►
From YouTube: RELOOK AT MICROSERVICES
Description
JakartaOne Livestream is a one-day virtual conference for developers and technical business leaders that brings insights into the current state and future of Jakarta EE and related technologies focused on developing cloud-native technologies for enterprise Java applications.
Featured Speaker: Emily Jiang
08:00-08:50 GMT-04
A
Is
it
okay?
If
I
continue
first
in
german
and
okay,
perfect
from
ebm,
emily
is
yeah
native
architect
by
ebm
and.
B
Thank
you
for
coming
yeah
great
to
be
here
and
hello.
Everybody
to
I
mean
hello
to
everybody
in
this
room
with
claudeline
and
my
greetings
to
the
people
who
join
remotely.
So
in
this
session
we
will
just
look
at
the
microservices.
B
Actually,
the
microservices
has
been
around
for
for
some
years
right.
So
it's
in
this
session.
I
want
to
look
at
the
history
of
macro
service
and
also
the
the
facts
and
amaze
about
macro
service
and
things
that
you
should
know
about
micro
service
and
then
we'll
look
at
its
macro
service
is
going
to
be
obsolete.
Again.
B
This
is
the
technology
like
a
repeated
itself
again
and
again
right.
So,
let's
first
rewind
back
back
to
like
back
up
to
the
like
older
days
and
and
let's
see
what
like
generated
micro
services,
so
it's
back
in
1997
ibm
released
the
edb
who
who
knows
egb
yeah,
that's
great.
So
it's
a.
I
remember
the
one
I
joined
the
ibm.
B
I
had
to
go
on
a
course
to
learn
about
edb,
so
you
have
to
use
ide
to
generate
stops,
be
able
to
do
the
remote
like
the
service
course
and
it's
very
complicated,
and
the
other
thing
is
only
like
the
technology
for
java.
Actually,
we
want
to
do
this
remote
code
between
the
services
in
the
different
languages,
so
with
that
it
came
as
aoa,
so
service
oriented
architect.
So
the
basic
concept
of
soa
is
about
language
agnostic,
so
services
can
be
written
in
any
kind
of
languages
and
et
cetera.
B
So
this
is
very
popular
in
2000
and
then
also
is
soe
came
with
the
enterprise
integration
bus.
So,
basically,
the
idea
is
using
the
messages
between
the
services
and
for
these
communications.
It
uses
a
protocol.
Soap
simple
object,
access
protocol
and
is
for
soap.
Actually,
the
original
design
is
simple.
However,
it's
not
simple
and
everything
needs
to
wrap
up
in
xml.
B
B
Actually
I
I
kind
of
fortunately
or
unfortunately,
I
worked
on
the
ws
addressing
on
ws
policy
and
dealing
with
the
xml
all
the
time,
and
you
have
this
rootstock
to
java
all
kinds
of
different
tooling,
it's
really
complicated
and
then
it's
like
in
the
2008
that
kind
of
period
time
and
people
think
okay.
This
is
too
complicated.
We
need
something
simpler,
so
that
came
with
the
rest,
so
representational
like
a
state
transfer.
So
basically
it's
with
rest.
Basically,
you
have
the
get
put
and
post
on
a
delete.
B
So
this
is
the
rest
res
electric
stakes
is
become
popular
and
popular,
so
it's
a
and
then
service
that
will
be
able
to
like
use
rest
to
communicate
with
each
other
and
then
like
once
is
stable
and
we
need
the
name
for
these
services,
so
in
2011
or
2012
is
the
micro
services
that
came
up.
So
basically
it's
okay,
this
this
is
just
a
services
they
communicate
and
sometimes
via,
like
originally
via
this
rest.
B
So
we
call
it
the
microservices
and
because
of
the
micro
service
name
was
created,
and
then
we
need
a
new
name
for
the
traditional
application
and
then
we
give
it
a
a
kind
of
a
name
called.
Let's
call
it
a
monolith
right
like
what.
What
is
the
monolith?
Okay,
the
bigger
blob
and
then
is
the
kind
of,
and
then
you
know
it's
looking
all
this.
We
always
like
the
new
things
and
then
we
just
say:
okay,
all
the
things
is
is
bad
and
is
it
is
monolith
bad?
Is
it
older
fashion?
B
It
is,
is
evil
like
people
in
the
past?
Oh
the
evil.
Monolith
is
evil
when
you
need
to
go
to
medical
services.
So,
let's
look
at
what
is
the
micro
service
first
and
then
we
can
go
back
to
think
where
the
model
is.
It
is
evil
or
not.
So
it's
according
to
the
macro
service,
dot
io,
so
micro
services
need
to
be
maintainable
and
testable
and
the
other.
The
very,
very
important
thing
about
microservice
is
need
to
be
independent.
