►
From YouTube: Geo Introduction with Sunjung and Fabian
Description
Fabian, the Product Manager for Geo, explains the high-level functionalities of Geo to Sunjung, the new Product Designer for Geo and Distribution
A
B
Everyone,
this
is
Fabian
and
our
new
product
designer
I'm
talking
a
little
bit
about
Gio
and
how
how
it
works
and
what
it
does
and
how
the
interface
looks
and
essentially
giving
a
quick
tour
of
of
the
the
product
with.
Hopefully,
a
lot
of
questions
on
why
things
are
the
way
they
are
and
how
they
look.
Ok,
so
the
first
thing
is
I
think
for
for
Gio.
The
way
we've
we
structured,
the
the
product
at
the
moment
is
essentially
to
cover
two
very
broad
use
cases
and
they're
actually
quite
different,
and
they
have
different
requirements.
B
The
first
one
is
about
geo
replication.
This
is
why
it
is
named
geo
and
the
the
idea
from
sort
of
a
let's
say
user
perspective
is
that
if
customers
of
gitlab
have
offices
in
several
different
countries,
then
having
a
single
gitlab
instance
in
one
country
will
potentially
decrease
the
productivity
of
of
teams
that
are
remote
right
so
in
other
geo
locations
and.
C
C
C
B
They
will
have
a
primary
located
relatively
close
by,
for
example,
on
the
East
Coast,
and
then
they
have
a
secondary
node
in
Europe,
which
the
European
developers
can
actually
use
in
order
to
download
data,
for
example,
and
that
means
that
they
are
in
a
position
where
it
essentially
is
just
as
fast
to
work
with
with
gitlab
in
Europe
as
it
is
in
the
United
States.
That's
kind
of
the
first
really
big
use
case
for
4G,
oh,
and
you
can
already
see
this
is
accomplished
by
different
mechanisms.
B
B
You
have
artifacts
stored
and
object
storage
so,
for
example,
things
that
people
upload
to
get
lab
or
stuff
that
is
generated
when
continuous
integration
pipeline
is
built
and
also,
obviously,
we
have
repositories
so
the
actual
get
with
posit
or
E's
and
wiki's
that
need
to
be
that
need
to
be
replicated
so
geo
essentially
allows
to
do
this
on
a
continuous
basis.
So,
every
time
a
change
happens
on
the
primary
it
will
be
forwarded
to
the
to
the
secondary.
B
The
important
point
here
is
that
the
secondary
is
read-only,
so
you
cannot
actually
change
any
any
sort
of
State
on
the
secondary.
You
can
only
read
from
it,
and
that
is
maybe
something
we
can
change
in
the
future.
But
at
the
moment
the
complexity
involved
with
having
several
different
like
lead
and
right
right
above
it
lab
instances,
is
quite
significant,
given
that
we
use
phosphorus,
so
you
could
imagine-
and
for
example,
if
both
of
those
were
writable-
and
you
make
a
change
here.
B
B
We
don't
really
want
this
to
be
very
explicit
for
users
in
Europe
all
right
at
the
moment.
A
lot
of
the
complexity
of
all
of
this
year
is
kind
of
exposed
to
to
users,
because
they
kind
of
need
to
know
in
many
cases
that
they
are
using
a
secondary.
The
secondary
may
have
a
different
name
right
than
the
primary.
B
B
But
the
request
actually
gets
sort
of
proxied
over
here,
so
nothing
happens
on
the
secondary,
but
it
gets
sort
of
to
the
other
primary,
which
is
quite
neat,
because
you
are
essentially
able
to
to
use
the
secondary
as
if
it
was
writable
for
those
operations
which
is
already
a
big
improvement,
and
we
also
now
support
certain
sort
of
deployment
modes
where
you
can
give
like
a
single
name
for
your
primary
and
your
secondary
for
get
traffic.
So
it
would
be
something
like
git
dot,
customer
comm
and
there's
a
load
balancer.
That
can
actually
say.
B
Oh
when
the
users
in
North
America
say
this,
they
should
go
to
the
primary.
But
if
somebody
says
this
in
Europe
they
should
go
to
the
European
server
right.
So
it's
like
automatically
decides,
which
is
quite
neat
because
then
you
don't
need
to
remember
like
what
name
you
have
to
use,
and
so
that's
sort
of
the
the
number
one
scenario
does
that
does
that
make
sense.
Yeah.
