►
From YouTube: Serverless Computing for E-Commerce Applications on JD Kubernetes Platform - Yuan Chen
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
Serverless Computing for E-Commerce Applications on JD Kubernetes Platform - Yuan Chen & Xin Tong, JD.com
Fibonacci is an enterprise level serverless computing service on JD.com’s e-commerce platform. It is built on top of OpenFaaS and includes significant enhancements and new features that make it suitable for large scale production use. Specifically, Fibonacci provides seamless integration into JD’s internal container cloud platform, the world’s largest Kubernetes production cluster. The new features include FaaS-optimized elastic scheduling, web control panel and GPU support. Fibonacci has been powering applications on JD Mall, such as intelligent product image processing, automated IT operations and chatbot, greatly simplifying the application development and deployment, and improving resource efficiency. We will present Fibonacci’s design and implementation, share our experience and lessons in developing and applying serverless computing to large scale e-commerce platforms.
To learn more click here: https://sched.co/FuLt
A
B
Afternoon,
everyone
that
will
fund
the
new
wager
not
all
should
charge
on
one
night
in
here
with
the
park
I
will
join
one,
that's
easy
robot,
that
is
my
bunny
doll.
Okay,
whoa
whoa,
hey
hyung
gosh!
You
weigh
like
a
20
good
afternoon.
Everyone.
It
is
my
pleasure
to
be
here
with
my
colleague,
home
sing,
well
from
JD
dot-com,
and
it's
my
honor
to
be
here
to
share
about
our
experience
in
Subway's.
B
I
think
most
of
our
peers
are
aware
of
JD
dot-com
I
want
to
emphasize
that
we
are
very
large
company
and
we
have
very
rich
application
scenarios
and
we
have
a
very
large
customer
base
in
addition
to
traditional
e-commerce
business.
We
also
have
business
in
the
logistics
and
the
financial
business
and
we
have
a
it
is
also
in
an
IT
platform,
and
we
have
cooperation
with
large
internet
companies.
We
have
the
strategic
partners
with
these
companies
like
Google,
a
Walmart
and
Tencent.
There
are
large
partners
and
I
want
to
introduce
our
team
I.
B
Don't
know
how
many
of
you
attempted
a
keynote
session
this
morning
and
Joe
did
account
just
1
and
user
award
and
I
think
it
is
a
recognition
from
the
community
I
think
from
the
JD
that
comes
containerized
platform
and
also
I.
Think
JD
is
one
of
the
earliest
company
using
this
technology
and
a
2016
I
think
Google
started
to
make
the
kubernetes
open
source
in
2015.
C
B
Many
of
our
systems
are
based
on
those
source
and
in
our
practice
we
also
contribute
to
many
use
cases
and
also
feature,
and
so
we
have
made
the
contributions
to
the
to
the
Kuban
a
place
and
also
the
system
and
I
hope
that
in
future
we
have
more
exploration,
and
we
would
like
to
share
our
experiences
and
use
them
to
the
community,
and
the
most
important
is
that
we
can
have
the
contribution
in
technology.
I
think
the
surplice
is
also
a
new
era
and
the
way
exploring
how
to
make
the
most
advantage
of
it.
B
B
This
is
a
platform
for
the
infrastructures
service
and
we
also
have
gos,
and
this
is
optimized
communities
of
the
jetty,
and
we
also
have
the
manager
system
and
also
the
middleware,
RPC
and
caching
and
other
database.
All
of
this
are
running
on
our
container
platform.
In
addition
to
the
online
business,
we
also
have
the
AI
and
big
data
analysis
and
the
recently
we
also
have
an
alpha
line
in
stores,
and
we
also
use
this
kind
of
technologies
for
our
offline
stores.
B
As
I
mentioned,
we
apply
these
technologies
for
our
businesses
in
GE,
including
the
international
business
logging,
talent
or
the
Indonesian
and
other
countries,
and
now
I
have
introduced
a
lot
of
background.
So
I
hope
that
you
will
know
the
current
context
about
a
JD
that
is
JD
has
a
very
large
containerized
the
platform
and
in
this
condition,
in
this
context,
I.
C
B
Know
do
you
have
a
very
good
knowledge
of
the
past?
If
you
do,
please
raise
your
hand.
Well,
many
of
you
maybe
not
know
a
lot
of
it.
I'll
give
you
a
brief
introduction
that
I
think
it
is
an
August.
Just
give
you
a
brief
introduction
and
if
you
take
a
high-level
perspective
and
from
traditional
big
application
to
as
small
a
small
services,
then
can
be
divided
to
even
smaller
unit
and
not
it
I
think
it's
model
is
driven
by
the
message
and
and
the
surface
and
the
will
consider
it
our
front.
B
Two
aspects:
I
will
mention
why
we
pay
attention
to
it.
As
I
mentioned
that
in
jetty
we
have
a
rich
uses
Narrows,
so
the
function
services
are
driven
by
events,
and
once
you
have
smaller
services,
we
can
find
a
specific
user
scenario.
No
matter
from
the
back
end
from
the
big
data
analytics
are
even
the
edge
computing
on
IOT.
