►
From YouTube: GitLab 12.5 Kickoff - Geo
Description
Fabian Zimmer (Product Manager, Geo) highlights new features and direction for GitLab Geo in 12.4.
Geo Product Strategy: https://about.gitlab.com/direction/geo/
Geo Roadmap: https://gitlab.com/groups/gitlab-org/-/roadmap?scope=all&utf8=%E2%9C%93&state=opened&label_name[]=group%3A%3Ageo&label_name[]=geo%3A%3Aactive
A
A
Alright
super:
can
you
see
my
my
spin?
Yes
great,
alright,
so
I
usually
go
through
the
high
level
things
that
we
are
attempting
to
do
so
these
are
our
currently
active,
epics
and
I
will
talk
through
some
of
them
in
more
detail
than
others,
but
I'll
also
highlight
a
few
of
the
core
issues
that
are
relevant
to
those
I
start
at
the
top,
and
then
I
worked
my
way
through.
A
So
the
first
thing
is,
and
I
already
highlighted
that
last
time
for
12.4
we
are
investing
significantly
more
energy
in
geo
to
actually
enable
geo
as
the
disaster
recovery
solution
in
comm.
So
I'll
actually
have
a
second
roadmap
for
this.
To
give
you
a
little
bit
more
like
detail
on
what
we're
doing
here
and
so
I
think
the
first
important
thing
to
to
highlight
is
that
adding
and
replicated
data
type
is
always
important
for
dr
just
because
it's
not
replicated
you
can't
recover,
so
that's
ongoing,
but
more
interestingly.
A
At
this
moment
we
are
working
on
rolling
out
geo
to
staging
kept
calm
and
then
also
defining
the
specific
Geo
architecture
on
github.com.
So
when
this
goes
to
production,
so
the
Geo
architecture
is
essentially
figuring
out
how
exactly
we
are
going
to
deploy
it,
organizing
some
some
work
with
infrastructure
and
also
creating
a
server
architecture
diagram
for
staging
com.
So
this
is
something
that
we
are
going
to
do
in
12.5
as
sort
of
a
pre-google
to
actually
deploying
it.
Also
yeah.
We
are
rolling
it
out.
A
So
there's
a
lot
of
more
like
technical
issues
in
here
that
are
relevant,
but
what
we're
trying
to
accomplish
is
automating
as
much
as
possible
and
working
very
closely
with
the
infrastructure
team
to
accommodate
that.
So
this
is
just
a
high
level
overview,
and
once
that
is
done,
we
are
going
to
actually
rolling.
You
are
starting
to
roll
as
out
at
dr.com.
A
A
It
could
take
significantly
longer
I'll
keep
you
posted,
but
this
is
the
state
that
we
have
at
the
moment.
This
is
one
of
the
highest
priorities
in
NTU
at
the
moment,
just
because
it
will
resolve
a
lot
of
issues
that
we
we
have
we're
you
know,
scaling
is
difficult
and
having
it'll
calm
is
the
the
way
to
go.
So
let
me
quickly
close
those,
then
verifying
all
replicated
data
I've
spoken
about
that
a
few
times.
I
think
everybody
in
the
team
agrees
that
he
replicates
something.
We
should
also
verify
it.
A
However,
we
are
kind
of
taking
a
step
back
from
some
of
the
initially
plant
work.
I
spoke
about
verifying
uploads
last
time
and
while
looking
at
verification
for
uploads,
but
also
verification
for
design
repositories,
we
realize
that
verifying
data
at
the
moment
is
actually
a
lot
harder
than
we
would
like
to
be
so,
rather
than
finding
sort
of
localized
solutions.
We
are
now
actually
thinking
about
a
that's,
a
potentially
more
framework
like
approach.
The
team
is,
is
thinking
about
solutions
here
and
from
a
product
perspective.
A
You
know
how
we
can
how
we
can
actually
make
sure
that
verification
is
simple
and
straightforward.
So
you'll
probably
hear
me
talk
about
this
a
little
bit
in
more
detail
after
12.5,
based
on
the
outcomes
when
we
are
thinking
about
product
icing
this,
but
at
the
moment
it's
a
little
bit
more
discovery
and
I
believe
this
is
valuable,
because
ultimately,
one
of
the
challenges
we're
facing
in
geo
is
that
we
are
one
team
and
there's
a
lot
of
new
data
being
added
to
get
lab
all
the
time
that
we
would
like
to
replicate.
A
And
so,
if
that's
hard,
we're
always
going
to
lag
behind,
and
so
one
of
the
challenges
that
we
have
is
deciding
how
we
can
make
it
really
easy
to
add
new
data.
Maybe
at
some
point
also
have
a
more
self
service
oriented
framework,
but
that's
in
the
future
and
the
team
has
some
work
to
do
to
come
up
with
some
sort
of
proofs
of
concepts
and
and
sort
of
initial
discovery,
then
the
improving
of
the
upgrade
documentation.
So
this
is
almost
done,
and
actually
we
are
releasing
quite
a
bit
of
working
here.
A
You
can
see
we
spent
a
significant
amount
of
time
working
on
improving
the
upload,
documentation
and
I
think
this
is
valuable
to
our
customers,
who
we
encourage
top
write
very
regularly.