B
So,
as
you
can
see
here
like
on
the
top,
so
basically
the
each
micro
service
has
a
clear
identity
right.
It
should
be
able
to
be
released.
I
mean
independently
shouldn't,
I
need
to
negotiate
with
others,
and
it
should
be
owned
by
the
single
team.
Like
a
small
team,
even
with
one
team
you
own,
like
100
micro
service,
you
did
it
wrong,
you
haven't
done
the
right
architecture.
B
We
have
done
micro
services
for
a
decade,
so
it's
a
there's
kind
of
some
myths
about
microservice
and
then
often
people
ask
me
I
don't
know.
Maybe
you
also
go
to
these
like
questions.
How
small
I
should
do
microservices
should
be.
Is
it
10k
big
enough?
Is
that,
like
a
one
class,
there
is
two
classes
too
big?
Is
that
just
one
class,
so
this
is?
This
is
all
wrong
questions
is
this.
Is
this
is
if
you
go
to
this
question,
that
means
that
person
haven't
understand
the
microservices
correctly.
B
B
B
Actually,
in
preparing
for
this,
for
this
like
presentation,
I
did
a
survey
I
asked:
do
you
think
a
microservice
going
to
replace
the
monolith?
You
can
see
more
than
half
says?
No,
so
usually
rest
assured,
if
you're
doing
monoliths-
and
you
do
not
have
any
problems-
you
do
not
have
problems,
don't
rush
to
macro
services,
just
for
the
sake
of
micro
services,
so
the
because
we
can
only
hold
three
things.
I
just
mentioned
the
top
three,
so
servicing
is
kind
of.
B
If
you
have
some
organization
problem,
you
said:
okay,
my
team
is
not
performing
well,
probably
because
we
are
doing
monoliths,
we
should
do
micro
services
to
increase
the
performance
or
your
application
is
not
running
fast.
Maybe
I
should
like
to
do
micro
services
like
you,
should
have
to
look
at
the
problem.
Why
is
it
slow
is
that
you
are
not
using
the
correct
runtime
right?
If
you
like,
do
not
look
at
the
problem
and
write
it
to
macro
services.
Eventually,
you
will
create
a
distributed.
Monoliths
is
even
worse.
B
B
They
they
just
did
a
talk
about
why
they
moved
from
monolith
to
micro
services
and
found
out
actually
their
team
performing
badly.
They
they
have
like
a
performance
is,
is
kind
of
deteriorate,
so
they
have
to
merge
the
micro
services
together
into
monoliths
again
and
also
because
there
is
also
a
tweet
about
macro
services.
So
this
this
company
is
not
alone
and
also
adam
bin
is
also
as
such,
I
mean
suggested
something.
So
you
have
funny
ways
to
merge
microservice
into
monoliths
and
then
give
it
a
cool
name.
So
it's
the
other.
B
So
what
has
the
micro
service
effects?
Microservice
is
about
independent
deployment,
and
it's
a
it's.
Basically
the.
If
you
cannot
remember
everything
I
said
remember
just
one
thing:
microservice,
it's
a
it's
not
about
the
size.
It's
about
independent
deployment.
You
need,
I
mean
ci
cd,
to
enable
the
independence.
B
So,
basically,
you
need
a
continuous
integration
and
continuous
deployment
and
also
is
owned
by
a
small
team.
Even
if
you
have
a
big
like
infrastructure
on
a
many
macro
services.
Eventually,
you
all
aggregate
all
these
micro
services,
synchronize
micro
services
and,
at
the
same
time
again,
you
just
like-
are
not
doing
the
correct
micro
services
again.
B
So
the
other
thing
is
the
second
thing.
Micro
services
doesn't
do
color
sharing
like
back
in
the
2018.
I
don't
know
2019.
So
I
was
at
the
berlin
where
developer
conference,
and
actually
I
have
a
lot
audience
in
the
in
the
room.
Another
is
one
person
saying:
actually
how
can
you
enable
code
sharing,
so
I
need
to
like
do
a
shader
library
on
a
set
in
microservices.
B
B
You
need
a
new
technology,
new
programming
language,
to
go
with
it,
and
then
people
say:
oh,
you
need
to
use
a
go
or
you
create
a
new
language
to
do
it.
Actually,
micro
services
can
be
written
in
in
java.
Java
actually
have
a
great
ecosystem.
I
will
demonstrate
to
you
later
and
to
enable
to
do
really
really
good
microservices
development,
and
also
you
could
choose
the
goal
of
python
to
go
with
it.
B
Okay,
what
it?
Okay,
you
by
now,
you
you,
you
kind
of
roughly
understand
the
microservices.
What
kind
of
things
you
should
know
if
you
do
microservices
is
microservice
the
default
answer
for
your
problems.
