►
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
Running Blockchain as a Service on Kubernetes - Shan Yu, Alibaba Cloud
Blockchain is bringing innovations into many industries. At Alibaba, we believe Kubernetes is the ideal platform for running Blockchain-as-a-Service (BaaS) and managing blockchain applications. In this topic, we will share architecture and design of Alibaba Cloud's Blockchain service, which is developed atop Kubernetes. We will dive into how Kubernetes support Hyperledger Fabric with resource orchestration, elastic scaling, and uniform foundation across public, private and hybrid cloud. Moreover, we will talk about production-proven solutions and tips for solving typical pain points of running Fabric on Kubernetes, like cross-VPC consortium network, chaincode container mgmt, storage sizing & mgmt, high availability, etc. At the end of the topic, a demo will be presented to show how easy to build up a cross-company digital asset business application network using BaaS of Alibaba Cloud.
To learn more click here: https://sched.co/FuLn
A
A
First
of
all,
a
brief
introduction
of
broad
chain
and
past
I'm
sure
you
all
have
some
understanding
for
blockchain.
You've
listened
to
Leto
and
Henry
already
what
is
work
chain
in
an
essence.
It
is
a
distributed,
shed,
ledger
based
on
smart
contract
and
all
parties
of
a
transaction
or
not
to
reaching
consensus
and
the
transactional
history
is
tempered
proof
and
then,
apart
from
the
technological
side,
there
is
a
broader
perspective
to
approach
it.
We
think
blockchain
is
more
like
a
structure,
architecture
or
collaboration
model
connecting
organizations,
individuals
and
machines
throughout
which
network
values
can
flow.
A
A
A
The
second
type
is
a
totally
decentralized
systems
like
publishing
and
other
similar
systems
amount
organization.
Individuals
are
gathered
on
the
open,
Virginia
walk,
the
other
is
a
multicenter
systems,
not
an
alliance
chain
comparing
them.
We
believe
there
are
three
core
differences,
especially
for
an
assassin
shoot
with
access
control,
public
chain
and
alliance
chain,
they're
open.
Anyone
can
join
for
the
Alliance
chain.
A
A
Temporary
operations,
so
it's
a
complicated
system
and
it
delivers
a
complete
solutions
or
system
and
then
coming
to
the
applications.
There's
no
sets
and
standards.
We
might
find
various
types
of
applications
supporting
businesses
like
application
data
service,
staple
unstick.
For
so
that's
for
its
application.
A
If
combined
with
dhaka,
we
will
get
a
standardized
package
in
the
distribution
model,
guaranteeing
that
no
matter
how
complicated
the
software
is
no
matter
how
many
dependencies
there
are,
there's
one
set
way
to
package
them
and
then
deploy.
Then
it
gives
a
consistent
environment
so
that
it
can
run
anywhere
which
stuff
so
better
probability.
So
that's
the
advantages
of
providing
it
with
continuous
blockchain
with
kubernetes.
It
offers
a
lot
of
advantages
for
resources.
Now
all
maintenance
stay
for
unstable
support,
microservice,
architect,
seamless
integration,
safety,
segregation,
community
and
ecosystem.
A
That's
why
we
came
to
keep
calm
because
of
these
advantages.
Here,
I
wrote
micro
service
architecture
or
service
manager
for
a
blockchain
system.
How
can
it
be
integrated?
We've
done
some
analysis
for
the
boxing
system
itself.
Internally,
there's
a
strong
couple
architect
and
the
services
are
not
unique.
Universal.
A
Shows
the
history
of
either
MacLeod
in
terms
of
ball
chain.
Last
year
in
2017,
in
October,
we
launched
the
continued
service
option
solution,
which
enabled
many
module-based
a
one-step
deployment.
So
within
one
minute
we
can
deploy
a
network
for
corporate
and
then
in
December
we
transit
from
deployment.
We
migrated
to
Kira
Nerys,
so
we
were
able
to
use
and
orchestrate
the
applications
towards
end
of
July.