B
From
the
perspective
of
the
business,
I
think
the
function
service
softness
will
bring
some
benefits
from
two
aspects.
Firstly,
from
the
perspective
of
the
user,
I
mean,
if
you
are,
writing
the
application
and
you
want
to
deploy
it
in
a
container
of
course.
Today
you
can
use
the
kubernetes,
but
of
course,
the
company,
the
complicity
is
still
very
high
and
the
keynote
you
know
that
a
threshold
is
still
high,
but
if,
in
your
application
you
have
a
template
to
follow,
then
you
can
simplify
the
process.
B
You
just
arrive
code
and
logic
of
application
and
it
will
be
less
complex
and
we
can
increase
the
level
of
automation.
Of
course,
the
production
is
very
important,
so
engineers
resource
is
the
most
expensive.
If
we
can
reduce
the
cost
in
this
aspect,
then
it
will
be
very
important
for
right
industries
and
also
traditional
industries,
and
second,
is
about
the
surface.
B
If
we
can
make
it
a
more
elastic
and
to
increase
the
granularity
of
these
services,
then
it
would
be
better
if
we
can
put
more
containers
on
that,
then
we
can
utilize
the
containers,
which
means
we
can
reduce
the
costs
for
a
large
company,
like
the
kingdom,
I
think
the
utilization
improvement
will
make
a
huge
contribution
to
the
cause
of
reduction.
Therefore,
I
think
the
buzz
will
provide
benefits.
Further
complicate
the
complex
complexity
as
well
as
the
utilization,
and
it
is
also
event-driven
model
and
it
will
be
simplified.
C
A
C
A
A
C
A
This
already
platform
for
enterprises
to
news.
We
found
there
lot
of
pitfalls,
because
this
is
a
very
new
technology.
In
terms
of
the
cloud
we
found
many
mature
technologies,
for
example,
AWS
number
function.
However,
the
AWS
has
a
specific,
specific
operation
models.
We
are
an
ecosystem
and
we
have
the
f3
and
demo
DB.
We
have
our
own
ecosystem.
We
are
also
thinking
how
to
use
the
fast
platform
to
encourage
with
our
own
ecosystem.
Based
on
that,
we
have
already
implemented
some
of
the
past
and
we
have
done
a
lot
of
work
on
it.
A
C
A
C
A
Chef
who
has
introduced
a
little
bit,
some
of
our
JT's
function
as
a
service
platform
and
its
use
cases
and
now
I'm,
going
to
talk
about
some
of
our
practice
and
experiences
where
we
use
far
as
performant
ujd
in
2014
started
to
do
this
container
work
being
English
area
for
many
years,
so
we
are
successful.
Our
proportion
of
containerized
is
very
high.
Old
applications
can
be
now
implemented
and
deployed
on
the
containers.
There's.
C
C
A
C
C
C
A
C
C
C
C
A
A
A
A
A
C
A
A
A
C
B
B
C
B
C
B
B
B
C
B
C
B
C
C
C
C
C
B
B
There
are
some
applications
which
will
have
a
low
requirement
for
the
latency
and
then
the
rest
I'll
request
a
food
to
the
pool,
the
management
of
the
poor.
It's
like
a
management
of
the
thread
pool.
It
won't
do
some
recycle
and
reuse
this
resources.
If
a
resource
is
not
not,
they
will
create
new
port.
C
B
C
B
C
B
C
C
C
B
B
C
B
B
Originally,
you
have
load
your
code
to
the
gate.
It
will
also
have
the
code
scanning,
but
now
maybe
we
just
are
some
of
the
the
code,
a
small
section
of
the
code,
the
set
the
code
snippet
and
then
you
can't
just
go
it.
You
can't
stand
it
so
now
we
I
think
now
the
function
called
scanning
is
available
for
Java
and
the
Python
after
you
scanned
it
and
it
will
be
uploaded
and
released
on
the
platform,
and
we
also
have
the
test
and
also
have
a
control.
This
is
whole
set
of
the
security
mechanism.
C
B
Problems
in
the
JD
Dawkins
business,
because
the
JD
is
developing
fast
training.
Therefore,
the
we
need
this
kind
of
service
function
service,
one
to
provide
a
full
lifecycle,
function,
management
from
development
to
operation
and
monitoring
and
focusing
on
JD's
scenarios.
We
were
integrated
with
a
middle
world
and
also
underlining
services.
I
want
to
support
him,
multiple
lingo,
multiple
languages,
in
addition
to
Java
in
the
past,
Jenny
yocha,
yocha
Java
for
the
middleware.
C
B
C
B
B
Thank
you,
Tom
seeing
so
if
you
are
interested
in
this
and
you
can
communicate
with
us
after
the
presentation,
we
have
some
specific
use
cases
like
intelligent
speaker
and
also
some
intelligent
image
processing,
as
well
as
the
application
in
the
mobile,
and
this
is
our
application,
which
of
course,
we
want
to
have
communication
with
the
community
and
if
you
want
to
share,
if
you
want
to
communicate
with
me,
maybe
we
can
just
talk
after
this
presentation.
Thank
you.