When
you
have
problems,
you
say:
okay,
yeah,
I
have
a
problem
with
my
monolith
is
for
somehow
is
the
monitoring
is
not
doing
well.
I
can't
figure
out
what's
wrong.
If
something
is
wrong,
probably
I
should
go
with
micro
services.
B
You
have
to
ask
why
you
having
to
go
to
monitoring
if
you
directly
go
with
micro
services
without
adding
any
any
metrics
and
et
cetera,
you
still
run
into
the
same
problem,
so
you
need
to
find
out
what's
the
problem
first,
so
it's
don't
like
a
rush
to
microservice,
just
because
you
think
micro
service
is
going
to
fix
everything.
No,
the
second
thing
microservice
needs
a
really
good
design
like
a
micro
service
is
not
simple.
You
need
to
think
clearly
about
it.
You
need
to
learn
about
domain
driven
design.
B
So
basically
you
need
to
like
structure
your
mac,
your
your
like
application
in
a
nice
domain,
driven
this
kind
of
architecture
and
then,
if
you
want
to
do
the
synchronized
communication,
you
also
need
to
learn
about
event
driven
architecture.
B
B
We
got
to
use
it
and,
let's
use
that-
and
this
is
like
you
can
learn
in
your
spare
time,
but
do
not
put
into
just
the
use
the
personal
goal
and
driven
the
the
macro
service
design
and
when
you
do
the
micro
services,
obviously
you
will
have
a
few
micro
services
and
then
they
will
deploy
the
to
the
cloud
so
day.
Two
operation
is
very,
very
important.
B
You
need
to
have
a
right
matrix
and
you
need
to
have
the
choosing
availability
or
also
you
need
to
be
able
to
communicate
with
the
cloud
infrastructure
and
also
you
need
to
use
the
standard
programming
model
so
that
you
can
communicate
these
micro
services
no
matter
what
kind
of
technology
or
they
use.
So
it's
a
no
matter
which
run
I
may
use
the.
The
fourth
thing
is
the
kind
of
way
you
do
macro
services.
B
Obviously
you
you
need
to
adopt
agile
kind
of
the
the
kind
of
practice
and
also
you
have
to
change
your
team
structure,
make
sure
you
are
doing
micro
services.
Not
like
you
have
a
humongous
team.
I
manage
like
10
20
micro
services
and
then
the
fifth
thing
is:
if
you
want
some
refactor,
if
you
really
think
like,
because
you
have
a
monolith,
some
part
of
money
lists
need
to
be
scaled
out
or
scaled
down
like
a
independently.
B
This
is
really
valid
user
case
for
you
to,
like
top
your
micro
service
into
like
a
utopia,
is
into
micro
services,
but
you
need
to
be
mindful,
I
hope,
to
you
like
a
splitter.
These
monoliths
into
microservices
there's
also
some
patterns
like
a
strangler
pattern
and
some
other
patterns,
and
also
in
ibm.
Actually,
we
use
ai
like
how
to
direct
it.
Like
a
refactor
monolith
to
microservices,
we
have
a
pro.
We
have
a
project
called
a
mono
to
macro.
B
So
basically,
this
is
we
use
ai
to
monitor
like
the
real
business
operation
and
use
that
data
and
to
figure
out
the
the
context
and
etc
and
to
be
able
to
like
recommend
what
the
macro
service
boundaries
are
and
because
today
is
the
kind
of
micro
service
and
the
main
reason
development
today.
So
there
are
many
like
talks
about
microservices
and
all
the
best
practices
so
check
out.
B
Definition
so
it's
about,
like
you,
understand
the
business
operation,
and
then
you
divide
the
boundary,
so
each
boundary
is
so
that
will
be
become
like
a
single
macro
services.
So
it's
a
kind
of
independent
operation.
B
So
I
briefly
also
mentioned
the
edd.
So
udd
is
a
kind
of
event
driven
architecture.
When
you
do
the
event
driven
design
and
often
you
will
see
kafka
here
so
basically,
the
event
driven
design
is
like
a
micro
service,
do
not
directly
like
connect
or
do
a
http
call
to
others
the
the
communicator
via
messages.
So
here
it
will
be
message
broker
and
often
it
has
a
database.
You
need
to
persist
the
the
messages
so
kafka
is
will
fit
into
into
here.
B
B
What
do
you
really
need?
You
do
not
hang
up
with
that
with
the
term
monolith
or
microservice.
What
you
really
need
is
a
cloud
native
application.
What
is
a
cloud
native
application?
So
it's
the
cloud
native
application.
Obviously
it
needs
to
be
native
to
the
cloud
I
hear.
I
just
talked
about
the
application
itself,
so
this
application
is
need
to
be
restful.
