►
From YouTube: OCB: AI-enabled Proactive Management for Cost and Performance Optimization in Hybrid MultiCloud
Description
ProphetStor has been working on bringing AI/Machine Learning technology to the management and control planes of Cloud and 5G. The traditional IT infrastructures are passive, over-provisioned, and lacks of visibility. With the introduction of Machine Learning and Multi-Layer Correlation Analysis, we strive to make the management of OpenShift and Kubernetes management in general to be Proactive and Application-Aware in order to optimize the cost and performance from traditional IT environments to Cloud. In this episode, we will talk about why we think that the solution will revolutionize the cloud operation and greatly simplify the complexity in the management. We will also present how the OpenShift ecosystem, when used in cloud and 5G, can benefit the users with enhanced application performance and cost.
A
Okay
looks
like
we're
rolling
my
name's
mike
waite.
This
is
the
openshift
commons
briefings.
This
is
our
operator
hours
show
here
today
every
wednesday
at
noon.
Eastern
time,
and
today
we
are
fortunate
to
have
I'm
gonna
actually
have
to
share
my
screen.
Aren't
I
yes
today
we're
actually
fortunate
enough
to
have
with
us
two
gentlemen
from
profit
store.
A
We
we've
been
talking
with
him
all
morning
here,
and
there
goes
the
phone,
so
we
have
eric
chen
who's,
the
ceo
of
profit
store
and
ming
shu
who's,
the
evp
of
profit
store,
and
they
are
joining
us
here
today
on
our
show,
I'm
going
to
stop,
sharing
and
and
welcome.
Gentlemen.
Thank
you
for
coming
and
and
being
our
our
next
contestant
on
the
commons
briefing
hour
show
welcome
good
morning
good
afternoon.
Thank.
A
A
B
A
Yeah,
it's
been
a
long
year
yeah
and
I
I
gotta
tell
you
I
haven't
you
know
we,
our
offices
have
been
closed
for
physical.
You
know
presence
since
the
end
of
march,
and
I
never
thought
I'd
be
able
to
work
here
in
my
in
my
my
office
here
at
my
house
for
nine
straight
months
without
going
crazy.
How
are
you
folks.
B
Yeah,
I
still
remember
last
year
in
red
hat
summit.
You
know
we
have
a
big
audience
and
a
big
stage
and
the
ceo
of
ibm,
red
hat
and
microsoft
and
ourselves
all
are
invited
as
keynote
speakers
on
the
stage
and
suddenly
this
year
everybody
is
in
their
pajamas
right
and
locking
their
rooms
and
doing
all
the
conference
thing,
and
that
was
fun.
A
You're,
the
ceo
and
co-founder
of
profit
store,
and
we
always
like
to
find
out
a
little
bit.
Something
about
you
know
like
tell
us
a
little
bit
something
about
yourself.
I
know
when
we
were
talking
earlier
when
we
were
doing
one
of
our
you
know.
Dry
runs.
You
mentioned
that
you've,
you
founded
profit
store,
but
that's
not
the
first
company
you
started.
Can
you
tell
us
about
that?
That's
right.
B
B
and
and
of
course,
our
president
sunny
shoe
dr
xu
also
co-founded,
a
company
called
imi
and
later
was
acquired
by
nine
large
egg
and
then
by
broker
for
3.4
3.7
billion
dollars.
Right
and
ming
actually
has
been
a
serious
entrepreneur,
help
co-founded
several
companies
and
all
have
have
been
all
have
successful
access
and
most
recent
one
was
raucous
networks
and
that
was
later
acquired
by
brokay.
A
A
B
Okay,
the
name
profit
means
that
forcing
something
yeah
and
the
reason
why
we
wanted
to
to
name
the
company.
This
way
is
that
because
we
would
like
to
bring
the
foreseeing
something
into
it,
operations,
which
means
that
we
would
like
the
I.t
systems
or
crowd
operations
to
be
poor
active
rather
than
passive,
a
reactive
like
before
right.
In
the
past
we
have,
we
have
a
high
availability
systems
bill
built
for
ids,
for
enterprise,
gray
or
or
carrier
grade
systems.
B
But
now
that
h,
a
high
availability
will
kick
in
only
when
one
system
fail
and
the
other
one
will
take
over.
