►
From YouTube: Level Setting the Cloud Native Edge: Definitions and Core Principles - Keith Basil, Rancher
Description
Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon North America 2021 in Los Angeles, CA from October 12-15. 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.
Level Setting the Cloud Native Edge: Definitions and Core Principles - Keith Basil, Rancher
Everyone has a different definition of the “edge”. Collectively we’ve found it useful to establish a baseline definition of the edge so that meaningful discussions can ensue. In this talk we will review an example framework that has evolved over time during the natural interactions with engineering, partners and those doing real-world cloud-native edge deployments. Borrowing from the aforementioned framework, the session will also succinctly highlight core principals and attributes that are present in the majority of the Kubernetes edge use cases we see today. Attendees of this session will get an introduction to the edge challenges in the areas of application, cluster and operating system management at edge scale.
A
Hello
folks
welcome
to
kubernetes
on
the
edge
day.
My
name
is
keith
basil
and
I'm
the
vice
president
of
product
for
cloud
native
infrastructure
at
sousa.
I
also
helped
shape
souza's
global
edge
strategy
and
I'm
here
to
talk
to
you
about
edge
today,
I'd
like
to
thank
kubecon
and
the
cncf
organizers
for
this
opportunity
to
present
and
with
that,
let's
get
started.
A
So
I
wanted
to
build
on
the
quote
that
was
given
by
the
linux
foundation,
and
here
it
says
the
edge
computing
space
will
be
four
times
larger
than
cloud
and
will
generate
75
of
the
data
worldwide
by
2025..
That's
a
very
bold
statement
and
with
that
statement
it
gives
us
quite
a
bit
of
runway
to
work
into
that
space
as
we
move
forward,
so
walk
with
me.
A
If
you
will,
if
you
can
imagine
a
global
deployment
of
7
500
remote
locations
and
within
those
locations,
you've
got
a
thousand
industrial
iot
devices
at
each
one
of
those
locations
right.
So
when
you
do
the
math
you're,
looking
at
7.5
million
things
that
need
to
be
orchestrated
and
managed
at
the
edge,
and
so
what
I
want.
A
Here
we
have
the
facilities
to
tackle
this
challenge
and
before
we
dive
into
that,
let's
establish
a
framework
for
defining
what
the
edge
is,
because
everybody
has
a
different,
a
different
definition
of
edge
right
and
so
collectively.
We
found
it
very
useful
to
establish
this
baseline
definition
of
edge
so
that
we
can
have
meaningful
and
relevant
discussions
going
forward.
A
And
so
the
first
thing
I
want
to
to
walk
through
here
is
what
we
call
the
narrative.
So
just
walk
with
me
for
a
second
off
to
the
left-hand
side
of
this
screen
is
the
centralized
services
large
data,
centers
and
the
like.
So
that's
where
all
the
centralized
services
are
today
and
so,
as
you
move
from
left
to
right,
you
get
closer
and
closer
to
the
edge
and
first
thing.
A
A
That's
within
that
space,
and
here
again,
as
I
said,
this
is
the
realm
of
the
telcos
right,
and
the
last
thing
I
want
to
call
out
is
that
that
line
of
demarcation
is
critical
because
there
are
some
edge
solutions
or
applications
where
the
communications
providers
are
providing
appliances
that
go
up
to
and
sit
on
that
line
of
demarcation.
And
if
we
go
back
to
that,
ipspace
definition,
the
ip
space.
That's
attached
that
to
attach
to
that
device
is
managed
by
the
communication
service
provider.
A
In
fact,
that
gear
is
typically
owned
by
them
and
they
offer
services
to
the
end
customers
on
the
other
side
of
that
dmarc.
So
that's
a
really
critical
thing
in
terms
of
ownership
to
help
us
define
what
that
near
edge
scenario
is
so
let
me
move
to
the
next
portion,
which
would
be
the
far
edge,
and
so
again
we
move
to
the
on-premises
side.
A
This
is
the
remote
location,
and
this
is
where
things
get
really
interesting,
and
so
again
the
border
there
is
meant
to
represent,
let's
say,
a
layer,
two
domain
from
a
networking
perspective.
This
is
customer
owned
and
managed
ipspace.
This
is
customer
owned
and
managed
infrastructure
in
the
form
of
like
hardware
that
supports
your
kubernetes
clusters
right
and
so
we've
got
boxes
there.
A
If
you
see
visually
represented,
we've
got
boxes
of
various
sizes
to
represent
multiple
cluster
sizes,
so
in
some,
for
example,
manufacturing
use
cases
they've
carved
out
a
portion
of
the
data
center
to
sorry
a
portion
of
the
factory
to
act
as
a
small
data
center
and
they've
got.
