►
Description
Case Study: OpenShift at China Mobile
OpenShift Commons Gathering
Barcelona Spain 2019
Speaker: Xian Jia, Project Manager, Edge Cloud Computing Architect (China Mobile)
Title: OpenShift and Edge Computing at China Mobile
A
A
I'm
I'm
from
China
Mobile,
my
name
is
the
chuan
is
very
pronounced.
Hardy
I'm,
the
edge
cloud
computing
architecture
for
this
presentation.
I
will
to
tell
the
story
about
the
edge
computing
with
OpenShift.
So
so
I
want
to
tell
the
background.
So
what
is
H
computing?
The
is
computing
is
a
computing
and
network
deep
fusion.
So
I
draw
a
5,
a
4
pictures
here,
so
we
can
see
which
is
the
H
computing.
So
we
can
check
the
red
star.
A
It's
for
the
Aged,
the
environment.
A
is
very
little.
Its
didn't
like
the
data
center,
which
is,
can
concern
1000
per
machine
in
the
H
in
H
H
computing
center.
It
can
only
come
only
have
a
10
or
20
parameter
machines,
so
the
resource
is
limited
and
I
think.
This
is
why
we
want
to
use
a
darker,
a
container
technologies
in
this
field,
because
in
the
telecom
field
we
almost
used
OpenStack,
but
now
I
think
this
is
the
community's
opportunities.
A
So
in
the
age
a
we
celebrate,
we
Sturbridge
2
part
when
the
networks
network
side,
each
computing,
which
is
below
the
awareness
station
and
for
the
other
side,
is
computing
instant.
Like
the
kid
way
we
can
put
it
in
in
the
home
or
in
the
industry,
so
for
the
in
the
age
computing
we
can
provide
the
the
eyes
the
past
and
the
sass
in
telecom
field.
We
are
always
followed
the
ETSI
or
some
others
international
standards,
but
and
now
I
poured
into
this
the
this
architecture
from
ETS
that
we
see.
So
it's
very
calm.
A
A
We
want
to
use
these
technologies
in
China
and
in
this
in
his
right
eye,
a
laser
some,
the
pros
and
cons
so
for
the
pros
it
is
based
on
the
NIV
architecture
and
it
have
defined
the
network
infrastructure
in
information
and
the
location,
API,
new
identity,
API
and
bandwidth
management.
So
maybe
all
the
people
in
here
I
don't
know
what
Hayes
is
so
I
can
give
us
an
example.
So
when
we
open
our
phone,
so
you
will
see
there
is
a
4G
at
the
at
the
top
of
the
our
screen.
A
So
all
these
only
a
movie
clip
at
the
Chrome
application.
So
we
can
search
the
e
tonight.
So
already
in
title
flow
is
in
the
telecom
cloud.
We
call
it
it's
some
like
the
private
cloud.
The
oldest
data
flow
is
the
enterprise
private
cloud,
but
in
the
age
computing,
who
is
our
customers
in
China?
Maybe
the
by
to
aunt,
isn't
Alibaba
and
some
industry
company.
A
A
It
sound
like
a
today
defined
like
the
private
cloud,
so
it's
not
satisfy
our
requirement
so
I
gave
the
conclusion
here
is
the
I
mean
the
average
and
IV
architecture
is
to
complicate,
and
there
is
a
far
away
from
our
product,
but
we
can
use.
The
API
which
is
have
defined
in
the
idiom
is
e,
so
in
the
LAV
field
and
we
choosing
in
the
telecom
companies
it's
all
we
all
we
all
used
OpenStack,
but
in
its
computing
we
want
to
use
coupon
IDs.
So
I
try
to
refactor
the
architecture
from
the
ETS
imec.
A
A
There
is
the
H
computing
operation
managed
center
is
some
like
the
photo
and
can
manage
a
lot
of
the
did
each
computing
a
cluster
and
again
provides
the
beam
the
container
and
the
stamp
third
party
API
unfold
nmv.
It
only
put
fears
atrophic
into
a
in
into
the
age
computing
data
center.
So
we
so
there's
two
part
is
ace.
A
Is
a
beaut
it's
billed
separately,
so
why
we
want
choose
open
shift?