But
that
is
also
very
tense,
taking
process
and
and
things,
and
also
the
the
cause
of
making
it
right
and
also
maintaining
them,
will
be
very,
very
high
and
can
can
be
prohibiting
in
in
nature.
So
we
want
the
iot
system
or
collaboration
to
become
proactive.
B
All
these
machine
things
can
be
should
be
brought
in
in
order
to
make
it
happen,
and
therefore
we
need
to
collect
a
lot
of
data
and
we
would
like
to
provide
information
as
a
services.
That's
why
the
name
of
the
company
has
been
profit
store
data
services
since
day
one
and
we
started
working
on
storage
systems,
of
course,
and
all
the
way
up
to
it.
A
A
B
B
A
Okay,
yeah!
Well
what
so?
What
what
do
you
have
for
us
today?
I
know
we've
got
ming
here
on
the
phone.
I
know
that
you
folks
are
going
to
do
some
demos,
but
but
tell
us
about
tell
us
about
the
problems
that
that
your
company
solves
for
customers
and
how
you,
how
you
help
address
the
challenges
that
that
businesses
are
are
facing
as
they
move
into
multi-cloud.
B
B
In
addition
to
that,
workloads
are
mostly
dynamics
right
and
therefore
on.
When
in
the
past
you
are
handling
the
dynamic
of
the
workloads
with
with
the
constant
provisioning
of
the
resources,
and
that
is
not
right,
so
it's
actually
very
challenging
to
try
manage
and
optimize
and
it
needs
to
be
changed.
B
It
needs
a
huge
change
and
there's
one
interesting
stories.
I
want
to
share
with
you.
You
know,
especially
this
related
actually
related
to
what
initially
what
we
do
here
so
many
many
years
ago,
when
I
was
working
for
the
previous
company,
I
helped
co-found
it.
We
actually
sent
a
group
of
engineers
over
to
a
remote
customer
site
for
installing
the
software
defined
storage
solutions.
B
After
two
weeks
of
work,
they
finished
the
job.
They
got.
Compliments
from
the
user,
a
very
demanding
use
as
a
matter
of
fact,
and
then
return
with
triumph
right,
and
we
were
all
happy
and
pleased.
You
know
both
engineering
and
management,
and
we
have
been
thinking
that
oh,
this
is
yet
another
success
story
that
we
can
share
with
the
rest
of
the
world
and
we
are
our
ways
to
something:
that's
even
bigger.
B
So
one
week
later,
as
a
matter
of
fact,
I
visited
the
inside
partner
that
we
worked
together
with
together
with
to
address
the
the
the
need
for
for
the
customer's
case.
Thinking
that
the
ceo
of
the
si
parents
will
be
thanking
me
and
women
with
a
celebrating
party,
but
that's
the
things
actually
went
totally
wrong
when
there
actually
there's
no
cop
red
carpets,
there's
no
dinner.
The
ceo
of
that
company-
actually,
although
I
said
actually
told
me,
dana
eric
it-
was
a
great
project.
The
customer
is
happy.
B
We
make
good
money
and
you
have
a
very
good
support
engineers.
Okay,
however,
I
would
like
to
terminate
the
the
reseller
country
with
you
effectively
immediately,
and
I
was
shocked
right
to
say
the
list
right.
I
say
why
we,
I
thought
if
we
we
actually
onto
something
great-
and
we
have
been
doing
this
successfully
at
the
customer
side,
but
the
the
ceo
actually
replied
saying
that.
B
B
So
years
later
I
met
with
I
left
my
previous
company
and
I
met
with
dr
sunny
shu,
who
was
I
used
to
be
a
mit
professor
and
a
serious
entrepreneur
right?
We
started
to
talk
about
that.
We
need
to
bring
the
vacation
awareness
first
to
storage
management
and
then
to
id
and
the
cloud,
and
it
clicked
very
well
and
besides
he
was
expert
on
ai
machine
learning
and
and
from
that
point
on,
he
has
been
a
brand
of
what
we
do
so
at
the
time
of
2012.
B
A
B
B
Okay,
and
it
is
very
simple
to
say
that,
because,
basically,
all
the
things
that
has
been
invented
has
been
invented,
but
but
saying
that
I'm
using
ai
technology,
it's
a
very
simplified
thing
to
say,
because
you
know,
if
you
just
finally
throw
in
the
technology
to
the
the
issues
that
you
want
to
address,
you
would
end
it
up.
