►
From YouTube: SERVERLESS JAKARTA EE WITH MICROPROFILE AND CLOUDRUN
Description
How do you create a serverless EE environment using Google's CloudRun with Quarkus and Helidon? I will talk about what points should be considered in comparison with conventional EE environment.
A
B
B
B
B
This
core
profile
matches
the
Micro
Profile
completely,,
that's
not
the
case.
So,.
This
is
the
Micro
Profile
5.0
specs,,
but
the
green
part
I
colored
with
one
hand,,
but
the
green
part
is
familiar
to
everyone..
Cdi
was
also
there
today.
Jsonb
and
xrs
are
familiar
guys,,
but
this
blue
part,,
open,
tracing,,
open,
API,,
etc.
B
Isn't
it?
When?
It
comes
to
the
execution
environment
for
a
trial,
version,
for
example,,
Wellweblogic,,
Tire,,
Widefly
or
Liberty,,
IBM,,
etc.
basically
have
javn
and
application
salad.
There's
an
app
that
deploys
something
like
that.
I
think
it's
the
form
of
the
application
itself,
so
on
the
application.
Server
side,
how
to.
B
B
B
B
B
B
B
Guideline
that,
if
you
create
it
like
this
in
such
a
place,,
it
will
become
a
very
convenient
application
for
microservices
or
cloud
native.Based
on
this
guideline,.
Various
open
source
implementations
are
also
created.
Perhaps.
The
most
famous
thing
in
Java
is
the
Netflix
open,
OSS
stack.
It's,
the.
B
B
B
B
End
points
for
management,,
setting,,
management,
and
monitoring,
and
security,
etc.
Originally,.
It
was
an
AP
server,,
so
Web,
logic,,
tylers,
and
so
on
were
collateralized.
In.
The
recent
ecosystem
of
container
architecture,.
It
has
become
a
car,
and
in
terms
of
packaging,
things
in
the
Java
world,
are
solidified,
and
jakarta
gi,
moves.
But
in
some
cases,.
The
image
for
reading
comprehension
is
Windows
for
the
time
being,,
but
putting
that
aside,.
If
you
ask
for
a
Linux
image,,
anything
is
fine.
That's.
Why
the
technology
that
runs
on
top
can
be
Rails,,
Go,,
Python,
or,
of
course,
Java.
B
B
B
B
B
B
B
Putting
bones
aside
at
the
beginning,
that
portability
can
be
secured
by
the
unit
of
reading
comprehension,
image,,
so
I
like
the
packaging
and
the
image,
and
the
foundation
for
execution
is
also
a
micro.
Profile..
The
point
is
that
a
little
bit
of
API
is
fine,,
but
how
about
a
fairly
high-performance
cluster
like
web
logic
or
something
like
that?
But
when
it
became
a
micro-profile,
I
said
that
Car
News
would
provide
such
functions,.
B
B
Think
it's
possible
to
do
it
without
combining
Hellidon,,
Focus,
and
Kubalist,.
But
if
you
do
that,
you'll
lose
some
of
the
functions
that
Level
Logic
and
Fire
used
to
do.
I.
Think
it's
good
to
keep
in
mind
that
there's
a
different
system
that
requires
someone
to
be
in
charge,
and
if
the
quality
of
the
car
is
good,,
it's
going.
B
B
B
B
B
B
B
B
B
B
B
By
the
time
the
relayer
database
hits,,
a
connection
was
made
before
the
request.In
other
words,.
If
100
requests
were
made
per
second,,
100
connections
were
made,
and
if
it
took
10
seconds,
a
nationwide
connection
was
made..
Well,
rdb
was
acting
badly.
Currently,.
It
is
possible
to
avoid
it
by
recognizing
various
things
such
as
rds
proxy,,
which
is
out
of
the
mechanism.
Like
that
connection,
pool,
and
JCB,
cloud,
seekle,
proxy,,
etc.
I.
B
B
B
B
Feel,
like
I,
can
turn
around
about
3,
times,
and
moreover,.
This
connection
is
not
created
every
time,
and
when
it
is
no
longer
used,,
it
is
released
and
reused.
In
the
case
of
an
AP
server,.
It
is
possible,,
but
in
the
case
of
a
serverless,
the
connection
can
be
locked
when
6
processes
are
running.
B
B
B
B
B
Think
that's
one
reason,
isn't
it?
It's
like
this,,
isn't
it?
The
startup
time
is
included
in
the
response
time
to
the
user,.
So
if
it
takes
a
minute
here,,
it's
no
good
at
all,,
right?
Well,,
no
matter
how
long
the
quantity
is,.
If
possible,
it
must
be
less
than
1
second.
After
all,,
it
will
be
stressful
for
users.
B
B
B
B
B
B
B
B
B
B
B
B
Not
that
you
can't
include
configuration
files,
but
settings
that
work
in
the
same
way
in
any
environment,
such
as
the
log,
4G,
format,,
etc.
Of
course,.
There's
no
problem.
If
you
put
them
in
here.
