►
Description
Are Jakarta EE Servers Dead?
What is the relation between Application Servers and Docker? Are application servers suitable for cloud native environments?
What is the difference between Kubernetes and Application Servers? Is it still viable to run Application Servers on bare metal? Do we need dedicated runtimes?
Which application server’s features become superfluous?
Most of such questions can be answered best with a CLI and a few slides and a lot of live coding. Attendees questions are highly welcome.
///GET SOCIAL!
Subscribe: https://www.youtube.com/user/EclipseFdn
Follow us on Twitter: https://twitter.com/JakartaEE
Like us on Facebook: https://www.facebook.com/JakartaEE/
Join us on LinkedIn: https://www.linkedin.com/groups/13597
Tag us: #JakartaEE
Want more great Eclipse Community webinars? Become a member of our Jakarta Tech Talks Meetup: https://www.meetup.com/jakartatechtalks
A
B
A
A
Okay,
Oh
is
there
I
can
see
Adam
screen
perfect,
okay,
so
just
greeting
again
to
everyone,
it's
fantastic
to
see
all
you
on
the
chat
Adam
is
with
us
today.
I,
don't
think.
Adam
needs
a
particular
introduction,
but
I'll
leave
it
to
him
as
well.
I
just
want
to
welcome
you.
We
are
going
to
record
this
and
it
is
going
to
be
available
for
everyone
for
viewing.
If
any
questions,
please
type
them
in
the
chat
and
I'm
sure,
and
then
we'll
be
more
than
happy
to
answer
them.
B
B
So
this
nice,
you,
you
are
doing
the
marketing,
so
this
is
this
is
obvious.
So
now
Jakarta
is
service.
Our
Jakarta
is
service
dead.
So
I
hear
this
for
almost
fifteen
year
and
I
wanted
to
actually
talk
about
that
or
or
think
about
this
and
I
think
something
like
a
Tech
Talk
is
a
perfect
venue.
So
today
the
camera
is
not
running
justice
to
to
have
more
room
for
my
slides.
So
now,
I
really
like
Java,
still
and
I.
B
Think
in
2019
would
mess
with
my
B's
busiest
Java
years
and
I
do
a
little
bit
also
of
JavaScript,
but
yes
still,
80%,
Java
would
say
or
70%
and
what's
interesting.
So
something
new
were
is
my
podcast,
where
I
interview
interesting
people
like,
for
instance,
Tonya
and
my
blog,
is
more
or
less
like
notepad.
So
it's
just
personal
blog
I
get
lots
of
requests
as
a
guest
articles.
I
rejected
everything,
so
I
just
write
things
not
to
forget.
You
know
in
future
so,
and
it
has
to
be-
is
probably
most
interesting
to
you.
B
What
the
heck's
TV
is
is
like
the
monthly
question
and
answer
show,
so
my
mantra
is
learn
once
never
migrate.
This
is
the
basic
idea.
This
is
what
I
do
for
years.
It's
just
people
ask
me,
know
how
to
how
to
keep
being
up
to
date,
and
I
would
say
you
will
have.
You
know
to
learn
something
and
stick
with
it
for
a
while
and
and
the
beehives
are
coming
and
going,
but
something
like
standart
or
quasi
standard
mostly
survives.
B
So
this
is
what
also
happens
with
Jakarta
em
and
and
also
happens
in
web
standards,
for
instance
right.
So
this
is
the
x2v
this
year
it
starts
at
a
p.m.
not
6:00
p.m.
it
is
easier
for
me
because
of
the
workload
and
yeah.
This
is
a
first
Monday
of
the
month,
mostly
and
8:00
p.m.
you
don't
have
to
register.
You
can
just
attend
on
ask
questions
so
now
some
online
courses
so
about
this
this
session
here
and
Tonya
by
the
way.
How
long
is
the
session
1
1
hour
and
45
minutes?
What's
what's
planned?
Yeah.
A
B
One
hour
one
hour,
perfect,
so
I'm
we
had
a
chat
with
Bruno
Borges
about
micro
profile
in
the
clouds
and
service
and
for
me
Jakarta
he
was
actually
so
from
the
from
the
conceptual
perspective.
Is
this
a
server
less
technology?
Why?
Because
we
actually
don't
care
about
the
service
that
much
so
we
we
and
how
to
think
about
that.
So
in
my
project,
so
people
asked
me
know
what
to
choose
open
liberty
or
payara
or
right
fly
or
or
tommy,
and
I
don't
care
a
lot.
B
So
ok
pick
whatever
you
like
and
just
start
coding,
so
the
application
server,
every
application
server
comes
with
edit
value',
but
the
edit
value'
is
important
for
later.
Not
if
you
just
focus
on
business
logic,
you
know
the
edit
value'
are
like
no
supports
costs
or
specific
features
or
provisioning
or
deployment
or
integration
with
some
SS
O's.
So
this
or
or
cacistis
are
monitoring
stuff.
This
RT
the
the
edit
value',
but
it
doesn't
matter
if
you
start
coding.
B
B
If
you
have
our
War
this
web
archive,
we
can
put
the
war
to
deploy
the
war
to
application,
server
or
somewhere
else,
and
we
would
expect
the
war
to
run
and
if
you,
if
you
look
inside
the
war,
so
what
usually
we
have
there.
This
is
like
a
no
jax-rs
endpoint.
This
is
we
get
posts
and
whatever
was
a
little
bit.
Jason
B
know
to
expose
stuff,
and
it
just
gets
invoked
by
the
application
server,
but
it
could
be
invoked
by
a
cloud
or
whoever
we
don't
actually
care
about
the
application
server.
B
B
So
as
actually
for
me,
it
was
for
me
it's
a
prototypical
idea
of
of
7s
and
even
if
you
think,
about
Google
App
Engine,
which
was
one
of
the
first,
you
know,
prototypes
of
I
would
say
almost
serverless
and
it
was
also
very
similar
to
application
server
so
and
I
had
a
chat
with
Bruno
Burgas
about
that
and
what
I
found
out
is.
This
was
a
really
interesting
conversation
with
him,
because
I
talk
all
the
time
about
the
service
and
he
talked
about
runtimes
and
also
in
the
Jakarta
emailing
list.
B
I
remember
at
the
beginning
of
the
Jakarta
II
stuff,
there
was
no
discussion.
