►
Description
Taking OpenShift to the Edge with MicroShift
Henry Geay De Montenon (Red Hat)
This OpenShift Commons Gathering was held on July 6th, 2022 live in London, England
https://commons.openshift.org
A
So
how
is
it
different
from
openshift
already
being
into
the
edge?
Because
we
already
do
that
with
openshift?
In
fact,
the
aim
of
openshift
is
delivering
application
everywhere.
What
we
want
is
provide
consistency
for
itups
and
developers
alike.
We
want
to
provide
flexibility
so
that
you
have
architectures
to
run
everywhere
in
the
in
the
cloud
and
the
edge,
and
we
want
to
provide
you
with
choice
to
run
any
kind
of
kubernetes
application
there.
A
A
So
those
all
cutter,
four
different
use
cases
into
the
edge
single
mage
are,
for
example,
for
low
bandwidth,
size
and
very
low
footprint,
whereas
work
remote
working
nodes
are
best
for
space
constrained
environments
and
obviously
3.
cluster
will
provide
you
full.
So
we
have
all
those
edge
tiers
and
up
to
the
edge
server
chair.
A
A
So
right
now
we
have
red
enterprise,
linux
4h,
which
does
offer
a
first
solution.
So
what
it
gives
you
for
those
edge
devices
is
quick
image
generation.
You
can
basically
build
your
own
rail
image
with
whatever
software
you
want
to
embed
there.
Whatever
drivers
you
want
to
embed
there,
you
have
the
edge
management
we
have
for
now
an
online
edge
management
facility
that
allows
you
to
build
those
edge
images
to
manage
them.
A
You
can
manage
the
servers
you've
built
with
that.
You
have
efficient
over
the
air
updates,
but
I'll
give
you
more
on
that
just
in
a
bit,
so
this
is
brought
by
rpm.
Os3
risted
brought
up
this
a
bit
this
morning
talking
about
red
core
os.
This
is
something
we
actually
introduced
a
while
back
when
we
created
red
atomic
host,
and
the
aim
of
that
is
basically
to
have
an
immutable
os
image
where
all
your
os
binaries
are
stored
in
an
immutable
file
system
that
you
can't
modify.
A
So
basically,
when
you
do
updates,
you
update
everything
at
once,
so
that
allows
also
to
have
state
files
managed
outside
of
this
immutable
file
system,
because
it's
not
core
os.
This
is
rel.
We
want
you
to
be
able
to
do
whatever
you
want
to
with
it
and
therefore
to
be
able
to
manage
your
own
configuration
files,
there's
no
in-between
stage
during
updates.
As
I
said,
you
update
everything
at
once.
Then
you
restart
the
server
and
it
comes
back
all
updated
there.
Also.
A
This
also
allows
for
rollbacks
efficient
rollbacks
and
automated
rollbacks
as
well
as
in
so
we
have
this
system
integrated
into
rel
for
edge.
That's
green
boot
that
will
basically
detect
any
issue
during
the
update
and
if
something
wrong
happen,
it
will
roll
back
the
update
and
start
again
fresh,
and
this
also
allows
for
basically
rail
eight
to
round
nine
updates.
Major
distribution
updates
to
happen
without
having
to
worry
about
yeah
things
are
not
gonna
work
as
they
should
or
whatever,
because
you
still
can
order
back.
A
A
Where
you
will,
you
are
able
to
send
to
your
hardware
manufacturer
a
usb
key
with
your
operating
system,
and
then
they
can
deploy
that
in
the
manufacturer
they
ship
the
devices
and
then
once
the
device
are
plugged,
they
will
go
and
authenticate
to
your
server
and
fetch
their
configuration,
and
that's
it.
Zero
touch
needed.
A
A
A
Single
node,
open
shift,
that's
four
core
16
gigs
and
doesn't
need
anything
else
and
all
the
way
to
srinal
cluster
with
six
core
24
gigs
per
server.
But
what
about?
Having
openshift
on
the
smaller
edge
devices?