I
have
lists
to
the
four
regions
here.
The
overshift
is
on
the
most
major
of
installed
project
with
support
communities
nearly
before
I
joining
the
China
Mobile
I'm
I'm.
Worse
I
worked
in
your
head
of
four
years
ago,
so
Amber's
a
red
Hatter,
so
I
know
it
is
the
very
stable
and
now
we
start
to
develop
a
project
based
on
the
kubernetes.
So
we
choose
the
overshift
as
our
basically
kubernetes
an
oval-shaped
and
provides
at
least
ten
layers
of
the
continuous
security.
A
A
So
they
can
so
they
can
use
use
a
few
effort
to
move
their
application
into
our
platform,
so
we
can
build
up.
Who
can
build
up
our
own?
It
computing
data
center
and
open
our
urban
ships
always
brings
amazing
features
from
open
source
communities,
so
seems
a
lot
of
the
companies
have
using
the
urban
shift
of
four
four
or
four
pound.
Oh,
we
are,
but
now
we
are
using
the
communities
three
pumps
at
the
other,
so
we
have
to
catch
up
with
other
guys
so
Sigma
8h
computing
platform
in
China
bebel.
A
A
She
may
be
it.
We
can
think
it
is
in
the
in
the
center
of
the
area,
for
example
in
abating,
and
they
can
manage
a
lot
of
the
problems
in
China
and
for
the
green
part,
is
a
is
our
age
I'm
huge
in
the
center,
so
it
can
run
on
tin,
beeramid,
machine
or
2020
pyramid
machines,
so
we
developed
some
additional
modules,
for
example,
is
the
classroom
management
and
service
routine
model?
A
So
it's
not
like
the
openshift
routine,
because
it's
not
satisfying
our
requirement,
so
we
developed
it
by
ourselves
and
undoing
here's,
the
third
pratik
capabilities,
industry,
sdk
awareness
and
network
capabilities.
This
is
our
from
the
industry,
so
it's
very
company,
it's
very
complex.
We
do
it
back
his
keys
back
ease
and
we
are
using.
We
are
trying
to
use
three
skill
to
manage
our
api,
because
in
some
industry
we
have
to
provide
api
and
then
we
can,
we
can
serve
the
api
and
and
and
on
the
money
you
can
give
the
earth
the
billing
unfolded.
A
A
A
So
we
talked
with
the
Baidu
company
who
has
joined
their
own
Otto
Michaels
in
Beijing,
and
they
provide
to
the
requirement
and
the
we
build.
The
arch
actors.
So
at
top
of
this
is
a
beta
cloud.
It
sounded
obvious,
but
I
do
has
their
has
their
own
public
cloud
and
there
is
a
Sigma
Sigma
is
I
is
just
totally
at
Otis
to
talk
about
it.
It's
our
age,
computing
platform
and
it's
the
bottom
and
at
the
bottom,
this
this
program
we
can
see
there
is
a
5g
network
and
some
the
aware
network.
A
So
the
procedure,
the
workflow
is
a
video
stream
is
saved
in
the
H
node
and
Sigma,
and
the
AI,
which
is
the
application
running
in
the
Sigma
and
the
AI,
operates
the
video
stream
in
each
node
and
then
sent
to
the
read
out
to
the
to
the
ICU
I
see
we
can
record,
we
can
go
away
a
regard.
We
can
think
a
T.
Is
the
machine
in
the
inner
car
understand
the
uncertain
pack,
the
data
into
the
PI
to
cloud?
A
This
is
public
cloud
and
I
see
you
I
still
send
it
out
to
the
car
as
the
PC
vibe
interface.
So
this
is
the
the
architecture
we
test
weight
has
in
baking
in
Beijing,
so
one
we
choose
two
ways:
one
is
the
Fuji
real
news
network.
Another
one
is
the
wire
net.
A
wire
network
is
some
like
the
way
provides
the
VPN
to
two
by
two,
because
we
want
to
to
to
balance
if
the
Fuji,
the
price,
the
price
is.
A
A
A
So
the
requirement
is
the
banner.
Is
there?
Is
a
6th
Mbps,
a
blue
stream
for
camera
latency,
so
why
we
want
to
use
the
edge
computing
because
it
can
provide
low
latency.