Should
we
rename,
you
know
not
talk
about
Jakarta
servers
rather
than
about
Jakarta
EE,
runtimes
and
I
asked
you
know,
what's
the
point:
what's
the
difference
between
a
server
and
a
runtime
and
I,
don't
even
see
the
difference
or
what's
what's
what's?
What's
the
story
behind
and
and
then
occurred
to
me
that
application
servers
what
it
actually
means
is
it
it
in
the
name
it
implies?
There
are
multiple
applications
on
a
server,
so
a
server.
It
serves
applications
and
runtime.
B
It
implies
one-to-one
relation
between
an
application
and
the
server,
and
this
was
also
true
in
the
discussion
with
Bruno
and
I
thought.
Ok,
this
is
actually
right
so
and
the
problem
is
probably
I.
Now
talk
all
the
time
about
Jakarta,
II
or
Java,
EE
servers
and
people
say:
ok
servers
were
that
and
what
they
probably
meant
is
that
in
the
micro
service
world
there
is
a
one-to-one
relation
between
the
server
and
the
application
and
not
one
to
end
between
the
application,
server
and
the
and
and
multiple
Wars.
B
So
this
is
this
is
the
idea,
and
so
this
was
like.
You
know
interesting
conversation
if
you
like,
you
can
release
into
it,
but
we
cover
this
allowed
a
lot
about
that,
and
also
we
talked
about
cloud
events,
for
instance,
so
Bruno
said:
ok
in
the
cloud
we
get
events
and
they
are
executed
by
a
functions
and
I
say
ok,
but
in
application
server
it
could
happen
the
same
the
same
right.
So
we
have
a
jax-rs
resource
and
this
jax-rs
resource
is
invoked
by
someone.
B
It
could
be
a
cloud
event
or
or
not,
I,
actually
don't
care
who
invokes
my
jax-rs
resource.
Actually,
on
my
server,
this
is
the
Apache
web
server
who
invokes
the
jax-rs
resource,
for
instance,
ok.
So
now,
by
the
way,
this
was
the
announcement
on
Twitter
and
I,
just
retweeted
that
our
application
servers
not
first
reaction
from
marketer.
If
it
was
not,
they
are
not,
and
my
reaction
was
this
session
is
not
going
to
be
that
charge.
B
So
there
is
something
no
to
talk,
so,
of
course
they
are
not
that
otherwise,
you
know
payara
will
just
disappear
from
the
internet,
so
but
it
says
interesting
discussion
to
know
to
what
what
actually
happens
right
now.
So
now
we
have
Jakarta
II
and
we
have
a
micro
profile.
So
we
have
two
specifications
in
as
eclipse
or
two
specifications.
They
are
two
specification
collections,
I
would
say
and
for
me
the
Jakarta
ee
is
like
more
stable,
almost
like
a
operating
system,
so
I
would
expect
to
know
more
more
stability
and
not
too
much
innovation.
B
I
would
say
because
I
mean
the
problem
is
to
know
how
to
innovate.
If
you
look
at
CDI
or
jax-rs,
they
are
almost
you
know,
complete
or
feature
complete
specifications.
Really
it
is
really
hard
to
invent.
You
know
or
to
extend
the
already
existing
CDI
specification
or,
let's
say
JPA
so,
and
there
are
already
major
technologies.
Of
course,
there
will
be
always
some
some
features,
but
it's
not
like
I,
don't
expect
a
huge
revolution
in
Jakarta
e10,
let's
say
so
a
little
bit
different,
it's
the
micro
profile.
B
So
this
is
the
microphone
and
3-1,
and
if
you
look
at
that,
so
in
the
at
the
bottom
there
are
actually
Java
8
specifications
and
the
upper
part
are
actually
more
or
less
cloud
native
foundation
specification
where
we
cannot
control
that
much.
So
what
it
means
is
we
have
if
we
cannot
control
this,
we
need
more
frequent
releases
because
yeah
because,
let's
say
if
matrix
changes,
these
open,
telemetry
format,
we
cannot
control
that.
So
we
have
to
know
to
keep
up
with
the
non
Java
communities.
B
This
is
not
about
Java
exactly
the
same
with
chase
web
token.
Of
course
rest
client.
We
control
it
completely
configure
all
over
that
open
a
open
API,
not
that
much
open
tracing
and
fault.
Tolerance
is
also
a
set
of
patterns
which
we
could
actually
implement,
and
the
bottom
part
is
esteem.
I
would
say
the
integration
with
Jakarta
II.
So
now
what
is
that
this
is
a
18k.
This
is
one
of
the
slides
of
from
my
micro
service
workshop
and
what's
interesting
here,
as
you
can
probably
see
it,
this
slide
is
fairly
old.
B
So
if
on
single
machine
costs,
1
million
and
I
was
in
around
the
year
2000
a
project,
we
had
two
such
machines
in
a
cluster
so
to
ethan
cage
was
insurance
company
and
you
can
imagine
we
have
2
million
euros
in
two
machines,
so
the
application
servers
were
installed
on
such
machine
and
the
main
idea
was
to
insert
as
many
applications
as
possible
on
a
single
application
server
and
provide
isolation
for
the
applications
to
keep.
You
know
the
hardware
costs
low
per
project,
so
this
was
the
main
goal
of
application
service.
B
So
the
problem
with
this
is
the
hardware
becomes
cheaper
and
cheaper.
So
right
now,
I
think
the
same
machine
would
cost
like
20
euros
was
the
same
power.
It
is
probably
I
assume,
so
the
18k
is
probably
a
little
bit
faster
than
Raspberry
Pi
I
guess
so
you
could
have
no
equivalent
machine
for
30
30
euros,
and
the
question
is
no:
what's
the
point
of
application
service
and
the
remaining
point
from
my
perspective
is
ease
of
use,
so
what
happens
over
the
years?
Was
we
got?
B
So
if
you,
if
you
take
a
look
a
little
bit
on
the
history,
so
my
impression
so
I
follow
this
a
little
bit
at
the
beginning
or
I
had
to
follow
because
I
was
in
projects.
So
everything
started
with
JDBC,
which
was
a
part
of
JDK.
Actually,
and
my
impression
was
like
it
was
a
little
bit
driven
by
Oracle
and
the
Java
database.
Connectivity
was
similar
to
ODBC
back
then,
and
the
idea
was
to
abstract
from
databases.
Then
Jane
di
came
in
Jane
di
Java,
naming
and
directory
interface.