This
year
we
launched
a
best
supporting
it
was
more
point
to
which
naturally
alternative
offering
pass.
A
This
day
in
September,
we
launched
the
last
meeting
hyung
joong,
that
our
vast
supports
the
entrant
in
shows
book
chain.
It
is
also
available
in
Cuba
Navy's,
so
that
is
the
Alibaba
offering
in
respect
of
your
children.
This
is
a
clearer
picture
of
our
offerings
at
the
core
here.
That
platform
is
up.
Blockchain
is
a
service
starts
at
the
bottom
level.
A
It
is
the
Java
cloud
service
and
then
there's
security,
auditing
maintenance,
it's
not
contract
to
user
management
and
system,
entrant,
etc,
and
in
the
meadow
there
are
two
engines
hyper
ledger
fabric
that
shows
of
support
for
the
open
source,
blockchain
community
and
then
that's
the
end
of
chain.
So
there
are
two
engines
control
based
on
SDK.
A
A
This
page
field
shows
where
you
provision
with
bus
a
fabric
based
Alliance
chain.
The
architect
should
look
like
this
for
the
favorites
topology
or
the
peer,
your
honor
formula.
So
there
are
several
types:
we
support
a
cross
region,
construction
network,
so
different
business
organizations
can
be
located
in
different
regions
and
countries.
At
the
left
side,
it
is
a
Alliance
at
main
it
office
order,
our
CAFTA
zookeeper
ultra
and
their
peer
and
CA.
The
other
participants
can
find
the
nearest
location
and
deploy
their
pair
of
medication
and
eventually
join
in
the
insertion.
A
A
Blockchain
or
bass
by
nature
is
a
complicated
system.
It
includes
many
containers,
a
February
org
there
are,
there
might
be
children,
editions
or
four
PS
that
will
be
in
more
than
30
containers,
more
than
27
cents,
more
than
10
images.
So
that's
a
lot
of
volume
and
one
headache
is
that
the
service
has
strong
dependencies
redeploy.
A
A
A
Binaries
use
the
container
image,
that's
normal
and
the
charcoal
we're
using
the
Ali
OSS
object:
storage
for
image
storage.
It
is
with
the
container
registry
service,
so
the
charcoal
chat
museum
will
also
be
added
in
terms
of
configurations
because
of
the
complexity
of
the
system.
There
are
many
configs
and
secrets,
so
we
use
the
conflict
map
to
store
if
every
networks
and
then
the
secrets
will
keep
passwords
post
cigarettes.
A
Safety
related
things
which
are
value
can
take
inputs
from
the
users
and
with
child
value
we
can
customize
our
pan
chart
in
the
past.
Hem
try.
This
worked
really
hope
every
network
later
on,
we
divided
them
into
smaller
granules
pr1
chart.
So
it's
separated
more
flexible
in
terms
of
service
orchestration.
B
A
So
we
need
to
think
about
AJ.
There
are
many
facets
to
it
and
here's
a
breakdown
at
the
bottom
level.
It
relies
on
the
cloud
infrastructure
in
the
resource
service,
high
high
availability,
including
the
multi
AZ,
the
ECS
network,
storage,
etc.
On
top
of
it,
there
is
the
best
management
components
a
tree.
A
If
it's
single
point
error,
and
then
she
ran
an
is
Kosta
H
a
there
are
master
nodes
and
Walker
knows
to
guarantee
the
H
a
of
Kosta
knows
and
then
the
five,
the
fabric
components
the
Aldred
appear.
They
are
mutual
backups
and
then
it's
tough
for
us
to
do
to
find
sand
blocks
and
then
C
itself
can
be
a
cluster.
So
these
are
the
components
a
chain
guarantee.
A
B
If
we
want
to
have
a
good
that
process
notified
should
have
a
good
h8
design
and
when
it's
interacting
with
the
fabric,
which
we
should
also
provide
some
of
a
chain
capability,
because
if
they
were
first
get
access
to
the
port
note
what
this
is
health
checking
insurance.