You
know
classic
you
know,
2u
machines
racked
up
and
they
it
like
a
data
center
and
those
are
very
large
clusters.
We
also
have
locations
that
have
a
single
node,
a
single
node
cluster,
to
serve
their
uses
as
well.
A
That's
actually
pushed
to
that
location
where
it
can
do
the
best
good,
and
many
of
these
use
cases
have
the
local
kubernetes
clusters
within
the
premises
to
actually
aggregate
data
from
the
iot
devices
and
sensors
and
such
and
that's
actually
a
great
segue
into
the
third
segment
of
the
edge.
So
so
far,
we've
talked
about
near
edge,
we
just
covered
far
edge
and
finally,
we've
got
what
we
call
tiny
edge
and
I
absolutely
love
this
name.
A
I
heard
this
naming
convention
at
our
edge
conference
in
the
fall
given
by
folks
from
microsoft,
and
this
is
really
where
the
law
of
large
numbers
kicks
in.
So
this
space
is
early
and
we
are
encouraged
by
the
yeoman's
work
being
done
by
kate,
goldenring
and
edward
wong
in
this
space.
Both
of
those
are
microsoft.
A
We
want
to
be
involved
in
that
that
community
as
well,
and
we
would
encourage
you
to
take
a
look
at
that
and
I
believe,
later
in
the
day,
there's
or
at
the
session,
there's
a
talk
by
edric
that
speaks
about
ocry
and
so
now
that
we
have
a
definition,
a
working
definition
of
the
age.
Let's
actually
talk
about
the
three
pillars
that
make
up
the
solution
for
managing
kubernetes
at
scale.
A
So
there's
many
options
there,
where
you
have
a
lightweight
operating
system,
that's
container
native
or
container
native
friendly,
there's,
probably
a
better
way
to
say
that,
and
we
believe
this
is
required
to
provide
a
low
attack
footprint
from
a
security
perspective
and
also,
more
importantly,
to
allow
us
to
manage
the
full
life
cycle
of
that
operating
system.
Because
when
you're,
managing,
let's
say
707,
500
or
anything,
you
know,
managing
the
cycle
of
the
operating
system
should
be
done
in
the
kubernetes
way
right
and
then.
A
A
And
so,
when
you
look
at
all
three
of
these
pillars,
we
think
this
is
the
the
minimum
set
that
you're
going
to
need
to
have
an
effective
kubernetes
at
scale
management
solution,
and
so
again,
kubernetes
management
at
scale
with
get
ops
is
great,
a
lightweight
distro
such
as
k3s
or
any
other
related
variants,
and
then
a
lightweight
operating
system.
That's
focused
on
the
cloud
native
front
on
the
cloud
native
space,
and
so
when
you
look
at
it
all
together,
you
have
that
solution
at
the
bottom.
A
It
will
allow
you
to
attack
and
manage
all
three
segments
of
the
edge,
so
we've
got
the
near
edge
and
the
telco
space.
That's
more
of
a
classic
data
center
play
with
you
know.
You
know
small
data
centers
or
regional
data
centers.
We
can
rack
and
stack
at
that
point.
That's
something
that
we
know
very
well.
The
things
get
really
interesting
when
you
come
over
to
the
far
edge
where
you
start
deploying
a
lightweight
version
of
kubernetes,
a
lightweight
operating
system
and
then
managing
that
with
a
declarative,
get
off
source
of
truth
behind.
A
You
know,
let's
say
10
000
deployments
and
then
the
tiny
edge
is
emerging,
the
acry
project.
We
like
it
a
lot
inside
rancher
and
we
want
to
again
contribute
to
that
community.
But
it's
also
very
new
right.
So
there's
a
handful
of
protocols
that
we
support
on
the
industrial
iot
space
and
we
want
to
make
sure
we
can
mature
the
capabilities
or
the
adoption
of
those
protocols
that
are
heavily
used
in
the
in
the
industrial
iot
space.
And
so
this
is
the
complete
solution.
A
We'd
love
to
see
the
definition
of
the
framework
that
I
discussed
adopted
as
we
believe
that
it's
going
to
be
very
meaningful
to
have
discussions
that
are
efficient
in
guiding
us
to
the
solutions
that
work
and
overall,
I
think
we
should
strive
to
remove
the
complexity,
that's
inherent
in
our
systems,
right
and
so
with
that.
I
just
want
to
thank
you
for
this
opportunity
and
we
hope
to
see
you
face
to
face
at
the
next
coup
con
going
forward.
So
thank
you
again.