►
From YouTube: BTS OpenShift Max Neunhöffer(ArangoDB)
Description
Behind the Scenes at OpenShift Commons with Michael Hausenblas
Interview with Max Neunhöffer (ArangoDB)
B
B
I
should
say
that
some
of
the
things
we
do
10
or
above
because
we
know
the
mortals
like
us
with
resource
definitions
as
its
evolved,
exclaims
and
end
up
like
anyway.
So
the
first
thing
I
should
do
is
I
should
deploy
it,
our
authoritah
for
AA
TV
and
for
our
ability,
replication
and
therefore
I'm
executing
these
two
commands.
B
A
B
And
so
therefore,
I
prefer
to
write
so
the
operators
are
up
and
running
and,
as
we
can
see,
we
have
now
created
also
great
sauce
definitions,
namely
we
have
custom
resourced,
a
traditional
type
or
the
wrong
deployments.
That's
for
clusters
or
civil
service
law
public
assets,
and
we
have
one
for
replication
and
exactly
the
same
with
the
other
day.
Therefore,
actually
now
just
really
take
very
simple
Yahoo
file.
Let's
look
at
it.
It's
actually.
All.
Where
is
my
cluster
name?
B
I
set
up
for
those
TLS
certificates
at
a
rate
name
to
access
it
I,
say:
I
write
out
a
cluster.
They
say
that
I
wish
and
I
ask
for
load
balancer
configuration
for
external
access
for
the
database
as
well
as
other
synchronizations
and
notice
that
I
use
the
is
names
here.
But
Google
cloud
service
will
obviously
give
me
public
IP
addresses,
but
therefore
I
will
have
to
quickly
configure
my
DNS,
but
just
after
I
have
deployed
this,
that
we
have
the
public
right
of
an
author.
B
Because
we
can
watch
what's
being
created
actually
happening
is
that
the
system
will
now
deploy
a
test
balloon
to
find
out
what
around
OTB
version
is
actually
in
this
technology,
which
can
also
not
be
really
derived
from
the
tag
or
omission
it
that's
the
test
balloon.
It
will
then
know
that
this
is
a
remote
DD
e,
3.4,
dot,
release,
candidates,
3
or
something
like
that,
and
it
can
then
deploy
the
actual
parts
and
knows
the
version
of
a
robot
EPA
T.
A
B
They
are
our
central
roster
DP.
The
cluster
configuration
that's
a
higher
value
which
also
handles
fellow
fighter,
but
as
a
customer,
you
don't
have
to
know
that
what
you
need
as
a
customer
is
the
coordinators.
That's
the
CR
at
the
end,
and
that
is
where
the
balances
in
front,
because
client
requests
code
for
the
coordinators.
It
doesn't
really
matter
which
coordinator
you
attack,
and
that
is
what
does
the
community
planning
and
optimization
and
then
and
finally,
we
have
the
primaries.
That's
the
actual
database
service
for
the
roughly
cost
of
the
data.
A
B
A
B
And
setting
up
data
center,
the
data
center
replication
is
actually
quite
difficult
if
you
want
to
be
really
scalable.
So,
therefore,
if
you
look
at
these
pods
here,
the
system
is
again
a
digital
system
with
sync
masters
and
co-workers,
which
is
responsible
to
actually
you
replicate
everything
in
here
to
the
other
side,
we'll
see
this
in
action
in
a
few
minutes
so
that
far,
let
me
quickly
show
you
the
other.
B
B
B
A
For
the
ones
for
the
ones
of
our
viewers,
who
are
not
that
deep
into
those
equal
database,
whatever
it
looks
a
bit
like
like
MongoDB
like
you,
have
facing
documents
collections
whatever
is
that
that
can
uniquely
give
us
a
under
what
is
the
pay
look
here?
What
it's
like?
It's
a
key
value
store,
but
well.
B
B
B
Documents
in
here
with
a
collection
and
as
you
see
if
we
go
here,
we
can
see
that
the
collection
is
distributed
across
shots
as
readers
and
each
shot
is
replicated
two
times
so
there's
a
which
has
a
replica
right.
This
is
worth
nicely.
Let's
see
what
the
other
cluster
is
doing.
Yes
or
pods
are
up
and
running,
but
that's
one
device.
The
fixed
units
in
Korea
might
as
well
open
a
second
tab
where
I
go
destination.
They
have
a
lot
de
and
here
I
have
to
use
the
node.
What.
B
B
B
B
A
How
would
you
do
that
in
like
this?
Is
this
works
because
both
like
they,
they
are
both
in
your
accessible,
but
would
you
do
that
in
a
production
setup
in
the
same
way
or
how
would
you
actually
is
there
any
good
practice
on
how
do
you
actually
connect
these
two
clusters?