Basically,
it
needs
to
be
a
kind
of
lace
and
it's
like
a
cat
or
not
pad,
so
it
can
be
grow
or
shrink,
so
it
can
be
duplicated
or
all
kind
of
decreased.
B
The
other
thing
is
need
to
be
configurable.
So
so,
basically,
if
you
want
to
configure
the
value
of
the
any
configuration
properties,
you
should
not
need
to
like
a
redeploy
application,
so
you
should
adopt
a
12
act
up
so
externalize
your
configuration.
B
The
fourth
thing,
because
if
you
go
to
the
cloud
obviously-
and
you
will
run
run,
run
out
run
into
the
situation
or
you
couldn't
remember
which,
like
app,
is
doing
what
so
it's
need
to
be
discoverable,
so
it
can
document
itself.
The
fifth
thing
is
need
to
be
secure.
If
you
deploy
a
lot
like
a
cloud
native
applications
and
each
of
them
met
like
have
some
sensitive
information,
so
you
need
to
secure
them
and
also
in
the
need
to
monitorable
and
not
traceable.
B
If
something
goes
wrong,
you
know
which
one
is
wrong,
so
you
can
fix
them.
So,
finally,
in
ep,
smart
intelligent,
so
basically
it's
the
cloud
native
application.
Why
is
it
is
a
cloud
native,
so
it
be
able
to
like
a
communicator
with
the
cloud
infrastructure
tell
the
cloud
infrastructure
whether
this
application
is
ready
to
receive
a
request
or
not,
or
if
it's
a
sync.
Is
there?
Oh,
it's
a
I'm
not
going
to
recover
myself.
Please
restart
me.
B
So
this
is
the
kind
of
thing
and
then
you
may
think.
Okay,
this
is
a
cloud
native
application
is
really
difficult
right.
It's
like
let
alone
your
business
logic.
I
mean
we
are
paid
to
create
a
applications
like
with
these.
Actually
requirements
like
you
have
to
work
at
actual
time.
Probably
you
you
have
to
work
over
the
weekend,
etc.
So
it's
a
luckily.
We
have
the
cloud
native
api
standard
to
help
you
so
micro
profile
energy
cutter.
Yes,
this
is
all
standard.
B
This
is
all
clouded
native
standard.
When
I
call
it
a
standard.
The
proof
is,
you
have
a
lot
of
runtimes
to
choose
from
it's,
not
only
one
vendor.
For
example,
we
know
like
our
companies
have
been
like.
You
have
companies
moved
very
fast,
sometimes
the
one
company
being
bought
by
another
and
then
top
organizations
like
a
director
said:
okay,
we
need
to
use
a
different
rent
time.
B
If
you
use
this
open
standard,
that's
fine,
yeah!
It's!
If
you
previously
run
wildfly
now
you
can
run
liberty,
no
problem
right.
So
this
is
the
real
beauty
of
this
standard.
B
So
today
is
also
jakarta.
One
live
stream,
germany,
so
it's
also.
We
will
talk
about
a
lot
like
later
about
the
jihadi
eaton.
So
one
thing
I
want
to
highlight
jakarta:
eternity
has
a
new
profile
is
called
core
profile.
So
you
use
this
as
a
standard
from
this
core
profile
to
create
a
services
to
create
a
cloud
native
applications
plus
that
you
will
use
the
apis
from
macro
profile.
So
micro
profile
is
the
like.
B
Is
the
community
so
pin
also
like
a
specifications
together,
offer
you
a
very,
very
solid
cloud
native
application
design
pattern?
So,
as
you
can
see
here,
you
can
use
the
decoder
stacks
to
create
an
application,
and
you
use
a
micro
profile
register
client
to
connect
these
micro
services
together
using
rest,
if
you
say:
okay
yeah,
I
want
to
do
the
synchronized
communication,
that's
fine!
You
can
use
reactive
messaging,
so
this
is
directly
you
can
plug
in
kafka.
That's
fine!
We
will
do
that
and
also
the
config
is
directly.
B
This
imp
implement
12
factor
up,
basically
externalize
your
configuration,
and
then
you
use
open
api
to
document
it
and
you
use
for
micro
profile
photo
tolerance
to
secure
it.
You
use
the
no
use
this
one
to
make
it
the
resilience
you
need
use.
The
micro
provided
wt
to
make
it
secure
and
also,
if
you
have
a
frequent
like
a
query,
to
run
and
you
might
consider
to
use
a
macro
profile
graphql
and
on
the
top.
B
This
is
all
about
like
analytics
observability,
so
you
use
open,
choosing
to
directly
connect
all
the
trees
back
together
to
form
one
trees
and
also
you
use
micro
profile,
health
directly,
communicate
with
kubernetes,
and
you
use
matrix
directly
at
any
additional
application
level
matrix
and
also
give
you
a
provider
some
like
gvm
metrics
as
well,
and
as
you
we
know
like
open
tracing
and
open
sensors,
merge
together
and
become
open
telemetry
in
macro
profile.