The
most
important
outstanding
issue
for
for
12.5
that
I
would
like
to
highlight
here
is
the
documentation
for
euh
a
upgrades,
so
we
are
working
on
creating
a
step-by-step
instruction
for
upgrading
geo
nodes
in
an
H,
a
configuration,
and
so
we
have
a
NMR
at
the
moment
already
with
a
lot
of
that
information
included.
However,
this
needs
more
testing.
It
needs
to
be
robust.
A
Then
here
key
or
supports
a
single
geolocation
where
get
URL.
This
is
going
to
be
very
likely
released
in
12
point
four,
so
we
are
closing
this
off,
which
is
great
similar
here,
to
improve
object,
storage
support.
So
with
twelve
point
four,
we
are
releasing
a
beta
feature
that
allows
replication
of
data
in
object,
storage
through
Geo
itself,
so
rather
than
relying
on
the
replication
features
of
like
third
party
object-
storage
providers.
A
This
allows
customers
to
actually
replicate
natively
with
geo,
so
that
opens
up
possibilities
of
sort
of
a
multi
cloud
deployment
where
you
use
one
provider,
one
location
and
another
provider,
different
location
or
use
local
storage
and
I'm
quite
excited
to
see
this
being
closed
closed
off
at
at
this
stage.
The
main
reason
for
this
being
beta
is
that
we
have
to
test
it
at
scale
yet,
but
I
think
it
would
be
nice,
nice
thing
and
Gabriel
happens
to
be
on
the
call.
A
So
a
big
shout
out
to
him
who
worked
on
this
so
super
excited
to
see
this
then
here
support
for
design
repositories.
This
is
something
that's
also
important
to
to
highlight
where
we
had
initially
scheduled
this
for
delivery
in
12.4.
But
one
of
the
issues
here
that
is
taking
a
little
bit
longer
is
the
verification
which
I
explained
already
is
one
of
the
reasons
why
we're
taking
a
step
back,
but
also
we
have
to
handle
design
uploads
in
the
admin
area
for
geo
uploads
or
the
estimate
of
front-end
work,
which
zach
was
also
on.
A
The
call
is
going
to
talk
tackle
in
12.5,
so
this
will
take
a
little
bit
longer,
but
we
want
to
ship
this
as
one
as
one
feature,
and
you
know
as
I
see,
lots
of
small
issues
were
already
implemented,
but
we
would
rather
get
this
done
fully.
I
also
realized
that
I
skipped
this
more
robust
foreign
data
table
updates
I'm
not
going
to
go
into
too
much
detail
here,
but
I
can
show
you
the
epic.
So
essentially,
this
is
part
of
making
upgrades
easier
and
simpler.
A
A
Finally-
and
this
is
going
to
start
at
this
month
and
I'm
really
excited
about
this,
because
it
works
well
with
enabling
geo
for
disaster
recovery
on
get
lip
comm,
so
we're
going
to
start
product
discovery,
efforts
on
promotion,
demotion
and
failover.
So
these
are
all
interesting
topics
to
do
with
disaster
recovery
so
similar
to
simplifying
upgrades.
What
we're
going
to
do
here
is
actually
looking
at
the
processes
that
we
have
right
now,
the
documentation
that
we
have
revisiting
them
and
working
working
through
sort
of
an
initial
phase
of
you
know
what
are
potential
improvements?
A
What
can
we
do,
adding
the
our
demos
and
then
you
know
coming
to
a
stage
where
we
know
exactly
how
we
can,
how
we
can
iterate
and
what
the
next
steps
are.
So
in
12.5
you
would
probably
not
see
a
lot
of
feature
implementation
yet,
but
this
is
really
important
to
many
of
our
customers
and
I'm
excited
too
then,
once
we
have
a
better
better
overview
of
that
area,
actually
moved
into
implementation,
both
probably
on
the
user
interface
side.
A
So
this
is
something
to
to
stay
tuned
in
comparison
to
the
last
time,
I
spoke
to
you
where
the
promotional,
emotional,
failover
and
the
geo
should
be
simple
to
install
we're
sort
of
happening
in
parallel
I'm,
giving
the
amount
of
work
in
a
billing
to
you,
uncom
and
all
the
other
things
that
we
are
doing
at
the
moment.
We
are
delaying
this
a
little
bit
more.
That
also
makes
sense,
because
it's
a
it's
a
thing
that
is
larger
than
to
you
itself,
so
there
is
other
work
going
on
on
potential
provisionally
gitlab
in
HK
applications.
A
Little
bit
easier
and
Gio
falls
into
that.
So
there
may
be
some
movement,
but
before
we
fully
focus
on
this,
I
think
there
would
be
a
little
bit
of
time
spent
doing
for
sell
for
things,
and
that
is
pretty
much
it.
So,
if
I
had
to
summarize,
it
I'm
really
happy
that
quite
a
significant
amount
of
work
is
already
sort
of
coming
to
an
end
in
12.4,
and
that
was
a
little
bit
delayed
earlier.
A
So
this
is
nice
and
that
frees
us
up
to
start
some
new
things
and
also
double
down
on
some
of
the
existing
efforts
and,
as
you
can
probably
also
tell
there's
quite
a
bit
of
discovery
work.
So
if
you
have
any
questions
or
you
want
to
know
more
about
this
or
participate,
please
feel
free
to
reach
out
and
that's
the
Jo
for
12.5.
So
thanks
for
listening
and
have
a
good
day
and
now,
I
need
to
stop
recording
one.
Second.