If
there
is
any
problem,
then
they
can
turn
to
other
nodes
in
turn
and
transfer
to
port,
because
this
will
include
a
discovery
discovered,
maybe
you'll,
connect
into
the
connection
file
profile,
maybe
some
of
its
down,
but
on
the
channel
you
can
collect
some
of
the
available
services.
B
The
simpler
way
is
local
disk,
maybe
first
we
should
remove
those
that
it's
impossible
to
be
put
in
the
containers,
because
it's
not
physical,
so
it
will
easily
get
down.
Otherwise
they
even
you
have
good
and
she
designed
it
will
be
useless
in
kubernetes.
It
has
a
calling
function.
Originally,
maybe
the
pier
can
be
related
to
the
drivers
and
then
it
will
drift
it
to
another
on-premise
notes,
and
if
this
notes
is
not
on-premise,
then
that
will
have
some
problem,
because
the
people
will
have
some
extra
capability
to
solve
that
there
were
other
possibility.
B
Maybe
we
can
use
the
nose
data
to
nose
data
to
solve
that
problem.
For
example,
all
the
notes
has
been
has
failed,
then,
on
our
discover,
we
have
a
more
reliable
driver
to
recover
to
my
original
status,
so
we
have
selected
to
approach
first
and
file
system.
There
is
a
storage
there,
then,
and
also
the
block
storage
as
well.
For
example,
the
proudest
now
feet.
B
It's
very
common
to
us.
As
already.
Although
we've
already
used
some
storage
path,
fertile
ledger,
there
is
a
problem
and
this
ledger
cannot
be
deleted.
Along
with
the
growth
of
the
enterprises
that
the
letter
it
will
grow
at
the
same
time.
So
at
first,
if
you
don't
have
enough
storage
for
it,
then
how
can
we
address
this
problem?
B
And
we
know
that
there
was
Sibley's,
dramatic
scalar
for
this
scalar,
it's
transparent
for
the
applications,
the
application
would
not
be
disrupted
and
don't
have
to
restart.
So
this
would
be
very
friendly
to
their
users.
From
another
perspective
is
the
features.
Maybe
we've
already
have
a
100
gigabyte,
the
data
of
this
ledger
and
it
when
we
grow
bigger.
Will
it
reduce
these
pages,
so
you,
when
it
grows
bigger,
it
can
grow
together
to
get
together
with
the
features
any
can
have
better
features
in
improvement,
and
this
is
some
description
of
the
path
fabric.
B
We
will
use
the
cloud
disc
and
also
different
allocations
situation
of
the
cloud
disk.
So
just
now
we
have
talked
about
storage.
Let's
get
to
the
networking
consortium
networking
for
enterprises
blockchain
this
this
field,
it's
actually
what
they
focus
on,
because
in
the
real
conditions
we
it's
impossible
for
us
to
put
it
into
VC
P.
We
need
to
think
about
the
master,
the
consortium
networking.
There
was
some
existing
plan
already.
B
Maybe
you
can
bring
the
nodes
of
different
organization
to
and
vbc
and
communicate
the
BBC,
and
also
to
provide
a
description
of
different
different
notes
and
because
it's
a
centralized
structure,
so
it
cannot
show
the
features,
the
capability
of
different
enterprises.
It
can
also
satisfy
the
basic
demands
and,
for
example,
the
communication
of
different
major
network.
They
have
different
classes
and
through
the
through
the
UDS
new
Diaz
coming
ups
communications,
it
can
satisfy
their
common
communications.
B
As
for
the
Heinie's
and
for
those
some
like
key
enterprises
that
have
very
high
confidential
requirements
now
the
government's
and
also
the
bank's
there.
Maybe
they
will
think
about
the
security
and
also
for
the
financial
companies
and
institutions
they
will
if
they
can
pour
a
lot
of
special
lines
through
the
the
telecommunication
operators.
