►
Description
Lightning Talk: The Top 5 Kubernetes Metrics to Monitor - Jorge Salamero Sanz, Sysdig
Data! There is so much of it. Especially given the move to distributed microservices, you're faced with an order of magnitude more data to deal with. How can you simplify your top-level monitoring focus into something manageable?
In this talk I'll focus on 5 of the top metrics that I believe will give you insight into whether or not you may have a problem in your environment.
About Jorge Salamero Sanz
Jorge enjoys monitoring all the things, from his Kubernetes clusters to writing sensors plugins and DIY projects with Raspberry PI and ESP8266. Currently he is part of the Sysdig team, and in the past was one of the promoters of HumanOps. When he is away from computers, you will find him walking with his 2 dogs across the countryside.
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
Okay,
thank
you
for
the
small
wait.
I'm
Horace
Alameda
I
work
for
sis
take
a
bunch
of
different
things
there.
If
you
don't
know
sis
take,
we
are
a
container
and
cover
neat
this
monitoring
tool.
It's
awesome.
We
did
quite
a
few
beautiful
things,
I'm
not
going
to
talk
about
the
tool
in
the
next
five
minutes,
just
walk
by
our
booth
and
I'll.
Happily
show
you
the
tool
I'm
going
to
share
my
experience
today.
A
After
talking
with
many
of
our
customers,
how
they
are
monitoring
kubernetes,
what
the
thing
that
is
important,
their
experience,
the
challenges
they
found
and
I
put
this
like
top
five
metrics
to
monitor
in
kubernetes.
But,
first
of
all,
we
need
to
get
a
little
bit
into
the
scope.
Some
I
found
some
people
that
they
were
saying
all
infrastructure
monitoring
the
monitor
is
mostly
forces
have
min
order,
ops
and
some
other
folks.
They
were
saying
oh
well
developers
they
should
be
using
APM
and
nothing
else.
A
The
tree
is
that
more
and
more,
and
especially
using
tools
like
kubernetes,
is
developers
the
ones
that
they
are
in
charge
of
running
the
services
they
use
on
their
applications.
So
we
need
tools
that
they
cover
and
they
have
our
visibility
across
all
the
stock.
Then
using
cover
Nittis.
We
have
seen
a
few
challenges.
A
few
things
near
first
of
all
is
visibility
inside
of
containers,
but
obviously
our
kubernetes
is
bringing
new
layers
of
complexity
of
infrastructure
that
we
need
to
monitor.
A
We
have
new
services
and
containers
are
or
or
services
the
Y
are
not
coupled
with
hosts
like
before.
We
have
a
static
host
and
our
service
run
in
there,
and
that
was
completely
coupled
now
we
have
cover
needs
or
schedule
in
containers
across
the
different
nodes
and
then
to
have
full
understanding
of
your
infrastructure.
Your
monitoring
tool
needs
to
understand
how
is
design
the
different
services
deployment
namespaces
all
that,
so
it's
a
must
that
your
monitoring
tool
needs
to
understand
kubernetes
method
8.
A
So
that
said,
we
have
at
least
5
layers
that
we
need
to
monitor
when
bringing
cron.
It
is
interaction.
First
of
all,
application
application.
It's
not
different.
What
we
were
doing
before
are
the
top
working
metrics.
We
just
need
to
make
sure
your
application
is
working.
There
are
quite
a
few
things
we
could
monitor,
but
I
have
found
that
most
of
the
people
they
are
monitoring
request
time.
If
the
application
is
fast,
it
works.
That's
great,
the
second
layer,
its
services
are
those
databases,
storage,
the
different
services.
A
Your
application
is
running
it
smooth
the
most
important
here
again,
there
are
a
lot
of
few
metrics,
but
the
most
important
is
to
make
sure
that
connections
are
going
through.
So
probably
monitoring
connection
on
those
databases,
it's
a
good
thing
to
the
third
point
it's
going
to
be,
and
this
is
the
new
part
as
a
new
part.
A
Kubernetes
infrastructure
kubernetes
is
organized
in
the
different
services
and
it
is
task
to
make
sure
that
we
have
enough
instances
enough
pods
running,
but
how
we
can
make
sure
that
kubernetes
is
doing
his
job
properly
well
thing
to
do
that.
I
call
this
the
holy
check.
If
you
run
to
renew
this
in
production,
you
should
be
made.
Taking
this
number
of
instances
running
should
be
at
least
the
ones
you
desire
and
if
it's
less
she's
be
only
for
a
given
amount
of
time.
A
If
you
are
you
doing
a
running
deployment
or
something
like
that,
then
the
other
thing,
the
four
point
to
monitor
its
kubernetes
internals,
how
we
can
make
sure
kubernetes
is
working
with.
There
are
a
few
internal
services
like
ABCD
cube,
let
doc
docker
uncover
needs
api.
We
need
to
make
sure
those
are
working,
but
something
we
also
need
to
look
at.
Do
we
have
enough
in
physical
nodes
or
vertical
Uster?
If
we
are
overlooking
our
nose
things,
they
will
start
kind
of
collapsing.
Finally,
the
fifth
point:
it's
going
to
be
host
notes.
A
This
was
something
it
thought
it
was
really
important
before
when
services
they
were
a
couple
in
to
host
or
notes,
and
probably
if
you
were
on
call,
this
was
the
kind
of
alert
that
I
was
waking
you
up
nowadays
with
kubernetes.
It
is
scheduling
things
moving
things
around.
If
one
house
goes
on,
it's
probably
nothing
so
probably
the
most
important
alert
that
you
can
set
up
here
is
okay,
Hardy's,
archit
and
Phil,
or
not
so
these
are
replays.
Host
monitoring
has
been
replaced,
but
an
upper
layer
which
is
the
kubernetes
monetary.
A
So
if,
when
you
walk
out
from
this
five
minutes,
talk
very
quick
thing,
I
want
you
to
remember,
monitor
your
application.
Monitor
your
services
monitor,
cover
Nittis
is
managing
your
applications,
cover
Nettie's,
internal
and
notes.
What
yes
a
little
bit
thanks,
very
much
I
hope
you
learned
something
and
if
you
have
any
question
or
anything
like
that,
walk
by
our
booth
and
we'll
be
very
happy
to
discuss,
how
are
you
doing
with
your
kubernetes
monitoring?
Thank
you.