We
are
currently
working
on
the
micro
profile
telemetry
to
just
align
with
the
open
telemetry.
B
So
with
this
that,
with
this
kind
of
the
release
that
we
will
just
replace
the
micro
profile
patricia
so,
as
you
can
see,
you
have
a
full
stack
of
the
cloud
native
apis
available
to
you.
All
you
need
to
do
is
just
use
this
api
and
concentrate
on
your
business
logic.
Your
application
is
very,
very
small
and
kind
of
the
first
to
get
started.
B
Obviously
I
talked
about
a
lot
of
cloud
native.
Obviously,
cloud
native
with
the
destination
is
in
into
the
cloud
and
you
need
to
make
sure
it's
kind
of
in
the
cloud.
Natively
first
thing
you
need
to
do
is
containerize
this
application.
Obviously
you
either
use
portman
or
docker
who
use
docker
at
the
moment.
B
Okay,
great,
who
uses
portman
okay,
so
doka
is
still
on
the
right.
However,
I
think
people
gradually
probably
will
move
to
parliament
because
of
the
other
kind
of
the
license
and
et
cetera
and
with
the
one
thing
I
want
to
say
with
the
cloud
native
application
and
if
you
use
standard
apis
and
cetera,
it
can
run
perfectly
in
either
private
cloud
or
public
cloud
or
hybrid
cloud
without
any
single
change.
B
So
is
the
can
you
see
is
big
enough,
yeah
good,
so
here,
actually
I
use
micro
profile,
starter,
so
micro
profile,
starter.
Give
you
way
to
start
to
create
a
micro
profile
like
applications.
If
you
use
the
starter.macro
profile.o
or
if
you
use
any
like
a
ide,
we
use
vs
code
yeah
great.
So
if
you
use
vs
code
also,
we
have
the
macro
profile.
Extension
pack
basically
allow
you
to
create
a
micro
profile
application
and
who
use
intelligent.
B
Okay,
also,
we
have
a
plug-in
for
the
intellij
to
directly
create.
This
is
a
micro
profile
application.
So
at
the
moment,
so
I
have
at
the
moment
I've.
I
have
a
two
application
created,
so
it's
a
service
a
and
a
service
b
right
service,
a
like
a
basically
a
simulator
like
a
service.
A
is
a
one
micro
service
and
a
service
b
is
another
micro
service,
so
service
a
like
a
invoker
service
b.
So
in
service
b
it
has
a
two
api
so
why
api
is
kind
of
the
do
something?
B
So
it's
basically,
this
is
really
really
like
a
simple
operation.
It's
a
guess,
see
you
can
see.
This
is
rest.
Restful
call
so
this
is
get
and
there
is
a
path
we
use
the
dxrs
chicago
web
server
rest
for
web
services,
and
you
can
see
here
I
deliberately
put
in
some
sleep
and
through
some
exceptions,
to
yeah
to
demonstrate
that
sometimes
is
not
stable,
and
this
is
one
like
operation.
The
other
operation
is
kind
of
this
operation
is
secured.
So
this
is
the
endpoint
is
a
ghetto
operation.
B
However,
I
use
the
decatur
security,
so
basically,
I
only
allow
rules
belong
to
protected
his
invocades,
like
her
backhand,
so
is
the
now.
If
you
now,
if
like
in
the
service
a
how
could
I
be
able
to
make
a
service
a
yearbook
service
b?
So
it's
in
service
a
so
for
invocation,
the
service
b
like
endpoint.
So
basically
I
use
the
micro
profile
rest
client,
so
I
created
api.
B
I
created
the
api
for
micros
for
macro
service
b
cloud
native
application
b.
So
this
is
an
api,
do
something
it's
a
get
operation
and
it's
a
path
and
then,
like
I
annotated
with
the
registration
client.
So
this
is
the
director
from
macro
profile
apis
and
then
in
my
like
textiles
resources,
I
can
directly
inject
inject
this
service.
So
basically
I
do
I
inject
a
registered
client
and
interact
with
this
this
service
after
the
injector
this
service.
I
can
directly
call
this
service.
B
I
can
do
a
service,
don't
do
something
and
you
measure
you
measure
confused
like
no
is
a
service
b
is
up
running
like
in
the
one
port
a
service
says
here:
how
does
a
service
and
a
know
where
service
be
is
so
we
use
a
micro
profile,
config,
so
macro
profile,
config
so
directly,
like
provides
a
macro
profile
that
configure
properties
file.
So
basically,
you
can
specify
any
properties
from
here
and
it's
a
the
property
will
make
make
it
available
to
the
application.
