►
From YouTube: Visualizing Kubernetes - Patrick Reilly
Description
Recorded on Feburary 25th, 2015 at the Kubernetes Gathering in San Francisco, CA, USA
A
You
guys
hear
me:
okay,
I'm
Patrick,
Riley
I'm,
the
CEO
of
cos,
Matic
we're
a
new
startup,
we're
focused
we're
focused
completely
on
kubernetes,
where
we
jokingly
refer
to
ourselves
as
the
kubernetes
company.
That's
not
Google.
One
of
the
reasons
we
really
double
down
on
kubernetes
is
it's
lean.
It's
really
lightweight.
It's
simple,
its
accessible!
It's
written
in
go
people
can
understand
how
go
works.
You
can
look
at
the
code.
You
can
send
in
pull
requests,
it's
a
great
vibrant
community
and
it
seemed
like
a
great
place
to
start.
A
A
Try
things
out
and
you
just
want
a
way
to
kind
of
see
what
the
cluster
is
doing
and
and
visualize
that
so
we
decided
to
work
on
the
web
UI
and
come
up
with
a
a
nice
framework
for
the
web
UI
that
can
be
extensible
in
built
on
and
take
some
of
the
features
that
are
available
in
kubernetes
and
make
them
really
accessible
to
everyone.
So
with
that
said,
I'll
show
you
just
a
brief
intro.
A
So
when
you
first
launch
a
cluster,
you
have
various
notes
and
you
kind
of
want
to
add
a
glance,
see
the
health
of
them.
So
we
came
up
with
this
visualization
to
just
show
the
health
of
the
nodes.
So
what
you're
seeing
here
is
the
overall
CPU
memory
and
filesystem
utilization
on
each
one
of
those
nodes
and,
as
you
would
add,
more
nodes
to
the
cluster
you'd
be
able
to
see
the
health
statistics.
A
So
you
can
see
it
when
it
read
there
that
particular
node
had
an
issue,
but
now
it's
recovered,
but
you
can
see
that
it's
still
kind
of
like
the
shadow
red
color,
because
you
recently
had
a
problem.
So
you
know
that
you
might
want
to
take
a
look
at
those
nodes.
So
our
goal
with
this
project
was
to
make
it
modular,
be
able
to
write.
Anyone
can
write
components
and
they
become
new
tabs
in
the
interface.
A
And
if
you
look
over
in
the
it's
kind
of
hard
to
see
on
the
projector,
but
in
the
right
hand
side
we
have
views,
so
we
can
go
through.
All
of
the
different
commands
that
you
can
do
with
the
CLI
and
see
them
in
the
web
interface
and
again,
so
let
me
just
jump
forward,
so
you
can
see
you
know
we
can
dig
into
the
group's
page.
We
can
start
to
filter
everything
is
paginate
able
it
uses
the
the
rest
api,
so
we're
getting
back
to
json
information
parsing
it.
So
it's
easy
to
see.
A
We
have
the
control
over
here
on
the
side
to
get
to
more
features
if
you're
digging
into
a
component.
So
I
just
recorded
this
video
earlier
today
wanted
to
go
through
and
show
you
know
some
of
the
things
you
can
do,
but
our
goal
with
the
project
is
anything
you
can
do
on
the
CLI.
You
can
do
in
the
web
interface,
so
you
could
actually
administer
a
cluster.
A
You
know
do
everything
that
you
want
to
do
without
ever
having
to
touch
SSH,
and
you
know,
as
we
drill
into
into
some
of
these
sections,
you
get
more
information.
You'll
see
the
breadcrumbs
up
in
the
top,
so
you
can
actually
navigate
back.
So
you
know,
as
far
as
you
drill
into
something
you
can
go
all
the
way
back
and
some
of
the
integration
points
that
were
we're
wanting
to
do.
Is
you
know
this?
A
A
What's
going
on
replace
the
image
you
know
do
whatever
they
have
to
do,
you
know
for
maintenance,
so
we
felt
like
this
was
a
big,
critical,
missing
piece
with
kubernetes
and
we
wanted
to
have
this
available
and
the
project
is
all
written
with
angular
and
it's
a
material
design.
So
when
you
want
to
add
a
new
tab,
you
want
to
have
a
new
component
into
the
system.
A
We've
come
of
a
manifest.json
file
that
you
write
and
you
basically
create
your
application
and
once
that
that
manifest
is
loaded
in
the
system
it
auto
wires
up,
it
becomes
a
new
tab.
So
you
can
see
you
know
here
in
the
top
tab
bar.
We
have
the
graph
tab
which
Jack
Greenfield
from
Google
is
going
to
come
up
and
talk
a
little
bit
about
that
and
his
team
was
able
to
do
some
really
amazing
stuff
and
the
intersection
between
what
we
were
working
on
with
a
dashboard
and
what
he
was
doing.