B
It
was
like
abstraction
from
from
directory
service
like
LDAP
and
Novell
was
behind
back
then
NDS
novel
Novell
directory
services,
huge
huge
things
in
the
2000s,
was
like
an
OD
ability
to
look
up
a
printers
in
the
network.
So
it
was
a
huge
deal
back
then,
and
the
GMs
came
from
IBM
similar
story.
We
wanted
to
have
a
abstraction
to
talk
to
MQ,
serious
or
web
cmq,
and
we
we
got
JMS.
So
over
the
years
we
got
abstractions
for
for
for
many
different
products
and
concepts,
and
so
and
we
get
everything
at
once.
B
I
say
Java,
EE,
API
or
now
Jakarta
API,
and
this
is
convenient
for
me,
because
if
the
project
starts,
I
can
just
you
know
download
that
they
think
and
just
you
know,
keep
going
so
this
is
this
is,
what's
so
I
would
say
the
the
D
role
of
applications
of
a
change
to
be.
You
know
a
a
money
saver
just
to
save
hardware
costs
and
we're
a
usability
actually
didn't
met
at
all
at
the
beginning.
Actually,
no
one
cared,
whoever
the
application
servers
are
usable
or
not.
B
There
were
complex,
but
no
one
cared
and
actually
before
j2e.
There
was
actually
no
way
to
deploy
an
application
to
different
application
service
at
the
same
time
because
they
were
completely
different.
So
there
was
a
no
persistence
power,
tier
Paul
and
jbuilder.
There
was
a
gemstone.
There
was
a
Tenga
server
which
became
ba
later
and
they
were
complete
different
servers,
so
there
was
actually
no
way
to
to
install
to
deploy
something
to
multiple
servers.
B
At
the
same
time,
there
was
not
even
in
a
WebEx
ml,
there
were,
there
were
no
servlets
and
I
was
one
of
the
involved
in
one
of
the
first
application
servers
was
called
the
Java
web
server
from
Sun
and
back
then
there
was
no
website
what
Peaks
ml,
so
there
was
actually
no
way
to
configure
the
service
properly.
So
you
you
had,
you
know
to
put
one
char
and
and
and
this
was
a
deployment
of
of
your
application
to
to
the
application
server.
B
Okay,
so
now
over
time,
of
course,
the
application
service
became
more
and
more
usable
and
the
goal
shifted
to
be.
You
know,
like
the
only
one
operating
system
for
all
applications
inside
a
company
into
a
runtime
which,
which
became
also
smaller
over
time
because
it
gets
you
know
over
the
course
of
20
years.
It
got
to
know
optimized
over
and
over
again
and
of
course,
and
what
also
happened.
The
vendors
who
know
started
to
compete
with
each
other.
So
there
was
a
white
flathead
modules:
thin
glass,
which
became
OSGi
server.
B
B
Cool
so
I
just
to
take
a
look
now
what's
in
the
chat,
so
everything
on
the
track.
So
this
is
one
of
my
of
my
slides
from
the
micro
services,
and
this
was
like
one
of
the
best
practice
slides
and
what
I
said
is
we
have
a
self-contained
worse
one
operating
system,
one
server,
one
war,
no
dependencies,
no
jars.
So
this
is
a
like
my
default
slide.
This
is
how
to
run
how
to
run
micro
services
in
or
outside
clouds
or
clouds
do
not
matter
a
lot
and
by
accident.
B
B
So
this
was
the
hard
part
and
what
I
wanted
to
do
is
I
wanted
to
migrate
from
Linux
to
open
Solaris,
and
why?
Because
Solaris
had
containers-
and
the
idea
was
you
know
to
pack
to
to
do
to
to
to
put
my
application
service
to
containers
and
then
just
you
know,
export
them
somehow
as
a
cars
and
and
run
them
in
the
new
opensolaris
server
and
the
problem
was
back
then
Oracle
bought
Sun
and
opensolaris
was
not
option
anymore,
but
at
the
same
time,
I
just
searched
for
a
donor
for
containers
and
I
found
Dockers.
B
Okay.
This
is
not
opensolaris,
but
at
least
something
similar
and
I
actually
use
docker
just
for
Beca
purposes.
So
the
idea
was,
you
know
you
can
you
can
just
save
the
whole
container
sitar
and
if
the
new
survive
of
a
new
server
have
moved
everything
to
a
new
server
or
even
back,
then
what
I
did?
The
server
was
empty.
I
developed
everything
on
my
development
machine.
B
Just
pushed
everything
to
the
service
launched
everything
and
it
run
so,
and
what
I
recognized
back
then
that
the
application
service
actually
I've
worked
perfectly
well
with
with
docker
and
and
and
why
because
properly
of
the
slide?
Because
if
you
have
you
know
if
your
entire
application
runs
in
in
a
in
in
a
war
and
the
war
has
no
dependencies,
the
war
is
tiny,
because
the
entire
API
with
the
implementation
lives
already
on
the
server
and
if
it
lives
on
the
server.
There
is
nothing
to
ship,
because
everything
is
on
the
server
as
I
said.
B
And
by
the
way,
the
Jakarta
archetype
is
in
maven
central,
so
sometimes
it
takes
a
week
or
two
to
to
index,
but
now
I
double
check.
Then
it
should
be
available
for
everyone,
so
the
name
of
the
archetype
is
Jakarta,
essential,
archetype
and
I
will
switch
to
tech,
talk
to
tech,
talk
and
just
open
the
editor,
and
what
you
should
see
is
this
is
what
we
talked
about
today
is
the
Jakarta
e8
dependency,
and
this
is
the
micro
profile,
but
this
is
just
provided
in
meters
already
on
the
server.
B
Then
you
will
see
that
the
Tech
Talk
war
should
be
tiny
and
of
course
this
is
simplistic
application,
so
it
is
4k.
It
is
tiny,
but
in
my
projects
more
commercial
projects,
they
are
usually
still
tiny,
I
would
say
inverse
case
a
couple
of
megabytes,
so
there
is
not
usually
under
one
MEK
and
why?
Because
the
Java
bytecode
is
pretty
small.
So
now
we
did
this
and
now
watch
that.
So,
if
I
convert
this
to,
let's
say
payara,
why
not
and.