B
So
you
can
do
the
like
a
fully
qualified
class
name
for
the
service,
basically
commodore
examples
in
4g
client,
the
service,
slash,
npress,
slash
url,
so
I
put
the
service
bits
the
backhand
there.
So
you
know
is
that,
like
a
service,
no
a
know
where
service
be
so
in
this
way,
as
you
can
see,
I
started
the
service.
A
already
is
on
the
1980,
so
this
is
running
on
open
liberty.
Running
here
is
the
ninth.
Let
me
go
go
in
here:
here's
the
1980.!
B
So
this
this
is
an
appointment
directly
according
to
service
b.
So
here,
like,
let's
try
it
out
here.
I
said
hello
cloud
native
cloud,
dev,
2022,
service
b.
Count.
Let
me
like
go
to
service
b,
so
I
study
the
liberty
in
the
dev
mode
so
which
means
is,
if
I
change
something
it
will
automatically
figure
out.
I
have
some
change
and
it
will
reload
everything.
So
I
will
say
hello
again.
B
So
before
the
service
be
a
count,
so
if
I
directly
code
code
call
it
again,
so
you
can
see
straight
away
like
you
know,
flash
set
millisecond,
you
can
see,
it's
restarted
itself
and
reloaded
itself
is
said
to
start
start
again.
Okay,
so,
let's
me
like,
like
because
service
b
is
not
stable,
as
you
can
see,
here
is
really
slow,
right,
really
slow
and
then
sometimes
this
doesn't
work.
So
how
can
I
increase
the
resident
microservice?
How
can
I
create
a
resilient
cloud?
B
A
native
application-
you
say:
okay,
just
fix
the
service
building,
just
fix
that.
Sometimes
you
don't
have
an
authority
to
fix
that.
Sometimes
you
all
you
can
do
is
change
yourself.
This
is
seemed
true
right,
like
for
the
family.
You
just
have
an
argument.
Is
that
oh,
you
are
doing
that
wrong.
You
need
to
change.
It's
really
really
hard
to
change.
Others.
All
you
can
do
is
change
yourself.
So
the
same
thing
applied
to
the
software
development.
So
let's
change
that.
B
So
I
demonstrate
to
you:
how
can
you
easily
change
yourself
change
your
the
operation,
like
your
own,
so
in
here,
because
this
is
the
operation
I
have
so
the
on
client
side.
So
all
I
need
to
do.
Actually.
I
could
use
the
micro
profile
like
for
tolerance.
I
can
do
a
retry
if
something
doesn't
work.
I'll,
retry
and
also
we
saw
is
isn't
slow.
I
can
say
okay,
please
time
out
if
it
does
not
return
within
200
milliseconds.
B
As
you
can
see,
it's
no
longer
no
longer
like
slow
and
no
longer
give
you
an
example,
a
wrong
wrong
information.
However,
I
see
this
number
clearly
so
the
16
right
here
17
here
you
directly
again
19
what
happened
to
18
because
18
failed.
It
automatically
did
a
retry
and
give
you
17..
So
this
is
just
to
tell
you
is
how
simple
it
is
to
use
a
micro
profile.
The
other
thing
is
kind
of.
If
you
want
to
run
in
the
cloud,
the
other
best
practice
is
you
always
for
some
very
critical
operation.
B
B
B
So
the
other
thing
is
the
kind
of
I
want
to
show.
You
is
with
a
macro
profile.
Basically,
all
the
like
a
jax
s
register
for
operations
has
documentation,
so
I
have
the
localhost
1980
so
open
api.
B
So
you
can
see
all
these.
The
all
the
jacks
has
all
the
kind
of
like
the
rest
for
services
has
been
nicely
documented.
So
it's
a
this.
Is
the
kind
of
anybody
know
swagger?
Okay,
that's
good!
If
you
know
swagger,
you
know
this
is
identical
to
swagger.
That's
because
we
adopt
the
swagger
and
then
you
open
laboratory.
B
You
also
have
a
swag
ui
before
you
ask,
so
we
have
a
swagger
ui
and
you
can
you
can't
just
that
quote
this
is
the
kind
of
you
can
see
it
do
something
200
or
400
means
what
and
then
you
can
try
it
out
here
if
I
say
hello
and
execute
so
you
can
see
that
like
like
a
response
straight
away,
so
the
other
thing
is
also.
I
mentioned
that,
like
I
haven't
done
much
and
it
gave
me
a
lot
of
metrics
already.
So
if
I
do
localhost
1980
slash
matrix.
B
So
you
can
see
it
have
a
lot
of
metrics
for
you
and
you
you
can
get
like
the
requests
and
also
because
you
use
for
tolerance.