You
would
end
up
with
a
tremendous
resource
demands
just
to
solve
a
very
small
issues.
B
Okay,
so
I
actually
actually
we
have
seen
and
witness
the
alpha
goal,
beating
all
these.
The
the
goal
champions
in
the
world
right
and
that's
a
very
ingenious
task.
They
have
done
that
proven
something,
but
they
also
have
been
using
a
data
center
for
doing
that.
So
you
don't
want
to
use
a
a
gpu
server
just
to
to
predict
the
size
of
a
container
the
size
of
the
dynamics
or
container.
That
would
be
now.
The
benefit
of
that
would
not
justify
or
warrant
the
needed
resources.
B
A
B
Okay,
so
if
you,
if
you
look
at
what
happens
in
the
past,
that
that
every
layer
of
the
I.t
solutions-
okay,
when
I
talk
about
layers,
I'm
talking
about
horizontal
layers
right
that
things-
we
have
the
operating
systems,
we
have
real
right:
red
hat
enterprise
linux.
We
have
microsoft
as
operating
systems,
we
have
oracle
s,
applications
postgres
as
a
database
or
dbs
on
the
the
database
systems
and
underneath
we
have
in
the
infrastructures
right
and
also
we
have
clouds
okay.
B
B
The
whole
stack
will
be
a
very
complicated
process
to
to
take
so
in
order
in
order
to
support
applications,
people
just
throw
in
a
lot
of
resources
in
there
and
they
don't
and
and
also
each
each
and
every
one
of
them
they
are.
They
are
doing
their
best
effort,
job
trying
to
maximize
the
performance
of
that
layer
without
understanding
the
impact
or
what's
going
on
outside
of
the
scope
of
that
specific
layer.
B
What
I
mean
by
what
I
mean
by
westing
is
that
if
you
don't
provide
the
resource
and
the
right
amount
of
resources,
for
example,
if
you
are
over
provisioning
the
resources,
then
you
are
wasting
your
money
right
so
resource
it
actually
equals
the
cost.
But
at
the
same
time,
if
you
don't
know
the
right
amount
of
resources
and
if
you
underprovision
it
you
are,
you
will
run
the
risk
of
out
of
memory
or
even
the
stop
operations
of
the
applications.
B
So
it
will
create
tremendous
impact
for
the
performance
of
the
application.
So,
in
order
for
the
application
to
run
optimally
in
terms
of
performance
and
cost,
we
need
to
provide
the
just-in-time,
baited
resources
for
that
application
and
that's
the
foundation
of
our
thinking,
and
in
order
for
that
to
happen.
First
of
all,
you
need
to
understand
your
application.
B
So
that's
why
we
need
to
have
the
application,
awareness
and
everything
that
for
the
accelerations
industries
that
are
talking
about
providing
specific
hardware
or
software
combinations
in
order
to
accelerate
some
application
will
require
the
understanding
of
the
application.
That's
why
application
awareness
is
there
yeah,
so
those
are
the
things
that
we
have
there.
B
One
is
the
the
application
anatomy
you
gotta
understand
how
the
application
is
composed
of.
They
are
composed
of
multiple
micro
services
and
that's
actually
aesthetic
description
but
other
than
the
static
description.
You
gotta
have
the
workload
which
is
dynamic
in
nature
and
and
getting
to
understand
both
the
the
static
and
also
dynamic
configurations
of
that
application.
A
It
was
in
a
lab,
and
I
used
to
help
I
used
to
help
software
companies
port
their
applications
to
open
vms
on
alpha,
and
we
had
storage
arrays
in
the
labs
which
were
about
three
feet
tall.
It
was
called
the
ba
356
and
and
each
each
hard
drive
and
the
thing
was
like
the
size
of
it.
We
used
to
call
them
a
brick.
They
were
like,
like
this
big.
A
A
Four,
was
it
four
gigabytes
or
something
like
this,
and
so
storage
has
really
changed
a
lot
since
since
then,
back
in
the
day
when
everything
was
just,
you
know
a
data
center,
what's
it
gonna
be
like
in
the
next
24
months
I
mean
thing:
it's
been
30
plus
years
since
I
worked
at
dec
things
changed
kind
of
gradually
over
the
you
know,
the
last
you
know
20
and
have
been
accelerating.
What
do
you
think
storage
is
going
to
look
like
24
months
from
now.
