're
making
unique
are
no
unique
running
me
a
toaster
on
stage
and
you
know
making
it
those
East
Austin
State,
but
I
don't
have
time
to
do
that.
So
go
look
in
the
internet,
but
that's
very
interesting.
So
everybody
probably
heard
about
the
last
DDoS
attack
the
to
code,
the
internet
down.
A
We
were
a
great
blog
on
it
and
why
it
will
not
be
happen
if
which
was
running
the
unique
online
instead.
So
I
really
really.
You
know
I
really
really
encourage
you
to
go
and
look
at
this
because
I
feel-
and
this
is
on
the
wiki,
but
I
have
a
website
so
I
get
up,
so
you
can
just
go
there.
Okay!
So
that's
one
thing!
The
second
thing
is
that
when
we
look
at
that
announcer
for
Microsoft,
we
said-
oh
that's
interesting
right.
Microsoft
did
a
good,
very,
very
good,
huge
favor
for
us.
A
They
basically
make
dotnet
running
or
net
BSD.
Now
rump
language
is
my
favorite
unique
aroma.
Basically,
what
it
is
is
that
it's
net
bsd
they
did
a
rewrite
to
the
to
the
to
the
operating
system
and
they
basically
component
it.
So
now,
if
you
have
component,
you
can
pick
and
choose
what
you
need
right.
So
basically
they
using
the
drivers
or
net
bsd
if
the
using
the
driver
would
that
be.
A
Is
its
mean
that
it
shouldn't
be
so
hard
for
me
to
run
dotnet
on
on
a
unicorn
and
that's
exactly
what
we're
working
on
it
right
now
so
there
it
is
that
think
about
it,
no
license
to
microsoft
whatsoever,
you're
not
running
on
the
server
you're,
getting
a
very,
very
quick
demo
of
experience
and
you're
running
dotnet
application.
So
I
think
this
is
really
cool.
A
The
other
thing
is
that
we
talked
about
it
a
little
bit
several
s
right,
so
you
probably
all
know,
but
several
as
lambda
and
all
of
this
they
using
reusing
container
because
it's
hard,
you
know
it's
taking
time
to
put
them.
You
don't
need
to
do
that
with
you
Nicholas,
so
we're
working
right
now
probably
will
open
source
very
soon.
Basically,
a
several
less
implementation
with
a
back-end
or
you
Nakano,
so
you
will
make
a
choice.
A
You
can
still
run
container
if
you
like,
but
you
can
also
run
that,
and
that
makes
it
actually
easier
right
because
there's
no
using
it
so
on
it's
open
source.
It's
actually
we
open
source
it.
They
want
to
say
five
months
ago
four
months
ago,
and
this
is
my
repository-
it
make
team
repository.
If
you
will
go,
you
will
see
that
we
are
actually
a
thousand
stars.
There
is
people
coming
all
the
time,
getting
a
lot
of
attention
and
there
is
a
community.
So
please
please
come
help
us
make
it
better.
A
So
what
we
did
so
what
I
did
I'm
the
marketing
person,
apparently
also
so
I
did
a
a
slack
channel.
We
have.
We
are
always
there.
We
are
helping
question.
You
know
your
answer
will
be
always
answered:
I
open
an
a
Twitter
account
for
unique.
There
is
people
there,
it's
actually
probably
more
than
double
than
what
you
see
here
in
the
number
and
as
I
said
again,
they
get
up.
We
answering
any
issues,
and
this
is
my
team.
This
is
all
the
people
who
did
they
do
real
job
here.
B
A
B
A
A
Yes,
yes
and
I
wanted
to
show
it
so
in
AWS,
when
we
actually
building
the
unica,
are
we
creating
an
a
on
my
and
then
spin
it
up
now?
Just
you
will
know
it
doesn't
really
make
sense
to
run
on
AWS
I
want
you
to
know
that
and
the
reason
is
because
I'm
running
50
second
50
minutes
but
I,
actually
paying
for
one
gig,
doesn't
make
sense
whatsoever.
So
that's
not
the
cloud
that
I
will
choose.
Google
cloud
up
fully
can
help
us
better.
I
mean
to
be
fair.
A
B
A
Yeah
yeah,
look,
we
we
didn't
do
pol
quest,
yet
we
did
the
changes
that
we
needed
in
couplet,
and
so
on
I
mean
we
it's
there.
It's
open
source,
as
you
see,
no
one
is
because
it's
an
answer
go
help
help
us,
but
the
beauty
of
it
is
what
we
did.
We
basically
added
to
support,
so
you
for
unique,
but
because
we
wanted
to
run
the
actually
dashboard
and
song.
So
what
we
did?
We
also
running
unique
and
that's
running
a
docker.
So
what
we
did?