B
If
different
enterprises
have
taken
different
carrier
different
operators,
if
they
want
to
form
a
consortium,
how
can
they
connect
their
each
other
that
will
have
very
high
cause
if
you
have
form
the
original
enterprises
Network,
if
there
were
new
enterprises
want
to
enter,
then
there
will
be
a
threshold
and
also
quite
very
high
cost
as
well.
This
is
not
a
perfect
solution.
B
The
existing
light,
airy
cow
le
cloud
will
offer
some
of
the
solutions
and
products
we
can
combine
it
with
pass
to
provide
a
consortium,
networking
solutions
for
our
clients.
They
can
connect
different
VPC
on
the
network.
Then
this
is
a
kind
of
network
solution
that
connecting
different
nodes
and
through
the
taking
communicates
through
the
internal
networks
nodes.
B
Although
these
notes
is
not
existing
on
the
on
premise
database
or
in
not
on
premise,
they
can
connect
to
the
vbc
and
also
together
with
the
on
cloud
we
visit
it
communicates.
Then
that
is
a
hybrid
hybrid
cloud
that
is
across
different
across
different
enterprises
and
different
accounts.
It's
not
just
providing
ipq
different
attendees.
B
They
can
provide
descriptions
of
different
domain
names
like
level
shown
as
well.
That
is
a
Class
V,
PC
domain
name
services.
So
all
of
those
clients
and
users
who
want
to
SS
can
utilize
this
and
implementation,
which
also
think
about
this.
The
planning
that
we
beseech
session
organization
to
avoid
the
conflicts
of
the
names,
the
namespace
their
names.
B
A
B
Peer
when
they
get
in
century
drink
own
and
they
would
like
to
find
a
darker
demon
and
through
to
dr.
demon
through
the
fabric
czd
for
home
wand
image,
then
we
can
build
up
environments
and
also
compile
in
it
the
nose
or
gold,
the
the
other
chain
code
on
the
other
chain
code
inside
and
after
they
get
the
binaries
that
they
can
pass
on
pass
it
to.
Through
the
dark
demon,
we
will
generate
a
fabric,
OS
face
image
with
chain
code
with
matching
Cabana
and
then
I
was
started.
Then
this
will
become
a
start.
B
A
started
shrink
whole
container,
we'll
find
some
problem
in
it,
as
for
dr.
diamond
because
is
realigning
on
cabanas
kubernetes
host,
so
this
would
be
implemented
by
the
host
dr.
diamond,
including
some
of
the
containers.
It's
actually
separated
and
separated
from
the
system
of
the
kinetics
is
for
the
cycle.
Lifecycle
management
that
will
cause
some
promoters.
B
The
second
point
is
that,
because
it's
segregated
from
the
sub-q
Bonetti
system,
it's
difficult
to
use
the
name
space
of
the
kubernetes
to
do
the
segment
segregation
and
also
the
other
policy
description
or
the
definition
features
to
protect
its
network
and
it's
difficult
for
us
to
avoid
some
of
the
attacks,
access
attacking
access
from
the
external
Network,
and
we
also
explore
some
up
solutions
for
these
problems.
The
first
solutions
is
that
I
think
this
one
will
be
more
ideal.
B
If
I
put
in
this
the
service
environment,
my
elastic
and
highly
segregated
environment,
then
they
can
have
a
kernel
level
segregation
to
protect
the
chain
code.
So
that
is
wonderful
expectation
and
another
option.
Is
that
whether
possible
to
do
the
deemed
the
darker
inductor,
the
deemed
approach?
To
do
that?
We
also
do
some
testing
to
evaluate
this
option
and
we
have
use
of
yam
and
use
this
team.
You
can
you
can
have
a
look
at
the
code?
B
Maybe
you
cannot
see
clearly
here,
but
in
my
PDF
online
you
can
download
it
and
have
a
look
at
it.