B
B
So
done
so,
this
was
the
entire
docker
creation,
but
so
what
I
usually
do
is
I
to
show
the
Jakarta
EE
Tech
Talks,
Tech
Talks
and,
as
you
can
see,
the
magic
is
that
the
most
upper
image
are
4k
and
it
yeah
it's
just
4k.
This
is
why
it
was
so
fast
and
all
other
layers
are
ancient.
So
there's
like
no
30
months
and
the
this
was
three
years
so
three
years
I
created
the
base
image
which
is
properly
the
center,
a
seven
I
guess
then
I
installed
Java,
then
payara
was
refreshed
thirteen
months
ago.
B
So
I
was
pretty
lazy
in
this
case,
but
you
can
see.
The
uppermost
image
is
is
my
microservice,
and
this
is
what
matters,
because,
usually
regardless
private
or
public
clouds,
we
already
have
the
whole
application
server
in
the
private
registry
somewhere
in
the
clouds.
So
it
could
be.
You
know,
on
OpenShift.
This
is
the
private
register
and
openshift
in
cuban
it
is
you
have
your
own
registry.
It
could
be
in
my
enterprise,
project,
Nexus
or
j-rok
or
or
hardboys
also
interesting
registry,
and
the
point
is
so
this
most
upper
layer.
B
This
is
the
thing
I
would
say
this
is
actually
what's
what
I
found
out.
This
is
actually
the
key
to
be
cloud
native,
because
if
something
changes
I
only
ship
this,
what
changes
and
everything
else
remains
stable
and
also
interesting,
is
the
following.
So
if
you
run
this
in
production,
you
know
in
this
case
I
guess
it
is
payara
five,
one,
nine
three.
So
what
I
can
do
you
know?
I
can
buy
support
from
the
payara,
guys
and
I
say
hey
what
is
actually
the
libraries
inside
five
one,
nine
three
and
I
get
a
list.
B
The
question
is
know
what
is
inside.
There
is
no
clean
separation
between
business
logic
and
infrastructure,
so
as
way
harder
to
know
what's
inside
so
I
will
have
to
use.
There
are
lots
more
labels
and
annotations
and
Cuba
natives
are
open
shift
to
find
out
what's
actually
inside,
which
is
already
a
issue
in
some
projects
in
in
production
in
DevOps.
So
this
is
like
you
know
my
way
to
docker
as
an
application
service,
and
this
is
the
accidental.
B
You
know,
enlightenment,
actually,
application
servers
accidentally
because
we
have
the
deployment
process
are
perfectly
suitable
for
Dhaka
environments
and
so
of
course,
in
Cuban
it
is,
and
if
something
runs
in
Cuban
it
is,
it
is
actually
appropriate
to
run
well
in
all
clouds.
So
this
is
this
is
the
main
idea?
Ok,
nice,
so
we
did
that.
Of
course,
if
you
are
started,
will
it
start,
of
course,
and
by
the
way,
because
I
mentioned
that
so
what
I
usually
do
before
a
project
I
reinstalled
payara?
B
Now
my
payara
recent
one
I
think
is
one
nine
three
or
four
or
I
can
do
the
same
with
a
white
fly
and
what
happens
then
is
just
you
know?
The
old
server
is
deleted
and
the
new
one
is
installed.
So
this
I.
Do
it
all
the
time,
because
it's
so
easy,
so
I
would
say
so.
The
main
goal
of
application
service
became
ease-of-use,
usability
and
and
I
can
focus.
B
My
applications
of
my
code
becomes
simple
and
simpler
and
I'm
working
a
lot
with
with
startups
and
what
I
tell
me
the
lecture
is
is
very
easy
to
work
with
application
service
and
one
funny
question
or
a
question.
I
yeah,
question
I
get
is
why
the
application
servers
are
so
fast
there.
They
never
optimized
anything
and
they
just
one
fastest.
Okay,
okay,
because
if
you
focus
on
business
logic,
usually-
and
you
don't
write-
you
know
complex
code-
you
don't
try
to
optimize
things.
Usually
it
runs
fast.
Ok,
so
we
cover
that.
B
So
what's
the
issue
here,
so
the
question
is
no
our
application
service
that
right.
So
the
problem
is
what
I
see
is
the
multiple
wars
in
an
application.
Server
is
no
more
a
feature,
so
in
90%
of
all
times
there
is
one-to-one
relation
between
application,
server
and
the
war.
So
what
it
means
is
I,
don't
care
about
the
deployments
that
much
anymore
so
for
me,
is
very
important.
B
It's
the
clean
separation
between
business
logic
and
infrastructure,
but
whether
the
application
server
can
you
know,
isolate
one
war
or
50
I,
don't
really
care,
because
in
eighteen,
ninety
percent
of
all
cases
I
always
have
one-to-one
relation
between
both
okay.
So
this
was
actually
the
the
idea
from
these
slides
and
now
take
a
look.
What.
B
So
so
what
what
did
what
it
basically
means
is:
what
is
what
is
the
future?
The
future
in
my
eyes,
is
we
get
something
similar
to
let's
say
hollow
jars
with
with
pyar
a
micro
or
prior
on
tale
or
two
something
similar
to
Haledon
or
Quercus?
Why
that,
if
you,
if
you
take
a
look,
let's
say
on
Quercus,
just
to
compare
Quercus,
which
is
very
recent
runtime
and
compared
to
jakarta
e,
so
if
I
would
launch
the
a
quark
UCF-
and
this
is
the
old
one,
this
is
one
one
one
and
let's
say
a
hex.
B
So
now
fixed
now,
what
do
we
have
here?
Just
to
compare
that
with
the
Civil
War?
We
have
a
runner
which
is
170
K,
so,
as
you
can
see,
that
jar
is
small,
but
not
as
small
as
the
war.
Why
that,
because
this
contains
this,
contains
the
main
method
and
a
little
bit
of
bootstrap
code,
as
well
as
my
application
logic,
and
by
the
way
the
application
logic
is
also
tiny.
B
So
if
you
take
a
look
on
that,
it
is
very
similar
programming,
more
the
same
programming
model
like
Java
but
different
runtime,
and
if
we
take
a
look
on
that
main
Java
here,
so
this
is
like
a
jax-rs.
There
is
actually
no
difference
from
from
the
programming
perspective,
and
so,
if
you
take
a
look
on
that,
this
is
the
one
runner
which
is
executable.
So
there
is
no
more
application
server,
so
you
can
say:
ok,
you
preach.
You
know
one
hour
and
you
talk
about.