It
was
really
clean.
A
You
know
once
we
had
those
interfaces
you
know
designed,
he
was
able
to
just
have
his
team
go
off
and
build
it,
and
one
of
the
unique
things
that
we
have
with
the
kubernetes
cluster
and
by
using
angular
for
this
project
is
that
you
can
run
these
tabs
on
different
pods
in
the
cluster,
because,
through
the
8
through
the
API
proxy,
it
looks
like
it's
a
local.
You
know
it's
a
local
resource
so
that
you
don't
have
any
cross.
A
You
know
any
course
issues
you're
able
to
load
it
up,
just
fine,
so
this
particular
dashboard
could
be
running
across
many
different
pods.
You
know
in
your
system.
Each
of
those
tabs,
you
know,
is
fault
tolerant.
So
our
goal
with
this
project
is
to
put
it
in
the
main,
kubernetes
repo,
basically
replace
the
dub
dub
directory.
That's
in
there
now
and
make
it
available.
You
know
to
everyone
to
build
their
things.
A
You
know,
on
top
of
probably
one
of
the
next
things
that
we're
gonna
look
to
do
is
really
define
all
of
those
destructive
changes
that
you
can
do
so
have
the
edit
mode
that
you
can
click
into
when
you're,
not
just
viewing
the
cluster
and
do
any
of
those
commands
that
you
want
to
do
and
then
tightly
integrate
with
Google
storage
and
the
elastic
search
to
get
any
of
the
the
logs
or
anything
that
that
you've
stored.
So
weird
feedback
on
the
mic.
A
Ok,
anyway,
and
you
know
we're
hoping
that
in
about
a
month
we'll
have
this
open
sourced
and
in
the
repo
and
anyone
can
go
in
and
start
hacking
away
and
adding
their
components
and
we'd
like
to
see
both
paid.
You
know
proprietary
components
that
people
could
build
if
you're
building
something
to
interface
with
the
openshift
and
what
they
just
showed.
A
You
know,
or
if
your
particular
cloud
provider-
and
you
have
features
that
you
want
to
you-
want
to
put
in
here-
it's
easy
to
load
those
up,
and
you
know
we're
hoping
that
once
we
open
source
that
people
have
interesting
feedback
and
and
give
us,
you
know
direction
to
go
so
one
last
story.
They
wanted
to
tell
us
a
little
bit
about
these
bar
gauges
that
we
built
I
came
up
with
the
idea.
How
do
we
visualize
the
health?
A
You
know
of
a
particular
mania
in
her
note
and
started
it,
you
know,
did
a
rapid
prototype
and
then
one
of
our
people
at
Google
that
we
were
working
with
Kenmore
did
a
pie
in
the
sky
like
if
this
was
the
perfect
bar
gauge.
This
is
how
I
think
it
would
work,
and
he
posted
that
we
took
a
look
and
I
took
it
as
a
challenge.
A
I'm
like
I,
think
I
can
do
that
with
d3,
and
so
my
co-founder
BC
Broussard
at
kiss
Matic
and
I
both
started
hacking
away
on
it
and
our
goal
was
to
try
to
implement
his
design
as
fast
as
humanly
possible
and
about
12
hours
later.
We
had
it
working,
we
got
excited,
we
put
it
up,
you
know
another
12
hours
went
by
and
can't
actually
woke
up.
He
got
on
chat
and
he
was
blown
away
he's
like.
A
Oh,
my
god,
you
actually
did
what
you
you
know
what
I
thought
was
a
pie
in
the
sky,
design,
I
didn't
think
you
could
actually
pull
that
off
and
we
felt
that
it
was
a.
You
know,
a
nice
interaction
between
our
team
in
Google
and
trying
to
get
the
best.
You
know
stuff
into
kubernetes
and
and
just
in
closing
this
is
the
the
piece
I
really
thought
was
missing.
You
know
getting
people
to
try
new
stuff,
getting
them
comfortable,
going
to
a
CLI
learning
a
bunch
of
commands.
That's
a
lot!
A
That's
a
lot
to
do,
and
especially
when
there's
no
safeguards
like
you
know
sure.
If
someone
can
run
get
minions
and
get
back
a
list
of
minions,
but
do
you
really
want
to
trust
someone
with
that
same
access
to
you
know,
do
something
potentially
destructive
to
the
cluster
that
you
spent
days.
You
know
figuring
out,
you
don't
yet
have
you
know
a
good
way
to
replicate
it
and
bring
it
back
up.
So
hopefully,
this
project
will
will
be
a
good
starting
point
and
will
get
more.