The
advantages
of
it
is
that
they
don't
have
to
rely
on
the
hosts
notes
or
the
deductr
sucks
or
the
Dockers
demon,
to
generate
the
darker
docker
container,
because
for
darker
and
darker
you
can
call
it
from
the
calendar
and
for
all
the
twinkle
image
in
CCTV
or
from
the
hosts
notes.
Have
some
deal
with
Deb
preface
it.
B
Sec
prefix,
if
you
have
some
drinker
with
the
same
name
them,
they
will
have
some
conflicts,
but
in
the
dark
in
Dhaka,
then
you
don't
have
the
problem
of
the
trinkle
names,
because
all
the
Trinko
name
image
is
in
the
dark
and
dark
environment
and
they
steamed.
Indeed
they
are
going
to
clear
up.
So
that
is
very
convenient
for
us,
but
whether
there
is
some
limitation
here,
we
can
see
that
after
we
established
appear
or
when
I
want
to
restore
70-year.
A
B
A
In
the
road
and
the
forever
and
then
the
chain
hoist
ensign
becomes
slower,
that's
understandable,
because
in
the
Indy
there
are
corporations
and
then
that
transducer
the
and
then,
when
you
reboot,
the
PM,
know
there
can
be
problems.
One
also
very
important
thing
is
that
in
Pakistan
and
the
approaches
of
the
Euston
CSC
be
testing
continuous
integration,
etcetera,
but
for
production
environment
there
are.
There
are
too
many
challenges
with
the
stability.
That
is
why
we
haven't
said
in
the
production
environment
and
in
the
theater.
A
He
doesn't
stop
the
segregation
and
safety
issues
with
the
chain
code
and
then
what
we're
doing
today
in
walk
around,
as
you
say,
with
configurations,
we
got
to
satisfy
the
demands,
for
example,
guaranteeing
the
communication
between
Pia
and
chain
code
and
with
a
young
environment
the
core
configurations.
We
guarantee
that
pure
and
chain
communication,
and
you
by
using
dr
our
energies
of
our
mind.
We
can
manage
the
lifecycle
based
on
spirit
for
the
afterwards
clearing
you
can
use
diamond
set
and
then
lifecycle,
priests
are
EXA
command.
A
A
Nowadays,
baths
have
helped
many
of
our
clients
in
and
also
outside
of,
China
have
some
applications,
for
example,
product
traceability,
protections,
tough,
digital
copyright
and
supply
chain
finance,
and
also
data
access
sharing
because
of
the
real-time
I'm.
Not
elaborating
for
that
here
are
some
of
the
basic
patterns
of
the
blockchain
system
of
kubernetes.
The
application
to
Rome
itself
is
not
difficult
with
bus
or
any
other
minis.
You
can
come
to
an
alliance
chain,
trust
the
first
businesses
and
then
that's
the
VPC
interconnected.
A
A
And
we
need
such
an
approval
process
and
invitation
mechanism,
so
there's
a
prove,
and
once
we
use
that
channel
consortium
there
are
user
names.
There
is
a
text
message:
verification,
SMS
verification,
because
we
need
to
press
ctrl
to
make
sure
we
don't
lose
millions
overnight
because
of
a
simple
mistake
and
then
they
need
to
download
an
SDK,
and
we
also
need
to
download
a
demo
for
Mapo,
and
this
is
transfer.
This
is
connection
profiles
automatically
generated
the
demo
and
the
exponent
applications
to
not
use
it.
A
A
Once
we
cannot
were
connected,
we
see
the
chain
code
being
deployed
on
it
and
then
on
mobile.
We
get
to
build
the
virtual
assets,
the
colorful
models
and,
as
you
can
see,
the
blocks
are
any
number
each
time
you
create
and
shift
and
delete
the
chain
code
at
the
bottom
levels.
Notes,
because
we're
going
to
time
and
not
finishing
all
of
this,
we
need
to
introduce
you
to
our
home
page
and
please
let
us
know
what
you
think
so
much
for
my
presentation
today.
If
you
have
any
questions,
please
let
me
know.