B
You
know
how
great
it
is
that
we
have
the
separation
between
infrastructure
and
business
logic,
but
if
we
take
a
look
on
the
lip
target
lip,
this
is
where
all
the
libraries
reside.
So
the
interesting
part
is
what
happens
here.
Is
we
have
still
clean
separation
between
the
business
logic
and
and
the
infrastructure
so
of
what
is
very
easy?
You
can
just
pack
all
the
and
all
the
dependencies
to
a
base,
docker
image,
and
then
you
know
ship
several
times
a
day,
just
the
jar,
so
it
happens
so
it
happens.
B
B
A
B
So
this
Takuma
loose
also
a
similar
framework
which
supports
micro
profile
and
also
micro
profile,
plus
jakarta
a
technology.
So
there
is
a
both
are
supported
and
the
interesting
part
in
this
frameworks.
So
what
they
do
is
they
do
like
you
know:
application
server,
plus
plus
that's
ok.
If
there
is
only
one
relation,
we
have
still
the
separation,
we
cannot
ship.
You
know
two
applications
per
on
time
anymore,
because
no
more
necessary,
but
but
but
there
is
no
deployment
at
runtime.
So
what's
the
deploy?
What's
the
problem
with
deployment
at
runtime?
B
Actually
not
not
under
huge
problem,
except
this
is
unnecessary,
unnecessary
complexity.
So
what
the
application
servers
have
to
do?
They
have
to
scan
a
folder
and
the
folder.
You
know
what
usually
happens
disk
and
the
folder.
They
find
a
war.
They
read
the
war,
the
the
decompress,
the
war
and
pair
reflection.
They
read
all
the
stuff,
but
if
the
application
server
knows
ok,
there
is
only
one
application.
They
can
already
read
the
application
at
the
build
time.
B
There
is
no
deployment
involved
and
this
makes
sense
for
specific
environments
like,
for
instance,
micro
services
or
micro
services.
Like
you
know,
a
small
small
applications
or
clouds,
and
what
the
application
service
or
UG
newer
runtimes
begin
to
do
is
like,
for
instance,
Quercus.
They
say:
ok
if
there
is
no
more
deployment
at
runtime,
we
can
deploy
at
Build
time
and
they
try
to
optimize
more
and
more
at
Build
time.
What
I
get
out
of
that
is
because
there
is
no
more
reflection
at
runtime
and
lots
of
you
know
stuff
going
on
at
build
time.
B
They
can,
for
instance,
check
whether
the
dependency
injection
will
work,
because
if
there
is
no
unsatisfied
dependency,
error
or
ambiguous
dependency
error,
they
will
find
that
the
bill
time,
there's
there's.
No,
that
you
don't
have
to
wait
at
one
time
and
the
run
times
become
even
smaller,
because
the
whole
deployment
machinery
is
no
more
necessary
at
runtime.
So
this
is
what
starts
to
shift
so
I
would
say
the
the
name
application
service,
I,
guess
it
is
that
will
disappear,
but
we
will
get
a
similar
or
exactly
the
same
technology.
B
You
know
and
name
differently,
so
there
will
be
no
player
a
server
at
Empire,
a
runtime,
probably
or
Wi-Fi
runtime
or
Wi-Fi
gets
replaced
with
Quercus,
but
as
you
can
see
print
in
the
Quercus
case,
we,
this
is
a
lots
of
you
know,
nettie
Quercus
and
if
you
white
fly
so
this
is
actually
like
the
white
flight
technology
repackaged
and
re
implemented
differently.
So,
okay
I,
would
just
take
a
look
at
the
chat.
B
B
So
Josh
I
said
sorry
to
ask
you
a
silly
question
is
being
a
while
I
have
looked
at
J.
So
if
you
ask
me,
questions
please
are
short
as
possible.
Otherwise,
to
read
it
reread
them,
and
he
said
you
can
we
can
you
can
we
can
separate
the
business
logic
from
dependencies
before
it
was
harder
also
before
I.
Think
no
I
mean
how
to
do
that
right.
B
So,
with
application
servers
with
deployment,
it
is
a
clear
cut,
so
I
mean
there's
a
war
there's
no
dependence
of
application,
servers
and
and
the
DD
war
is,
is
explored
at
a
deployment
time
without
a
war.
It
is
some
way
harder
to
do
to
be
cleaned.
What
I
remember
is
the
OSGi
days
back
then.
It
is
a
know.
Most
of
the
project
were
not
that
clean,
because
there
are
always
tricks
like
body
class
loading.
No
to
so
I
think
it
is
really.
War
was
actually
a
huge
success
in
my
eyes.
B
So
what
the
nineteen
is
in
the
pipes
with
food
micro
professed
support
here
and
why
fight
eighteen
also
comes
with
partial
microprocessor
port
payara
comes
with
full
micro.
Professor
ports
is
perfect
and
by
the
way,
I
use
paya
all
the
time.
It's
not
like
I,
say:
okay,
it's
not
usable.
So
what
I'm
just
saying
is,
like
you
know
in
my
last
payara
of
projects,
there
was
always
you
know,
one-to-one
relation
between
pyre
and
war.
It
was
the
same
in
whitefly.
B
Ask
question
so
someone
asked
me:
are
you
think
that
application
server
in
Dhaka
is
a
good
idea
for
production
usage
in
the
real
world?
We
need
to
set
up
application
server
only
once
and
then
just
update
deployment
yeah,
but
it's
the
whole
point
with
application
service.
It's
a
great
story
is,
let's
see,
wait
a
second,
let's
say
whether
I
managed
to
do
that.
B
A
registry,
okay,
doc
Karimov
remove
registry.
So
now
I
would
like
to
try
to
answer
the
question
with
some
code.
So
now
the
docker
registry
is
running
and
docker
images,
so
there
was
one
image.
This
is
this.
One
I
will
have
to
take
that
so
doc
attack
this
one
with
one
localhost
localhost
five
thousand,
and
this
I
think
is
right,
and
then
it
was
a
little
bit
of
luck.
Let's
see
what
it
works,
it
doesn't.
Work
always
depends
on.
The
network
took
a
push
to
that
cool.
It
works.
B
So
what
I'm
doing
right
now,
I'm
actually
deploying
my
application
server
to
imaginary
cloud,
so
it
is
Klaus
runs
on
my
machine,
but
when
doing
this
a
remote
push
to
doc
container,
so
the
question
is:
is
adverse
to
use
docker
and
the
application
server
do
not
change
all
the
time.