You
see
using
this
one
secret
command
to
actually
add
alert
in
the
production
environment.
B
A
B
Sentences
when
I
now
deploy
this
and
when
it
works,
then
it
does
an
incremental
swing
in
the
beginning
or
if
a
part
of
the
data
is
already
there,
it's
not
again
copied
and
then
establish
establishes
as
a
situation
where
the
sting
systems
know
that
they
are
in
soon
and
they
just
tailed
the
writer
headlock
of
each
database
and
get
everything
they
find.
If
you
cross
over
to
the
others.
Now.
B
A
B
B
B
B
A
B
A
Awesome
Wow
max
tepid
is
really
impressive,
I
mean
I
I.
With
my
background,
you
know,
I've
been
working
in
the
distributed
data
and
data
engineer
is
basically
I'm.
I
know
and
I
appreciate
what
is
behind
bad,
and
this
is
really
big.
A
stuff
I.
Don't
think
that
there
are
many
solutions
out
there
that
comparable
to
that,
let's
talk
a
little
bit
about
the
custom
resources
you've
developed
there,
so
you
haven't
operator
that
does
a
lot
of
work.
A
B
B
Is
first-
and
this
all
requires
relatively
intricate
knowledge,
and
we
would
like
to
of
beat
down
place
of
minor
revisions
and
stuff
like
that,
and
if
you
think
about
that,
it's
really
better.
If
you
actually
control
all
these.
So
therefore,
we
have
chosen
the
route
that
there
is
the
custom
resource
type
or
around
deployments
and
there's
a
custom
resource
type
for
our
replication
and
our
operator,
which
is
actually
one
executable
which
fulfills
both
the
role
of
both
these
operators,
just
by
speed,
switch
to
a
different
mode
just
handles
everything
manually.
B
B
B
B
A
A
Right-
and
there
is
just
one
question
that
I'm
not
entirely
sure
how
that
would
work,
so
you
have
the
bearment
one,
which
is
the
kind
of
follower
with
only
replicating
from
from
the
Tanner
engine,
one
that
is
the
primary
and
one.
How
like?
Let's
imagine,
obviously
that
doesn't
happen,
but
imagine
TKE
goes
down,
but
how
would
the.
A
B
B
B
A
A
A
B
B
B
And
confirmed
exception,
I've
never
trusted
with
one
coordinator
and
edibles,
but
our
next
thing
I
would
like
to
show
you
that
I
can
simply
click
Nastya
in
the
UI
you'll
make
that
a
cluster
a
bit
larger.
So
what
happens
here
is
that
the
UI
of
our
database
puts
all
kind
of
ice
requirements,
namely
that
it
wants
to
be
a
bit
thicker
into
the
agency
and
our
operator
observes
that
and
it's
obviously
the
operator
or
it's
an
imposition
boy,
a
new
parts
to
make
this
happen.
B
B
B
B
A
B
B
B
So
now
we,
the
second
agent
and
so
on
and
without
any
services,
were
actually
through
seamlessly
behind
the
sea
operate
each
knows
about
auto-ship
upgrades
are
allowed
and
not
it
knows
that
at
the
end,
both
of
three-door
3.17
to
three
or
four
of
VDU.
It
knows
it
cannot
go
to
3.5,
and
it
knows
it
cannot
go
down
anymore.
So
that's
the
benefit
of
having
our
own
operator
that
we.
A
This
is
very
impressive
and
it's
it's
really
cool
to
see
how
you
leverage
existing.
You
know
core
resources,
you
know
the
parts
and
services
and
persistent
volumes
and
and
and
roll
your
own
stuff,
essentially
bending
queries
to
your
needs.
The
air
with
with
the
the
operator
with
the
series
and
the
controllers
there
and
that's
that's,
really
I-
think
that's
how.
A
B
To
use
it
that
is
fully
opens
cause
we
have
held
charts
so
that
was
and
about
the
status.
I
should
say
that
we
are
currently
running
our
so-called
acceptance
test,
which
really
tests
things
to
the
limit,
like
shooting
everything,
and
we
do
this
platform
by
platform,
and
we
call
the
operator
production-ready
on
gke
and
we
are
hard
at
work
to
notice.
A
B
A
Awesome,
thank
you
so
much
max.
That
was
really
impressive.
I
really
appreciate
you
taking
time
and
showing
us
that
great
that
everything
worked
this
and
expected.
I
never
doubted
it.
Obviously,
but
it's
really
great
to
see
this
awesome
work
and
yeah
I
hope.
A
lot
of
people
will
follow
up
there
and
at
least
have
a
look
at
the
operator
and
learn
from
that,
and
hopefully
also
go
and.