However,
in
production,
for
example,,
it
works
with
information,,
but
in
debug,
it
seems
to
work
with
debugging.
Such
log
levels
must
be
written
in
environment
variables
as
environment-dependent
information,
or
database
connection
information
or
API
connections.
The
basic
idea
is
to
define
the
destination
and
such
information
in
environment
variables.
B
B
B
B
B
B
4G
For
example,
application,
logs,
error,
logs,
trace
logs,
and
various
other
logs,
such
as
4G
Log,
Manager
2
can
be
used
to
separate
out
various
logs,,
and
since
this
is
a
large
volume,,
it
should
be
rotated
once
an
hour
or
backed
up
afterwards.
I
think
you've
done
a
lot
of
log
management,
such
as
sending
it
to
the
server,,
but
in
modern
times,
many
prerequisites
have
already
changed.
To
be
honest,
I'm.
B
B
B
B
Yes,,
if
you
do
this,
like
this,,
when
you
go
to
the
management
screen
of
the
log
that
is
built
into
the
cloud,,
you
can
see
the
log
as
a
normal
business
type,.
But
you
want
to
dive
deeper.
If,
you
have
a
question,,
you
can
jump
to
the
cloud
logging
screen
here,
and
you
can
write
queries,.
So
you
can
write
quite
detailed
information,
and
you
can
embed
tags
in
it
or
use
the
Jason
format.
.
With,
this
kind
of
Jason,.
You
can
create
a
structured
log
structure,.
B
B
B
B
B
B
B
B
B
B
B
Today,
too,.
We
talk
about
open
telemetry
as
much
as
that.
Survey.
As,
you
all
know,,
multiple
applications
work
together,.
So
it's
a
hindrance.,
Simple
stack
traces
alone
are
no
longer
prone
to
errors
and
performance.
Bottlenecks.
That's.
Why
we
need
a
mechanism
called
distributed
tracing,,
but
this
is
just
a
standard
specification
for
connecting
logs,
and
nothing
new..
I.
B
B
B
B
B
B
Well,
I
mentioned
this
earlier,,
so
it
might
be
okay
to
skip
it
for
a
bit,.
But
apart
from
Open
Trees,,
there
is
a
mechanism
called
Open
Census
advocated
by
Google.
Well,.
This
guy
is
very
similar,,
so
let's
integrate
it,,
so
Open
Telemetry
was
born,,
but
as
of
September
2022,,
the
support
situation
is
mixed,
and
it's
cloud
and
base.
B
B
B
Only
the
otlp
general-purpose
protocol
is
supported
as
an
exporter.
So
Google's
cloudtrace
doesn't
support
odlp
yet,
and
you
need
to
set
up
Google's
tracer,.
So
even
if
it
doesn't
support
it,
a
bit,
you
can
use
that
focus
to
cloudtrace
with
cloudrun
If.
You
want
to
link
it,,
you
have
to
use
the
open,
telemetry,
sdk,
yourself.,
It's.
B
As
an
implementation,.
It's
called
a
tracer.
In
other
words,.
The
destination
is
the
destination,.
So
this
time
I
want
to
send
it
to
Google,.
So
I
write
the
destination,
and
when
I
actually
initialize
the
settings
and
use
it,,
it's
called
a
span,
so
for
clothes,
I
put
the
actual
inside
I
want
to
measure
in
it..
B
This
is
Google's,
Cloudtrace,
screen,
and
I.
Think
it's
similar
to
the
screen.
I
mentioned
earlier,,
but
the
parts
circled
in
red
are
the
front-end
application
processing,
and
the
back-end
processing
as
well.
It's
written,,
and
you
can
finish
it
on
one
screen,
and
the
log
is
also
attached,,
so
you
can
do
a
drill-down,
analysis.,
I,
think
yes,
and
finally,
management
of
confidential
information.
Actually,,
it's
GCP
in
the
first
place.
B
Important
to
know
whether
you're
doing
this,,
but
for
the
time
being,
there
are
times
when
you
want
to
connect
to
a
minescale
that
is
placed
in
a
VM
you've
set
up
yourself,
or
other
than
that,.
There
are
times
when
you
want
to
throw
it
to
an
external
system,,
so
confidential
information,
After
all,,
it's
going
to
be
pretty
important
to
put
secret
information
in
the
configuration
file
and
push
it
to
github,.
So
it's
going
to
be
terrible,,
so
Google's
secret
manager
is
used.
It's,
very
convenient..
B
B
B
B
B
Think
it's
better
to
make
it
so
that
the
user
experience
will
improve
considerably,.
So
it's
better
to
take
on
the
challenge.
The
following
3
elements.
Microservice
container
office
Mistakes
are
called
serverless
Minimizing
the
startup
time
from
the
elements
and
breaking
away
from
the
server
And
finally,,
a
new
type
of
application,
execution
environment
that
adopts
a
micro
profile.
So.
If
you
know
the
background,,
you
should
know
more
I
think
you
can
understand
why
this
function
can
be
used.
Thank.
You.