So
now
the
cool
story
is
so.
If
I
go
to
the
Tech
Talk
again
and
perform
some
change.
Oh,
this
was
mistake
from
button
to
the
same,
let's
say:
I'm
in
the
talk
to
a
Tech,
Talk
and.
B
B
Now
it's
built
with
the
different
tank
and
then
let's
push
it
again,
as
you
probably
saw
it
took
like
you
know,
ten
milliseconds,
to
push.
Why?
Because
docker
recognized
you
know
there
is
only
the
last
layer
changed
and
only
what
change
was
pushed
is
very
similar
to
to
get
or
if
you
like,
blockchain,
almost
right
so
yeah
that
that's
the.
B
That's
the
difference.
So,
let's
see
what's
what
else
so
on
it's
a
very
good
idea
and
I
would
say:
80%
of
all
my
clients
are
running
application,
servers
in
docker
out,
say:
90%
I,
actually
can't
remember
anymore,
who
runs
application
service
on
bare
metal
I.
All
my
projects
are
running
in
docker
and
I
would
say,
like
50%
of
all
projects
are
running
on
Cuban.
It
is
like
environment,
most
of
them
openshift
and
the
others
in
the
clouds.
B
Next
question
is
from
phone:
what
do
you
think
of
running
service
on
AWS?
You
sure
I
mean
actually
at
AWS?
What
do
you
mean
by
that
is
ec2
or
eks
works
perfectly?
It
is
this
elastic
kubernetes
service.
So
this
you
can
do
this,
so
it's
similar
to
OpenShift
and
Google
cloud.
A
little
bit
more
convenient
I
would
say
and
Microsoft
Asia
a
little
bit
more
popular
and
the
question
is
lambdas
lambda,
this
server
less
or
function
as
a
service
and
the
funny
story
is
in
a
view
project.
B
We
find
out
that
running
application
server
all
the
time,
so
first
application
service,
the
runtime,
is
smaller
than
you
may
think,
and
at
this
particular
project
it
was
cheaper
to
run
payara
all
the
time,
then,
because
that
traffic
was
high
then
used,
for
instance,
the
lambdas
from
from
AWS.
So
this
is
the
question
the
answer
to
this
question.
So
next
one
Arion
asked
or
asked
is:
is:
is
it
impossible?
It
should
be
retired,
so
I
think.
Is
it
possible
that
it
should
be
you
retired,
yeah
I?
B
Think
we
first
my
personal
opinion
is:
we
should
retire
as
much
as
possible
and
before
we
retire,
ejbs
I
would
retire
soap
first.
This
would
be
my
absolute
favorite,
so
it
should
just
disappear
from
Jakarta.
We
can
keep
it
in
Java
8
and
should
disappear
from
Jakarta,
so
soap,
this
xml-rpc
and
all
the
stuff.
We
don't
need
it.
So
the
next
one
is
easy
tones.
It
should
be.
I
think
it
should
be
is
was
never
a
problem.
B
We
have
our
Ian
from
payara
I,
guess
that
you
know
their
entire
GB
container
on
pi
RS
like
one
Mac,
so
there
is
actually
non
non
issue
with
with
agb's,
but
we
should
retire.
A
job
is
because
Cydia
is
good
enough.
So
what
we
should
should
include
all
the
annotations
from
eg
B's.
We
should
no
pull
over
to
CDI
and
then
and
then
we
are
good
to
go
so
the
more
specification
we
can
kill
the
simpler
and
more
useable
jacket,
Jakarta.
He
becomes
and
I
really
like
it
to
be.
B
But
if
CDI
offers
me
the
same
or
rotations
I,
actually
don't
don't
care
whether
they
come
from
Java
X,
it
should
be
or
Jakarta
Jakarta,
Enterprise,
EDI
or
whatever
yeah
and
answers
exactly
that
was
allowed,
or
probably
it
would
be
good
idea
to
have
in
our
pool
able
CDI
beans
also
I
have
to
say,
with
with
micro,
profit
for
tolerance
and
bulkheads.
We
have
something
similar
already
in
in
in
micro
profile,
so
message
driven
bean.
B
This
also
can
be
a
CDI
replacement,
yes,
and
even,
for
instance,
in
in
micro
profile,
it's
called
it's
called
reactive
messaging.
I
think
we
are
very
similar
without
going
on
incoming
rotations
to
message
driven
beats,
so
Google
flouts
Alexander,
says
yes,
Google
Klaus
is
even
more
usable
in
AWS
in
my
eyes,
so
yeah
a
cool.
B
So
other
questions
so
Josh
your
question
was
not
silly
at
all.
Was
good
one.
It's
too
long!
Sorry
for
your
long
questions,
you
have
to
say
so
pipe
line.
19
the
pipes,
someone
from
from
red
head
told
me:
okay,
no,
the
micro
prefer
to
support
is
around
the
corner.
This
was
like
November,
so
I
would
say
right
now.
I
would
expect
you
know
micro,
Wi-Fi,
Wi-Fi,
delivering
micro
profile
already
so
now
what
else
question
I
can
answer.
B
So
maven
clean
install
is
my
bad
habit.
Yeah
I
know
that
I,
don't
trust
me
even
a
lot
I
had
already
some.
You
know
some
discussions
with
Maidan
committees
like
Robert,
alter,
for
instance,
but
yeah,
my
way
of
going
sure
maven
Peck.
It
should
be
enough,
sorry
for
that,
for
a
clean,
install,
a
running
joke
actually.
B
B
Okay,
so
what's
what's
the
gist
from
the
from
the
whole
story
is
I
would
say
in
my
opinion,
so
the
name
server
is
problematic
because
people
expect
server
like
serving
applications,
so
multiple
applications
from
one
server
and
by
the
way
as
I
remember
it
was
I
think
it
was
even
the
same
company
with
which
ran
the
Eaton
case
back
then,
and
in
one
point
of
time
someone
said
we
get
an
application,
server
and
I
say:
ok
and
someone
else.
You
know
from
the
department
with
non
Java.
B
Programmers
like
you
know,
cobol
and
pl/I
one
and
and
someone
has
ok
great.
Then
we
can
install
vert
on
the
application
service
acorrea
what
it.
What
you
would
like
to
do
is
yes,
yeah.