B
The
number
two
scenario
is
that
once
you
have
a
secondary
node
here,
you
may
be
in
a
position
to
use
this
as
a
disaster
recovery
solution
right
because
you
could,
if
you
replicate
all
of
your
data
in
case
the
primary,
he
actually
has
a
failure,
maybe
planned
or
unplanned
the
time
it
could
take
you
to
recover.
It
could
be
lower
because
you
can
essentially
then
take
your
secondary
and
make
it
into
a
primary
ok.
C
B
So
there's
obviously
complexity
involved
right.
You
need
to
think
about.
How
do
you
make
a
secondary
into
a
primary
when
what
does
that
involve?
Is
that
easy?
Is
it
difficult
right,
and
at
the
moment
this
is
all
relatively
manual,
so
people
need
to
go
to
the
command
line
and
they
need
to
enter
enter
commands
right
in
order
to
accomplish
this,
but
it's
not
easy
to
focus,
go
to
the
user
interface
and
say:
hey
I
am
in
this
situation.
C
B
That's
kind
of
kind
of
Geo
in
in
a
nutshell,
those
two
things
on
a
high
level.
There
are
a
lot
of
intricacies
to
this
right,
as
in
the
devil,
is
often
in
the
in
the
detail
when
it
comes
to
replicating
that
was
data,
because
replication
is
not
necessarily
super
straightforward
right.
You
need
to
you,
transmit
those
data
over
the
Internet.
It
needs
to
be
robust
right.
You
need
to
be
able,
we
try
and
verify
people
rely
on
on
this
being
quite
stable.
B
So
from
an
engineering
perspective,
this
is
an
interesting,
interesting
problem
right,
and
so,
if
we
go
for
example
down
here,
we
have
current
limitations
where
you
can
sort
of
see
all
the
sort
of
data
that
get
lab,
generates,
repositories,
projects
and
then
look
at
sort
of
the
support
for
for
geo
right.
So
we
don't
actually
replicate
everything
yet,
which
is
something
that
we
would
like
to
to
do
in
in
the
future.
So
that's
something
to
keep
in
mind.
Mm-Hmm.
C
B
B
C
B
So
if
you
just
switch
off
the
primary,
you
will
lose
some
data
right,
and
so
you
need
to
like.
If
you
do
a
plan
thing,
you
need
to
be
sure
that
everything
is
actually
copied
right
and
you're
you're
sort
of
in
a
state
where
everything
is
identical
now,
which
means
that
you
have
to
restrict
what
happens
from
the
primary
in
some
form,
and
so
at
the
moment
that's
also
only
possible
in
sort
of
a
command
line
mode,
but
it
doesn't
allow
for
for
a
sort
of
like
a
maintenance
mode.
B
C
B
I
think
we
have,
we
have
essentially
like
from
the
user
perspective.
There
are,
let's
say
there
are
two
groups
right:
we
have
systems
administrators
who
essentially
are
tasked
with
setting
up,
get
lab
and
setting
up
get
lab.
You
know,
maybe,
with
disaster
recovery
capabilities
or
with
geo
replication,
they
they
need
to
administrate.
The
system
maintain
the
system,
install
it
upgraded
all
of
those
different
tasks
so
and
when
you
will
see
when
we
talk
to
our
customers,
often
we
talk
to
like
when
I
talk
to
them.
B
C
B
And
they
are
just
that
they're,
not
systems
administrators,
they
have
no
access
to
the
geo
administrative
panel
or
they
may
even
not
have
an
understanding
of
what
you
is.
They
just
want
to
use
gate
lab
for
whatever
task
they
they
need
to
perform,
but
they
get
frustrated
because,
let's
say
they
are
in
in
a
different
location,
Europe
or
Asia,
or
wherever
far
away
from
the
primary
and
Goodlett
may
just
be
really
really
slow,
and
so
for
them.
It's
a
performance
question
right
and
they
want
to
do
their
tasks
quickly.
B
A
C
B
Yeah,
so
that's
that
I'll
show
you
geo
in
in
a
second
like
a
GU
installation,
so
you
can
actually
see
from
a
Systems
Administrator
point
how
it
how
it
looks,
but
I'll
also
show
you
a
little
bit
how
it
looks
from
from
a
user
perspective.
I
think
one
thing
to
know
is:
if
you
look
at
this
from
sort
of
a
product
perspective,
we
have
a
lot
of
documentation
right,
but
it
is
actually
not
necessarily
very
like
there's
a
lot
of
information
for
users
to
digest
right.
So
I'll
give
you
an
example.