It
even
tells
you
how
many
times
being
timed
out
how
many
times
the
retry
has
been
done
and
all
the
like
operations
like
ajax
operations.
B
Rest
for
api
course
is
being
traced
as
well.
So
it's
towards
the
end,
and
also
because
I
added
like
a
special
matrix
like
other
matrix,
like
a
mole
matrix,
also
give
me
more
matrix.
So
I
show
you
here
is
because,
like
a
simply
time,
that's
from
directly
macro
profile
matrix,
so
here
more
matrix,
you
can
get
more
metrics
and
also
you
have
to.
If
you
want
to
count
how
many
times
has
been
invoked,
your
ic
ad
counted
as
well.
B
So
this
is
the
kind
of
the
the
the
major
thing
I
want
to
mention.
The
other
thing
I
mentioned
because,
like
you
eventually
want
to
deploy
this
application
to
the
cloud
like,
obviously
you
have
a
kubernetes
and
there
is
there's
a
something
in
the
caller
in
kubernetes
we
know
we
have
a
laminate
and
readiness
and
a
startup
attack,
and
then
your
microprofile,
we
have
three
endpoints
slash
health,
slash,
live
slash,
radiance,
large
startup
to
directly
like,
like
a
correspond
with
this
kubernetes,
the
probes.
B
Okay,
this
is
a
laminate
check
and
then
you
can
implement
this
health
check.
This
is
like
as
a
method
the
implementation
you
could
say.
If
you
round
a
ram
means
this
that
this
one
is
not
going
to
leave
but
just
destroy
this
one,
and
also
for
the
readiness
check
all
you
need
to
do.
Annotation
with
readiness-
and
here
you
might
say,
okay,
when
is
your
cloud
and
native
applications?
Ready,
maybe
is
the
wins.
Kind
of
database
can
be
connected
and
the
the
third
thing
is
startup.
B
Why
we
introduce
the
startup
in
the
kubernetes.
Anybody
know,
okay,
so
in
a
startup
is
because
I
think
gradually,
people
all
realized
that
not
all
cloud
native
applications
can
start
in
the
split
millisecond.
Some
of
them
will
start
within
a
few
min
a
few
seconds,
like
maybe
20
seconds,
even
but
kubernetes.
You
know
it's
kind
of
the
constantly
trying
to
like
do
a
lemonade
attack
and
also
do
readiness
attack
and
then
find
out.
B
Okay,
you
are
still
not
alive,
and
maybe
is
your
your
roundhouse
memories
need
to
destroy
you
and
then
kubernetes
run
into
the
data
log,
basically
before
you're
ready
to
start
and
accurately
and
then
restarted
again.
So
that's
why
it
came
up
with
a
startup
attack
so
basically
do
not
attack
alumni's
readiness
be
before
this
application
said.
I'm
started
so
this
is
about
starter.
B
So,
as
you
can
see
here
in
the
micro
profile
health,
you
can
see
localhost
1980
like
here
I
can
see
health
start
live,
so
you
can
see
all
these
json
like
a
display
and
also
you
can
see
ready
and
et
cetera
and
with
I
listened
to
a
podcast
regarding
about
best
practices.
B
One
company
has
been
doing
in
their
like
companies,
he
said.
Actually
you
and
my
company,
I
established
best
practices.
All
the
micro
services
they
all
need
to
have.
One
single
class
is
a
they
need
to.
Like
a
provider,
one
endpoint
blah
blah
slash
health,
so
that
I
can
plug
into
kubernetes
with
the
macro
provider
health.
You
don't
need
to
do
anything,
it's
just
it's
automatically
there.
You
just
implement
it.
So
this
is
why
standard
is.
Why
is
so
important?
B
I
save
a
lot
of
effort
to
like
to
negotiate.
What's
the
best
api,
like
what
kind
of
the
class
structure
should
be
like?
Okay,
so,
prior
to
this,
like
to
this
presentation,
I
already
like
deplored
this
containerized
is
the
two
services
and
also
published
to
like
my
local
docker
desktop.
B
B
Okay,
if
you
do
not
know
check
it
out
is
the
android
eclipse
foundation
is
really
really
cool
and
then
you
you
run.
This
is
the
like
optimize
the
the
image
and
then
you
create
a
deployment
yamo
in
the
deployment
yamo.
We
have
two
app
when
the
app
is
so
basically,
you
just
say
the
container
to
this
container
image,
and
this
is
important,
so
you
may
be
asking
and
in
service
a
I
directly
hardcoded
a
service
b's
url
right.
Do
you
remember?
B
I
said
localhost
in
1981,
but
once
you
deploy
to
kubernetes
is
no
longer
a
local
host
is
so
here
you
can
configure,
and
you
say
why
you
change
it
to
be
underscore,
because
the
thought
is
not
not
a
legal
character
in
like
as
an
environmental
variable.
