►
From YouTube: Operator Framework Moves to CNCF for Incubation
Description
Operators are a design pattern made public in 2016 to put operational knowledge in software to communicate and integrate with Kubernetes concepts and APIs. The Operator Framework is an open source toolkit to manage Kubernetes native applications in an automated and scalable way. The Operator Framework takes knowledge previously trapped in the minds of administrators, various shell scripts, or external automation software and transforms them into automated Day-1 and Day-2 activities. Learn of the two main components making up the Operator Framework and what’s next for this new CNCF Incubated project donated by Red Hat.
A
Hi
everyone,
I'm
rob
sumski
from
coreos
and
now
red
hat
and
I'm
thrilled
to
announce
that
the
operator
framework
is
joining
the
cncf
as
an
incubating
project.
Coreos
introduced
the
operator
concept
in
2016,
followed
shortly
thereafter
by
the
operator
framework,
which
is
a
set
of
tools
to
help
you
build
ship
and
discover
operators.
A
Operators
are
operational
procedures
and
best
practices
that
are
codified
into
software.
They
make
automated
day
two
operations
possible
on
kubernetes
and
model
the
complexities
of
today's
distributed
systems.
For
example,
there
isn't
a
concept
of
data
rebalancing
inside
of
kubernetes,
but
that
can
easily
be
built
on
top
of
the
kubernetes
apis.
With
the
operator
framework,
these
types
of
concepts
are
required
for
running
what
we
call
the
third
wave
of
applications
on
kubernetes
that
are
these
really
complex
distributed
systems.
A
These
tools
tie
into
other
cncf
projects
like
coupe
builder,
helm,
cuddle
and
popular
open
source
like
ansible.
The
framework
is
loosely
coupled.
So
if
you
have
a
testing
tool
that
you
like,
for
example,
you
can
keep
using
that
and
if
you
want
to
build
your
operator
outside
of
the
sdk,
that's
totally
fine,
you
can
still
run
it
with
the
lifecycle
manager.
A
The
operator
functions
correctly
when
it's,
you
know,
manipulating
these
very
complex
applications
and
our
experts
in
the
community
have
modeled
all
types
of
applications
as
operators.
We
can
help
you
save
time
and
avoid
some
bugs
plus
it's
just
been
really
fun
to
see
all
the
projects
come
through.
Our
community
call
and
help
people
out
now,
once
you've
written
your
operator,
you
need
to
hand
it
off
to
your
colleagues
or
your
customers
to
actually
run
it.
A
This
is
where
the
lifecycle
manager
comes
into
play.
There
are
actually
a
lot
of
tricky
problems
here
that
you
might
not
even
know
you
have
yet
like
collision
detection
for
crds
picture.
You
have
a
crd,
that's
managing
a
database,
but
then
there's
another
operator
that
wants
to
also
manage
that
database.
No.
B
A
Same
thing
with
like
a
machine
learning
application,
the
lifecycle
of
the
crd
itself
is
also
important,
though
just
going
to
their
v1
status
in
kubernetes,
and
so
the
operator
framework
can
help.
You
update
your
operator
and
manage
your
crd
as
part
of
that
process.
There's
too
much
coming
up
to
cover
it
all,
but
I
want
to
give
you
a
sneak
peek
at
where
we're
going
in
2020.
A
First
we're
moving
over
to
a
new
bundle
format.
This
is
how
the
operators
are
collected
into
catalogs
and
shared
between
clusters.
We
have
a
new
tool
opm,
which
will
allow
cluster
admins
to
curate,
that
set
of
operators,
including
for
disconnected
clusters
and
last
a
new
operator.
Api
is
going
to
land,
which
is
a
newer,
simpler
way
to
interact
with
operators
on
your
cluster
for
both
admins
and
developers.
B
If
updating
and
maintaining
containerized
applications
requires
manual
processes
and
regular
human
intervention,
then
it
makes
managing
applications
at
scale
difficult
worse.
It
leaves
those
systems
vulnerable
if
they
are
not
properly
configured
or
up
to
date.
Kubernetes
operators
change
this
an
operator
extends
kubernetes
to
streamline
and
automate
installation
updates
and
management
of
container-based
services.
B
It
does
this
through
a
control
loop
that
continually
checks
your
cluster
to
ensure
your
described
ideal
state
with
the
right
number
of
instances
on
the
right
infrastructure
in
the
right
places,
with
the
right
deployment,
rollback
and
failover
strategies,
kubernetes
automates.
Most
of
these
tasks
for
stateless
applications,
but
most
software
isn't
stateless.
B
An
operator
is
a
scalable
custom
version
of
this
control
loop
for
stateful
applications.
It
determines
if
your
application
is
running
correctly
according
to
your
instructions
and
in
line
with
best
practices
from
the
experts
in
running
that
app.
If
it's
not
the
operator
automatically
acts
to
correct
it,
operators
also,
let
users
interact
with
a
complex
application
as
a
single
object
on
the
cluster.
Thanks
to
custom
resource
definitions
or
crds
users
interact
with
the
operator
through
those
objects
which
expose
only
the
options
that
make
sense
for
the
application.