B
B
So
I
would
bring
the
attention
everybody's
attention
to
what
ibm
is
trying
to
do
here.
Ibm
recently
have
a
actually
a
merger
with
a
red
hat
right
and,
and
they
are
bringing
the
whole
company
forward
as
a
cloud
company.
Okay.
So,
as
you
can
see
here,
that
ibm
is
very
focusing
on
multi-cloud
and
ai.
Now
and
if
you
look
at
the
traditional
storage
companies
such
as
emc
or
netapp
right
gmc
is
now
becoming.
Dell
has
actually
have
become
dell.
That
netapp
is
a
very
respectful.
B
So,
as
people
are
moving
to
the
cloud,
the
asset
of
the
storages,
the
demand
for
the
asset
is,
is
becoming
diminishing
and
to
the
crowd.
Then
you
know
we
need
to
understand
how
cloud
is
working.
So,
as
you
can
see,
the
the
the
openshift
and
also
kubernetes
platform
are
supporting
more
distributed
scale
out
storages
such
as
ceph
and
that's
exactly
what
what
what's
moving
and
also
high-speed
storages
right
and
also
pci
based
storages.
A
B
One
thing
that's
very
interesting
is
the
storage.
Virtualization
is
moving
from
a
controller
that
actually
service
multiple
storages
into
the
storage
or
the
so-called
disk
controllers,
and
and
to
that
level
of
over
translation,
and
that's
a
very
interesting
development
that
we
will
be
seeing
in
the
future.
That
will
service
cloud
very
well.
A
So
I
I,
when
I
asked
you
guys
to
come
on
the
show
here
today
I
was
like
hey
we're
going
to
be,
you
know
doing
this
live
we've
got
people
who
are
on
the
bridge.
We're
also
live
on
on
we're
streaming
on
youtube
and
and
facebook
and
twitch,
and
we
just
had
a
question
come
in
and
I
was
wondering:
if
he's
was
it
a
s
asks?
B
People
tend
to
just
trust
the
native
auto
scaling
thinking
that
it
will
take
care
of
everything
right
for
the
scale,
but
the
fact
is
is
it.
It
is
not
like
that.
You
know
when
you
are
at
the
at
the
kubernetes
layer
that
is
a
virtualization
and
also
management
layer,
providing
services
to
microservices
or
local
applications.
B
B
Okay-
and
that
is
a
very
that-
is
very
device
oriented,
not
application,
aware.
Okay,
so
I
will
use
one
example
that
you
know
if
you
are
at
the
counter
of
mcdonald
right
and
there's
a
cash
register
that
how
many
cash
registers
are
usual
you
should
have
in
order
to
service
in
order
to
service
the
customer
in
the
queue
in
the
line.
B
Here,
if
you
depends
on
the
number
of
queue
size
to
decide
the
number
of
cash
register,
then
that
will
be
a
misleading
things,
because
because
it
takes
time
and
resources
to
open
up
a
cash
register,
okay,
that's
costs
associated
with
that.
But
if
but
the
the
fact
is
that
you
actually
need
to
need
to
base.
You
need
to
decide
the
size
of
your
number
of
cash
registers
based
on
the
traffic.
B
B
I
said
that
we
actually
have
a
brain
from
sunny
right,
the
the
brain
in
there,
so
that
we
would
take
in
the
the
data
from
the
monitoring
services
or
monitoring
solutions,
and
then
do
the
data,
analysis
and
understanding
the
workload
well
having
the
predictions
and
then
based
on
that
workload.
Predictions.
B
A
A
A
A
C
C
A
Just
one
question
ming
I
apologize
for
interrupting.
You
know
I
was
reading
your
documentation
earlier
in
the
week.
Actually
I
think
it
was
last
week
on
your
site
and-
and
you
mentioned
you
know,
ai-
for
open
shift.
You
didn't
say
kubernetes
or
you
know
various
are
you
folks?
Is
this
only
running
on
openshift
or
is
that
just
the
one
that
you
were
talking
about
in
your
documentation.
C
Oh
you're
right,
so
we
actually
can
run
on
many
kubernetes
deployments,
so
not
just
openshift
so
for
that
document
may
be
specific
on
openshift,
but
in
general
we
support
all
kinds
of
kubernetes
deployments,
including
you
know
the
gkp
and
aks
and
eks
and
all
the
different
communicative
version.