The
latency
of
the
data
flow
from
camera
to
age
computing
server
is
the
less
than
100
millisecond
the
latency,
with
edge
computing
a
server
give
the
result.
The
Houston
to
the
response
is
the
less
than
50
a
network.
Latency
is
less
than
2
miles
purely
understood
to
disinterest.
A
A
A
So
in
this
experiment,
environment
we
choose
to
Street
Internet
intersection
and
employ
to
totally
26
camera
there's
a
lot.
They
are
interesting
in
Barcelona,
I
didn't
see
any
camera
in
the
street,
but
in
China
just
lots
of
the
cameras,
so
26
camera
in
to
Street
intersection,
and
it
calls
a
lot
all
the
data.
A
The
Cameron
Nico
needs
the
electron-electron
the
power
because
when
because
they
had
our
the
the
machine,
teenies
need
their
power,
but
when
we,
when
we
set
up
the
camera
in
the
light,
London
single
light,
who
provides
the
the
leg
electronic
power
so
how
to
talk
to
with
the
garment
and
the
government,
provides
the
the
Leghorn
Ingo
power.
But
it
will
cost
a
lot
of
money.
A
A
A
So
we
have
to
use
some
other
technologies
to
to
reduce
the
data
in
each
computing,
so
why
we
choose
to
send
a
video
stream
instead
to
image.
So
we
we
talked
with
bed
companies
and
they
said
that
the
if
we
use
the
video
we
have,
we
can
use
some
some
model
modern,
some
new
technologies
to
truly
to
to
reduce
the
the
data
floor,
but
but
yeah,
but
one
second,
a
most
1mb.
A
So
it's
so
lot
for
lunch
ham,
so
we
will
sink
it
later
and
the
talk
wizard
about
to
if
they
can
use
the
I
use
image
instead
of
another,
is
a
GPU
in
because
in
the
age
computing
the
data
center,
the
the
machine
is
not
like,
as
not
like
in
at
the
machine
in
the
in
the
data
center.
So
it's
it's
more,
it's
more.
A
We
have
we
have.
We
have
ensured
that
each
server
needs
at
least
three
GPUs,
so
stretchy
peels
is
okay,
so
the
machine
can
have
the
three
PCIe.
So
you
can
put
the
three
GPU,
but
the
GPU
cost
a
lot
of
electron
in
power,
so
we
will
need
to
reboot
our
eight
computing
center,
because
the
edge
computing
Center
is
not
like
the
the
the
the
computing
Center,
because
that,
because
the
edge
computers
in
term
maybe
is
near
our
building,
our
home
near
the
our
industry
and
the
electronic
is
limited.
A
For
example,
if
in
China
there
is
a
old
building
and
can
only
support
three
key
W
I,
don't
know
how
to
pronounce
it.
Okay,
three
key
W,
but
if
we
put
a
lot
of
GPU
they
did
the
datacenter
let
each
the
center
can
support
it.
So
we
have
to
reduce
the
resource
we
have
used
in
H
computing,
but
in
the
to
the
center
we
don't
match
these
problems.
We
have
lot
of
space,
it
2
2,
2
2.
A
We
have
a
lot
of
space
to
put
our
pyramid
machine
in
here
and
they
were
doing
to
worry
about
the
electronic
power,
but
in
it
computing
we
have
choose
thinking
things
like
how
to
reduce
resource
the
container
technology
is
a
better
one.
So
what
we
want
to
use
it
open
shift
here
and
to
be
continuous,
that
we
are
doing
a
lot
of
the
things
in
the
future.
So
now
we
collaborate
collaborative
ways
that
tritium
companies.
They
were
two
prior,
the
API
and
the
China
state
construction.
A
We
only
have
three
pyramids
machine,
but
there
we
built
the
upper
shift
here
and
enabled
Robert
to
provide
the
wash
machine
and
give
the
the
virtual
machine
environment,
so
they
can
put
their
application.
Monument
and
other
is
ID
by
two
companies
and
the
blockchain
and
the
in
the
next
month
will
deployed
all
the
things
in
Judea.
It
is
a
4G
amidst
the
environment
and
in
the
the
later
we
will
he
was
the
overshift
boy.