The
idea
was
back
then,
the
day.
Instead,
you
know
the
worst
vert
or
Microsoft
Excel
on
the
application
server
and
they
will
just
magically
be
distributed
to
the
entire
companies.
It
was
the
very
first
idea
of
application
service
so
and
now
I'm.
So
therefore,
I
know
the
imagination.
B
If
it's
Rama
says
application
server
application
server,
it
means
okay,
they
are
multiple,
so
this
is
problematic.
So
I
would
say
from
this
point
of
view:
application
servers
you
know
are
conceptually
that
it's
a
problem
with
the
application
service
or
the
feature
of
application
servers.
They
are
just
you
know,
very
well,
optimized,
very
good,
runtimes
and
very
usable
runtimes.
B
So
now
what
about
in
Jakarta?
Is
there
any
plans
for
testing
I
mean
I
mean
testing
is
a
soft
problem,
so
you
can,
you
know,
launch
all
these
specifications
separately
and
we
had
lots
of
discussion
over
around
testing
I
even
recorded.
The
entire
course
about
testing
and
testing
I
would
say
is
soft
issue.
What
usually
the
only
problems?
Testing
are
the
statistics.
B
So
most
of
my
clients
try
no
to
unit
as
everything
to
get
nice
statistics,
but
what
really
counts
in
such
projects
is
more
I
would
say,
system
test
blackbox
tests
and
then
what
usually,
what
we
do
in
testing
is.
We
have
a
pipeline
like
Jenkins
pipeline,
and
you
know
the
application
gets
deployed.
We
wait
until
readiness
probes
are
green
if
they
are
green.
We
just
launched
our
system
test,
which
is
written
in
unit
tests.
J
unit
and
jax-ur
is
client
and
we
test
the
application
from
outside.
B
So
now,
Majid
Mustafa
asked
me
it
should
be.
Performance
is
better
than
CDI.
How
does
CDI
make
performance
better,
so
first
CDI
gets
more
and
more
optimized
and
in
run
times
like
Quercus,
for
instance,
it
doesn't
matter
anymore
because
they
don't
use
reflection
for
for
dependency
injection,
so
the
performance
of
request,
copped
injection,
is
almost
as
good
or
I
didn't
measure
that
but
I
said
I
actually
expect
to
be
as
good
as
the
performance
of
eg
B's.
B
What
about
running
Tomcat
on
docker?
So
Tomcat
is
an
interesting
one,
because
so
funny
story
with
Tomcat
I
think
was
also
5
to
10
years
ago
was
a
project
in
Switzerland
and
they
say:
ok,
we
have
a
light
way
to
run
time
because
they're
running
Tomcat,
ok
and,
as
you
probably
may
know,
Tomcat
is
actually
a
small
lightweight
runtime.
B
It's
a
problem
with
Tomcat
is,
of
course,
there
is
just
solid
API
on
jsps
on
its
own,
nothing
else
and
the
war
back
then
was
500
geeks,
so
they
ever
no
sorry
1505
and
it
keeps
no
not
500
keys.
500
makes
half
a
geek
exactly
half
a
geek,
it
was
incredibly
large
and
it
took
forever
to
deploy.
So
what
happened
with
Tomcat
begged
dentist,
you
know
that
the
runtime
was
small,
but
the
war
was
larger
than
the
application
server.
B
All
which
is
crazy
in
my
eyes,
so
the
deployable
should
be
smaller
than
the
runtime
and
hit
me
Tomcat.
It
was
larger,
so
I
never
actually
I
never
got
the
point
of
Tomcat
or
jetty
back
then,
because
for
me
the
simplicity
was,
you
know,
have
larger
runtimes,
a
usable,
runtimes
and
ship
smaller
apps,
my
testing
presentation.
If
you
go
to
a
xio,
there's
Java
EE
testing
cores
like
50,
you
know
50
50
sessions
just
about
testing
and
on
my
youtube
channel
you
will
also
find
lots
of
testing
stuff.
B
B
It's
a
problem
with
what
I
can
tell
you
is
so
average
application
server
you
can,
with
average
application
server,
you
can
have
several
thousand
transactions
per
second
and
recently,
if
you
go
to
my
youtube
channel
I,
measured
quark,
whose
vessel
whitefly
and
performed
a
small
benchmark,
and
there
were
5,000
transactions
per
second
right
away,
and
this
is
what
I
also
expect
or
see
in
projects.
This
is
the
performance.
B
Exactly
it
totally
makes
absolute
sense,
so
tell
me
where
I
could
actually
run
Tommy
in
a
docker.
Just
running
Tomcat
is
pointless
in
my
eyes
so
which
micro
profile
capable
application
server.
Do
you
recommend
preferred
the
good
news?
Is
all
application
servers
except
to
hight
fly,
which
is
funny
actually
because
they
have
Quercus,
they
have
smaller
I.
It
was
also
a
retina
for
than
the
application
servers,
not
fully
micro
profile
capable
so
I
good
I
made
very
good.
A
very
good
experience
have
very
good
experience
with
payara,
so
yeah
lots
of
payara
projects.
B
The
reason
being
is
I
did
lots
of
GlassFish
back
then
then
open
Liberty
is
just
great.
So
it's
very
fresh,
very
modern,
very
fast
application.
Server,
open
Liberty,
then
you're
asking
about
application
service
Tommy.
It's
also
nice
and
white
fly
with
some
smaller
eye.
Patches
is
also
great
and
we
all
waiting
for
for
white
try
19
seems
like
and
by
the
way,
hire
a
micro
I.
Never
you
spire
a
micro.
B
My
problem
was
the
pyre,
a
fool
who
was
small
enough
and
because
payara
fool
is
based
on
GlassFish
I
expect,
or
my
idea
was,
if
you
just
used
by
our
food,
we
will
find
more
documentation
and
we
will
find
more
answers
to
problems
even
from
GlassFish
days.
This
is
why
I
always
proposed
by
are
a
fool
and
always
aspire
a
fool,
and
it's
good
enough.
A
B
Cool,
so
if
you
have
just
room
any
questions,
ask
me
otherwise:
I
will
wrap
up
so
what
I
actually
I
could
code
something
but
I
code,
usually
all
the
times
we
just
end
up
having
another
hello
world,
but
the
the
says
you
has
a
question.
Okay,
what
is
the
our
experience?
Who
your
experience
using
Jakarta
in
combination
with
openshift
is
actually
great.