I
mean
we're
all
about
telling
openshift
gives
you
this
ease
of
management.
This
is
of
managing
your
workloads
across
all
your
footprints.
A
You
can
see
all
their
current
projects
at
next.redder.com.
They
everything
they
do
is
upstream.
It's
all
community
based
either
we
do
participate
on
other
upstream
projects
either
we
have
our
own
projects
that
we
do
set
on
our
public
github
for
imaging
text
they're
not
always
followed
by
products.
So
everything
you
will
see
on
next.rtrade.com
doesn't
mean
it's
going
to
become
a
reddit
product,
a
lot
of
those
actually
don't,
but
sometimes
some
do
and
they
have
multiple
areas
of
focus.
So
there
are
chances
they
are
working
on
stuff.
That
would
interest
you.
A
One
of
their
latest
projects
is
microshift.
So
what
is
microshift
microshift
is
basically
an
experimental
flavor
of
open
shift.
That's
meant
for
those
small
edge
devices.
What
it
does
is
provide
a
complete
kubernetes
environment,
it's
not
just
podman
and
rail,
which
is
what
you
would
do
today
to
have
containers
running
on
these
edge
devices.
A
A
It
also
aims,
obviously,
for
a
very
low
resource
footprint.
As
I
said,
we're
talking
about
non-server
components
there
very
small
devices
like
this
one,
there
nvidia
jetson,
so
the
minimal
requirements
currently
for
microshift
are
two
cores
and
two
gigs
of
ram,
and
actually
the
core
requirements
are
even
lower
than
that,
but
we
want
to
have
some
room
for
the
os.
A
It
supports
many
architectures
right
now
we
have
built
for
arm
and
x86.
It
can
also
easily
be
built
for
risk
or
whatever
other
architecture
it
got
low
over
the
wire
payload
downloaded.
It
is
not
gonna
cost.
You
lots.
If
you
have
like
4g
network
on
the
edge,
for
example,
low
storage
requirements,
you
don't
need
to
have
huge
disks
attached.
A
So
let's
look
a
bit
about
what
differentiates
microshift
from
openshift
so
openshift.
Basically,
as
it
is
running
on
x86
servers
with
bmcs
or
cloud
platforms,
basically
it
needs
to
be
able
to
control
the
underlying
operating
system,
the
underlying
hardware.
So
we
have
all
those
cluster
operators
which
are
there
to
provide
this
facility
of
managing
either
the
vmware
environment.
Openshift
is
running
on
the
azure
aws,
whatever
you
name
it,
this
is
not
really
needed
for
microshift,
because
we
don't
need
to
manage
the
underlying
environment.
Actually,
we
don't
even
want
to
manage
it.
A
We
want
you
to
be
in
charge
of
the
underlying
operating
system
because
some
of
our
customers
in
the
edge
space,
actually
they
they
need
to
be
able
to
customize
their
operating
system
to
a
larger
extent
than
what
we
can
offer
with
eric
coreos,
because
we
have
to
manage
that
from
openshift.
So
we
are
limiting
the
options
there
so
there
the
idea
is,
you
are
completely
unlimited
in
how
you
manage
your
os,
your
infrastructure.
A
A
What
we
did
as
well
to
lower
the
footprint
of
microshift
is
build
it
as
a
monolithic
exit,
basically
binary.
So,
basically,
you
can
have
it
either
as
a
binary
that
can
be
embedded
in
a
package,
especially
if
you
use
rpms3
or
you
can
have
it
as
a
container
that
you
would
just
run
using
padma,
and
this
gives
you
also
the
kind
of
flexibility
and
yeah
low
over
the
wire
payload
that
we
were
looking
for.
A
They
can
have
open
cluster
management
to
manage
all
the
apps,
for
example,
or
you
can
ship
everything
like
in
a
usb
device
for
devices
that
would
be
completely
disconnected
and
then
have
all
your
application,
microshift,
etc.
Everything
in
the
os
image
on
a
usb
stick,
for
example,
and
then
you
can
ship
another
osb
stick
for
the
updates.