So
we
also
call
that
so
it's
not
just.
C
C
I'm
sure
no,
no
anytime,
you
can
interrupt
me
anytime.
So
so
I
was
saying
right
so
knowing
when
the
workload
is
going
to
change,
I
can
assist
us
in
good
results
right.
Our
system
allows
user
to
observe
the
workload
position
at
different
layers.
So,
for
example,
you
can
see
over
here
we
can
do
the
prediction
at
the
cluster
linspace
application
or
controller
and
with
different
prediction
granularity
and
with
together
also
different
prediction
layers.
C
C
So
I
want
to
take
kafka
as
an
example,
and
you
may
have
you
know
in
your
system.
Many
kafka
producers
sending
messages
to
kafka
brokers
at
different
rates
at
different
time
of
the
day,
and
you
want
to
make
sure
you
have
enough
kafka
consumer
to
receive
and
process
those
messages
in
time
without
causing
a
lot
of
delays.
C
C
C
C
C
C
So
our
system
will
be
able
to
analyze
your
usage
and
based
on
prediction
to
know
how
you're
going
to
spend
at
the
name
space
level,
and
you
can
see
in
this
example.
There
are
many
resources
wasted
in
the
idling
time
and
that's
the
reason
we
can.
Actually,
we
can
provide
a
recommendation
to
find,
what's
the
right
instance
type
and
how
many
instances
how
many
customer
resources
you
need
to
do
your
job.
C
A
C
C
C
We
are
looking
to
try
to
see
the
relationship
between
different
layers
of
the
from
the
application,
all
the
way
down
to
the
lower
level,
network,
storage
and
stuff,
but
we
are
not
getting
into
the
inspection
of
what
the
specific
application
is
wrong
at
that
level.
So
this
is
a
different
kind
of
domain.
Okay,.
C
We
are
able
to
provide
the
future
looking.
What's
your
application,
look
like
how
how
much
resources
you're
going
to
use-
and
so
as
you
see
over
here,
is
our
application
and
we
provide
a
cpu
recommendation
memory
recommendation
as
well
as
what
the
predictions
look
like,
and
one
interesting
to
see
over
here
is
in
the
data
dock
showing
over
here.
They
cannot
draw
what
is
in
the
future.
So
then
we
are
actually
presenting
a
table
format
of
what
the
features
look
like
for
the
recommendation
prediction.
C
We
have
also
other
you
know,
integration
at
the
cluster
level,
an
application
level
and
other
thing
we
just
show
thing
in
our
regular
ui.
We
actually
presented
those
information
into
the
data
dashboards.
So
there's
a
strong
integration,
a
customer
can,
you
know,
benefit
from
a
single
plan
or
gas,
a
class
for
management.
B
B
Okay,
thank
you
yeah
mike
one
very
interesting
thing
to
see
here
is
that
in
the
past
the
monitoring
services,
they
all
provide
a
one-way
street
data
flow.
For
example,
they
collect
the
data
by
the
agents
and
then
send
over
to
the
cloud
services
and
then
do
the
do
the
ui
for
assembling
accumulated
data.
B
A
B
Okay,
of
course,
we've
been
working
together
with
dire
dog
and
they
are
all
very
professional
people
and
elon
and
myself.
We
are
all
sitting
sitting
on
the
red
head
customer
advisory
council
right,
a
solar
panel,
sorry,
solar
partner
advisory
council,
so
so
red
hat
actually
provides
very
good
play,
phones
and
ecosystems
for
not
only
openshift
itself,
but
also
for
companies
or
vendors
that
are
working
together
to
make
the
openshift
for
more
popular
or
feasible
platforms
for
enterprises
customers.
B
So
we
all
appreciate
that
and
datadog
is
from
our
perspective.
They
offer
very
good
monitoring
services
and
and,
as
a
matter
of
fact,
we
we
actually
have
with
datadock.
We
come
up
with
a
layer
called
data
adaptation
and
that
data
adaptation.
With
some
revisions,
we
are
now
be
able
to
support
other
vendors,
such
as
distinct
and
data
trace,
and
we've
been
working
on
others.
For
example,
we've
been
considering
the
collaboration
with
instanta
special
ibm
now
right.
A
B
B
A
Yeah
he
pizza
off
got
a
new
project
going
now.
It's
gonna
be
interesting
to
see
what
what
what
they
come
up
with.