So
this
is
the
beauty
about
macro
profile
config.
So
it's
able
to
search
the
demand
variable
with
the
normal
like
variables
and
if
the
character
is
illegal,
it
automatically
converts
to
the
go
and
continue
do
the
search.
So
it
knows
okay,
we'll
go
go
in
here.
B
B
How
to
code
it
in
my
application
here,
like
I
inject
interact
as
a
like
a
conference
and
a
conference
is
different
in
this
micro
dash.
Configurator
properties
is
the
cloud
dev
2022,
but
in
the
like,
once
it
deployed
to
the
cloud
I
can
overrate
it.
I
can
say
java
land,
no
javelin,
javelin,
2022
and
chicago
one
live
stream,
german
and
now,
if
I
like,
go
to
this
directly,
go
to
this
like
endpoint
notice
on
the
which
port
is
on
the
380.
B
B
B
Okay,
so
now
is
that
probably
you
by
now
everybody
have
heard
of
service
right?
Have
you
all
heard
our
service?
Are
you
thinking
you
are
going
to
do
service?
Anybody,
no
or
thinking
you
might
ask.
Is
the
cloud
native
application
obsolete
again.
B
So,
first,
what
is
the
service?
Actually,
services
doesn't
mean,
serverless
doesn't
mean
no
server.
Often
it
has
server
most
time
it
has
server.
It
only
means
you
don't
need
to
worry
about
server.
It's
very
small
and
the
other
thing
is
people
often
rush
to
like
a
surplus.
They
are
not
even
in
public
cloud,
they
are
on
prem,
they
say:
oh
okay,
we
need
to
do
service.
Do
you
understand
a
service
service
is
a
peers.
You
go
patent,
so
basically
you
are
renting
resources.
B
If
you
are
on
prime,
you
already
got
resources.
Why
do
you
need
to
do
service?
The
third
thing
is
again
is
because
the
the
service
is
paid
as
you
go.
Obviously
you
are
not
using.
You
don't
want
to
pay
for
it.
So
that's
why
you
need
a
need
to
scale
to
zero,
and
also
you
want
to
start
up
fast
as
well.
B
Okay,
you
might
think.
Okay
is
the
cloud
native
application
service
enabled.
Actually
this
cloud
native
application
can
be
very
fast
started.
Have
you
all
heard
about
raw
vm,
okay,
a
few
of
them
so
grow?
Vm
is
one
way
to
enable
the
like
the
application
startup
fast
and
like
some
runtime,
like
a
quackers
and
halogen,
a
micro
naught
and
trying
to
like
use
the
grow
vm
capability.
However,
there
is
some
downside
with
the
grow.
Vm
is
because
it's
a
hardcore
is
kind
of
almost
like
the
shield
when
used
after
the
build
time.
B
Often
when
you
build
application,
take
a
long
time,
because
it's
a
builder
kind
of
executable
form
and
also
you
couldn't
instrument
anymore.
If
something
goes
wrong,
you
don't
know,
what's
going
wrong,
all
the
atm
tooling
is
no
longer
to
be
inserted
and
then
there's
a
new
technology
up
here.
It's
called
a
linux
cryo,
so
checkpoint,
restore
technology,
checkpoint,
restore
in
user
space,
so
basically
openg9
utilize.
That
and
also
open
liberty
is
trying
to
utilize
the
openg9
the
capability
and
now
openability
like
this
is
the
macro
profile.
B
Application
can
start
in
a
70
millisecond
less
than
100
milliseconds.
Sometimes
it
can
be
even
even
like
a
faster.
The
other
thing
is
there's
a
lot
of
cloudant
like
a
public
cloud
service.
Some
like
technologies,
such
as
I
think
yesterday,
you
might
have
been
to
a
scale,
zero
talk
from
richard
like
ibm
code,
engine,
amazon,
lander,
google
function,
azure
function
and
etc.
All
enable
you
to
run
the
surveys.
B
B
You
need
to
find
your
problem
and
fix
the
problem
if
microservice
is
a
way
to
solve
it
through
microservices
and
the
monolith
is
not
evil
and
when
you
do
to
to
microservices
use
the
best
practices
and
also
learn
them
like
a
cloud
native
apis
to
be
able
to
use
like
like
to
be
to
be
able
to
develop
your
mac
cloud
native
applications.
B
B
So,
finally,
let
me
finish
up
with
the
shameless
plug
about
my
book.
I
wrote
last
year
published
a
book
on
micro
profile,
practical
cloud
native
java
development
with
a
macro
profile,
so
it's
available
amazon
check
it
out
with
that.
That's
the
end
of
my
presentation.
Thank
you
very
much
for
coming
and
support.