So
what
it
means
right
now
is
if
I
do
payara
wait,
a
second
so
fire
as
to
I,
and
it's
called
it
Tech
Talk.
B
B
B
It
took
a
little
bit
longer
than
expected,
so
I
was
a
little
bit
unsure,
but
now
the
tech
talk
is
deployed
with
spire
a
full-on
openshift
I
use
it
all
the
time-
and
you
see
here
the
Jenkins
ephemeral
if
I
would,
if
I
would
prepare
myself
for
the
better
for
the
talk,
I
will
show
you
a
pipeline
running
on
it,
and
so
the
experience
is
just
great,
so
you
can
run.
We
run
I've
hired
a
fool,
usually
what
I
run
on
OpenShift
and
pyar
a
micro.
B
B
As
you
can
see,
this
is
the
modified
version
Jakarta
in
9th
wonder
ker
and
on
openshift
and
yeah.
The
experience
is
just
great.
Why
was
so
fast
because
the
payara
was
already
installed
in
the
in
the
private
docker
registry
in
on
openshift.
So
if
we
go
to
the
open
shift
and
go
to
images,
you
see
the
payara
s2i
5
days
ago,
so
was
already
installed.
B
B
Have
to
scroll
you
see,
so
the
application
servers
are
easier
to
use.
Then
then
the
chat
software
for
me
so
now
go
from
so
this
was
this
one,
so
we
covered
hopefully
from
Martin,
so
the
expenses
of
Jakarta
and
whitefly,
so
OpenShift
comes
with
white
Flyers
to
I
as
well,
so
out
of
the
box
and
yeah
it
works
perfectly.
So
this
is
what
I
can
tell
you
and
with
of
course,
mini
shift
is
even
mini
shift
micro
profiles
even
better,
because
we
get
readiness
and
and
liveness
probes
and
with
both
probes
they
are
crucial.
B
You
know,
readiness
pros
in
order
the
OpenShift
has
to
find
out
when
the
application
is
ready
to
serve
them
by
the
way.
This
is
one
of
my
criticism
of
normal
application
service.
I
would
say,
because
if
you
deploy
application
server
to
openshift,
they
become
ready
and
then
the
war
gets
deployed,
so
it
was
fixed
in
micro
profile.
But
at
the
beginning
the
application
server
said
I
was
ready.
B
Also,
the
application
was
not
deployed
right
yet
and
if
you
go
in
the
with
the
runtimes,
like
no
quirk
ooh
salad
on,
there
is
no
more
separation,
so
either
everything
is
deployed
or
nothing
yeah,
which
micro
profile
capable
application
server.
You
recommend
actually
all
application
servers
except
whitefly,
because
we
have
to
wait
until
white
line
19.
All
others
are
really
greatly
supported
and
you
saw
the
Jakarta
ie,
API
or
Jakarta
a
project.
What
I
brought
with
the
archetype
I
always
use
no
Jakarta,
8
and
micro
profile.
B
I
never
had
just
a
micro
profile
project
or
just
Jakarta.
A
project
doesn't
make
sense
to
separation.
I
always
use
both
so
next
question.
What
about
the
new
micro
profile?
Reactive,
messaging
spec?
It
was
not
included
in
the
last
microphone
update,
it
was
not
included,
but
we
already
use
it
in
in
one
Quercus
project
and
what
about
reactive
messaging
I
mean
if
your
application
is
already
reactive,
it
makes
sense
if
you're
using
reactive
sources,
for
instance,
in
one
project
we
have
to
use
Kafka,
so
it
makes
absolute
sense.
B
You
know
to
read
from
Kafka
topic
and
transform,
for
instance,
the
messages
to
words.
Okay,
so
this
reactive
programming
is
great,
but
if
you
just
you
know,
have
a
normal
database
with
normal
JDBC
driver
and
the
normal
JavaScript
app,
it
doesn't
make
to
be
reactive
at
all,
so
how
you
prefer
to
share
state
in
case
not
stateless
application,
so
in
case
the
application
is
not
stateless.
The
question
is
no.
What
is
your
state?
B
Is
your
HTTP
session,
probably
not
because
we
don't
need
HTTP
session
anymore,
except
you
have
JSF
and
in
this
particular
case,
if
the
database
is
fast
enough,
storage
in
database,
if
not
probably
in
something
like
in
finish
spending
kinds
of
white
fly
or
hazel
Carson
cans
in
case
of
payara.
That's
what
I
will
store
the
state,
but
I
will
try
to
keep
their
state
yeah.
Exactly
I
would
like
to
keep
the
state
the
application.
Stateless
first.
B
B
B
Think
we
are
done,
I
cannot
find
any
questions
anymore.
So
so,
what's
the
conclusion,
our
application,
server,
debts
or
not
I
think
the
conclusion
is
I
also
always
also
have
to
be
a
little
bit
careful.
What
I'm
talking
about
so
my
project
I
always
said
you
know:
I'm
an
enterprise,
development
and
I
recognize.
Actually,
then,
all
my
and
enterprise
large
projects
they
look
more
or
less
like
Stata
project,
so
there
is
actually
what
also
happened
in
the
recent
years.
B
A
And
Tanya
appears
on
the
screen
and
yeah
we're
at
the
top
of
the
hour.
It
was
amazing,
as
always
Adam
thank
you
so
much
and
we'll
talk
to
you
on
Tech
Talk
soon,
but
I.
Just
don't
like
to
take
the
opportunity
to
tell
people
that
at
the
cube
con
Amsterdam,
there
is
going
to
be
seen
for
Jade,
a
so-called
native
Java
dia,
where
you
can
hear
more
about
Jakarta,
EE
and
micro
profile
as
well,
and
we're
really
we're
pretty
sure
we
will
see
out
of
there
as
well.
So
yeah.
A
And
we
may
I
will
chase
you
about
that
ASAP,
because
we
want
to
start
publishing
and
promoting
so
for
everyone
out
there.
Please
have
your
eyes
open
for
that
and
we're
extremely
hoping
to
see
you
there
and
before
even
that
event,
I
will
have
additional
tactics
get
in
touch
with
me.
If
there
is
something
you
want
to
hear
about,
or
you
want
to
present
I
would
love
to
put
you
in
the
schedule
and
we
can
get
you
participating
in
Jakarta
tactics
and
for
now
again
thanks
Adam
thanks.