So
how
lo,
when
you
get
this
all
set
up,
is
it?
Is
it
fully
optimized
right
away,
or
is
there
some
learning
curve
for
federator.ai
to
to
actually
sort
of
like
train
itself
before
you
know
giving
out
workload,
accurate
workload,
predictions.
A
C
C
Now
so,
if
you're
going
to
do
a
daily
prediction,
then
a
few
hours
later,
we
still
we
start
getting
some
basic
data.
Then
we
can
start
do
prediction
but
of
course,
the
longer
the
period
it
is,
the
better
the
result
is
and
so
looking
at
a
weekly
monthly,
they
will
actually
take
a
few.
You
know
for
the
monthly
you
probably
take
you
know
a
week
or
two
weeks.
If
we
start
to
have
the
data
showing
up
and
the
longer
you
go,
the
the
better
we
our
petition
is
okay,.
B
But
you
know
we
can
start
working
on
something,
even
within
a
very
short
period
of
time,
like
ming
said
that
you
know
for
the
operations
the
performance
wise,
so
it
probably
doesn't
need
to
have
a
long-term
forecast
for
that,
and
but
for
the
long
term
the
cost
analysis
it.
You
probably
need
to
have
a
have
a
longer
longer
term
of
observation
data.
In
order
for
this
to
be
more
accurate,.
A
Okay,
well,
we're
we've
got
just
a
couple
minutes
left
and
I
know
from
looking
at
the
people
on
on
the
bridge
here
on
the
call
I
know
we've
got
tad.
Lobeck
has
been
on
the
whole
time
and
I
believe,
he's
your
you
know,
vp
of
marketing
for
the
company.
So,
given
that
we
have
just
about
two
minutes
left
what
what
would
you
would
you
like
to
throw
out
there
mention
so
that
you
don't
have
to
have
the
phone
call
from
tad
right
after
the
show
saying
ming?
A
A
B
But
the
key
thing
he
chose
to
join
prophet
story
is
that
he
sees
the
the
great
potential
or
exponential
growth
in
the
coming
years,
because
in
the
in
a
phase,
one
of
the
adoption
of
the
kubernetes
platforms
right,
no
matter
if
it's
in
cloud
or
5g,
the
first
issue
will
be
will
be
the
visibility
and
security
right.
But
the
second
phase
of
that
will
be
the
day
two
operations,
which
is
which
will
be
many
on
the
performance
and
the
cost
of
the
operations.
B
And
that's
what
our
profit
store
has
been
working
on,
and
we
want
to
bring
automation
into
the
picture
because
all
the
solutions
we
have
seen
on
the
market,
they
all
require
the
human
engineering
and
that
complex
operations
warrant
newer
solutions
that
would
involve
machines
and
machine
learnings
right.
So
that's
exactly
what
a
profit
store
is
good
at.
What
means
just
demonstrated
is
that
we
are
using
our
partners
data.
B
So
so
we
believe
this
will
be
the
next
phase
of
a
big
and
explosive
growth
of
adoptions
and
in
order
for
kubernetes
and
openshift
platforms
to
be
widely
adopted,
and
also
the
success
of
of
the
multi-cloud
to
be
warranted.
We
believe
federated
will
be
something.
That's
really
needed
solutions
and
that's
what
interest
tad
and
what
ted
has
been
telling
people.
A
B
A
It
up
here
for
for
the
for
the
day,
I
can
believe
that
we're
having
a
conversation
with
you
and
datadog
about
doing
a
customer
or
webinar
coming
up
sometime
in
either
january
or
february,
when
we,
I
think
that
should
be
pretty
interesting
when
we
put
that
together.
If
anyone
wants
to
find
out
about
getting
in
touch
with
us
about
when
that
webinar
is
with
profit
store
and
datadog,
you
can
just
send
me
an
email.
It's
just.
It's
just
wait.
Redhat.Com
wait
at
redhat.com.
A
B
So
you
can
take
a
look
at
that
check
it
out
on
the
red
hat
marketplace.
It
has
been
certified
and
available
there
and
also
we
provide
free
trial.
So
this
is
the
link
to
it
and
you
can
also
follow
us
on
twitter.
We
constantly
put
out
the
updates
about
the
technology,
also
availability
of
products
and
also
the
application
cases,
studies
etc
so
stay
in
touch
and
stay
healthy.