B
Set
up
Qi
open,
you
have
like
seven
different
like
pages
what
you
need
to
go
through
right,
so,
for
example,
you
need
to
do
like
a
gate
lab
installation
first
right
in
whatever
way
you
choose
to
install
it
right,
then
you
need
to
set
up
database
replication
at
some
point
right,
which
is
there
is
a
lot
of
you
know.
Like
I
mean
you
can
see
here,
there's
a
ton
of
command
line
in
interaction.
So
from
my
experience
and
looking
at
the
support
tickets,
we
get
a
lot
of
support
questions.
B
B
Yeah,
there's
there's
there's
a
lot
of
manual
like
work
to
be
done
and
it
gets
complicated
because
you
can.
You
can
deploy
geo
in
various
different
configurations
right
with
high
availability
scenarios
and
I
can
walk
you
through
that
it's
a
little
bit
involved,
but
it
is
not
like.
Most
of
these
things
are
not
actually
able
to
be
performed
in
the
user
interface
other
than
the
command
line,
so
graphics,
graphical
user
interfaces.
B
C
B
So
I
have
sort
of
a
hobby
on
the
demo
version
of
of
gait
lab
signed
out
as
installed.
That
I
can
walk
you
through,
and
so
one
thing
it
you
can
see
is
like
I'm
here
on
sort
of
the
secondary
geoserver
right.
So
if
we
go
back
to
our
or
think
I'm
on
this
note
here,
the
secondary,
if
you
can
see
already
it's
over
here,
it
gives
you
a
little
notification
saying:
I
am
on
a
not
quite
sure
what
happened
here.
I'm
on
the
read.
B
Only
to
you
note
and
what
you
can
do
is-
and
this
is
not
the
perspective
from
sort
of
a
regular
user
right,
you
can
walk.
You
can
go
through
the
the
interface
here.
You
can
look
at
things
right.
You
can
go
to
your
project.
For
example,
you
know
you
can
first,
let's
say
you
can
create
a
project
or
that
this
is
interesting.
First
design
challenge.
B
Okay,
yeah:
these
are
some
of
the
challenges
that
we
have
in
this
interface,
because
I
think
I
have
done
this
at
least
10
times
myself.
Where
you
know
you
you,
you
see
the
banner
of
course,
but
you
can't
actually
like.
Then
you
can
still
sort
of
edit
the
interface,
but
you
can't
really
save
anything
because
it
is
a
it's.
A
weak
only
make
posit
Ori,
but
what
we
can
do
is
we
can
go
to
the
primary
node
and
now
we're
here
right.
The
notification
is
gone
and
I
can
say:
okay,
I'll
create
a
project
now.
A
B
B
A
B
A
B
B
That's
a
very
basic
question
to
establish
yeah
so,
but
that's
that
you
can
also
and
I
think
this
is
now
that's
kind
of
the
the
use
of
it
because
I'm
an
admin
but
I
can
go
to
the
admin
area
now,
which
is
here
yeah
and
you
can
see
we
need
to
update,
and
maybe,
if
we
have
a
tiny
bit
of
time,
I
can
actually
do
that
for
you.
But
here
you
see
the
geo
admin
interface,
and
this
is
where
you
can
see.
B
Stuff
is
other
information,
so
how
many
replication
slots
there
are.
This
is
sort
of
the
information
that
we
display
to
two
systems-
administrators
right
in,
for
example,
this
here,
it's
quite
important
because
it
essentially
tells
systems
administrators
how
much
of
my
data
that
was
generated
on
the
primary
was
synchronized
to
the
secondary
yeah
all
right,
and
so
this
is
sort
of
our
our
dashboard.
You
can
see
the
events
that
are
like
processed
right,
so
you
can
see
like
the
last
change
was
two
minutes
ago.
B
You
can
set
some
verification
information,
because
not
only
about
about
replication
is
also
about
making
sure
that
what
was
transferred
is
actually
not
corrupted
right.
So
that's
another
another
thing,
yeah
some
other
information
here,
yeah
you
can.
You
can
perform
some
actions
right.
You
can
change
some
settings
for
for
the
node
here,
but
there
are
many
more
settings
in
the
command
line
and
get
labardi.
B
You
have
this
new
node
button,
which
is
a
little
bit
of
a
lie,
because
if
I
just
give
it
a
name
and
a
URL,
it
does
not
actually
add
a
new
node
right.
You
must
have
a
lot
of
configuration
already
done
previously
in
the
command
line
to
be
able
to
use
this
okay,
all
right.
So
that's
another
sort
of
question
it's
like
if
somebody
wanted
to
add
a
new
node
here.
B
We
own
the
primary
right,
so
this
is
everything
that
is
displayed
on
the
on
the
primary,
and
you
can
see
here.
This
actually
has
has
been
checked
on
now,
so
it
updated
it's
a
little
bit.
It's
not
yeah
I,
don't
know
how
this
is
implemented
in
the
rails
side
of
things,
but
it
sometimes
takes
a
little
while
to
upgrade
to
update
now.
B
If
you
go
to
the
secondary
I
can
use
this
like
open
projects,
link
here,
I'm
being
moved
to
the
secondary
and
the
secondary
looks
almost
the
same,
but
the
the
top
interface
here
is
very
similar
right,
but
you
have
more
information.
So
if,
for
example,
you
have
projects-
and
you
can
now
see
on
the
second
area-
it's
like
okay,
what
was
actually
synced
when
was
it
synced
do
I
want
to
rethink
or
we
verify
so
there
are
certain
tasks
that
systems
administrators
can
perform
here,
even
like
some
batch
operations.
B
B
B
So
I
think
there's
a
lot
of
opportunity
to
redesign
this
and
make
make
it
easier
for
people
for
systems
administrators
to
perform
the
tasks
that
they
want
to
perform
yeah,
and
so
there's
number
one,
and
it's
also
about
sort
of
surfacing.
The
right
information
in
the
in
the
right
way
right
so,
for
example,
like
this
verification
information
here.
A
C
C
Is
there
any
like
hierarchy
for
this
information
or
like
something
is
more
important
for
them
and
something
is
less
important
for
them,
and
the
other
question
is
like
now:
does
health
status
is
fortunately
healthy,
but
this
has
changed
to
something
else
like
yellowish
or
the
red
like
warning,
then
probably
some
of
the
information
might
be
more
important
than
the
others.
Yes,.
B
That's
actually
100%
true,
so
a
I
don't
know
necessarily
for
all
of
our
users.
What
the
hierarchy
of
information
is
I
have
some
personal
guesses,
but
that
we
would
need
to
validate
repositories.
Often
quite
important.
Lfs
objects
are
more
important,
but
it
really
depends-
and
it
may
depend
on
the
customer,
but
I
can
say
for
sure
that
if
people
see
the
screen-
and
let's
say,
repositories
have
like
the
way
this
is
being
shown
as
you
can.
Basically
I
can
show
you
right
now.
B
It's
like
here
will
be
like
a
little
red
portion
right
off
like
non
synced
right
and
so
as
soon
as
things
are
not
a
hundred
percent
synced
and
like
people
want
to
know
why
yeah
right
and
then
the
systems
administrators
in
my
experience,
want
to
be
able
to
do
something
right
and
say
like
okay.
The
reason
why
this
is
red
could
be
this
or
this
or
this,
and
in
order
to
do
this,
I
can
now
ideally
go
to
the
interface
and
say,
like
hey
I'm,
going
to
let's
say
resync
the
last.
B
The
last
changes
and
I
can
solve
my
problem,
but
more
often
than
not,
people
are
not
sure
what
to
do
in
order
to
in
order
to
sort
of
solve
this
for
them,
which
then
means
they
contact
support
because
the
and
I've
seen
this
I
can
send
you
some
support.
Tickets.
I've
seen
this
like
several
times
where
you
know.
Essentially
we
get
a
screenshot
of
this
and
people
say
like
hey.
B
My
LFS
objects
are
only
70%,
synced
I,
don't
know
why,
but
I
really
need
them
to
be
synced
yeah
or
you
know
like
like,
and
how
how
you
arrived
at.
That
is
also
interesting.
It's
like
was
it
a
hundred
percent
at
some
point,
did
it
never
actually
hit
the
hundred
percent
right
like
what
can
be
done
like
all
of
those
are
really
interesting,
interesting
questions.
B
People
like
something
changes,
right,
they've,
upgraded
to
get
lab
version
right,
or
you
know
like
having
monitoring
capabilities
right
off,
saying
like
for
the
last
seven
days
right,
everything
you
know,
Mike
was
always
a
hundred
percent
or
you
know
if
something
new
happened
it
actually
like
synced
within
five
minutes
or
the
average
lag
time
between
my
primary
at
my
secondary
over
the
last
day
was
two
minutes
now
it's
ten
minutes,
but
maybe
that's
an
alert
right.
So
there's
a
lot
of
sort
of
monitoring
or
it
like
yeah
I.
Think
it's
the
fool.
B
It's
the
full-blown
requirements
like
people
want
to
know
what
the
current
state
they
want
to
see
the
history
they
want
to
know
what
is
normal
right
bye.
What
definition
you
know
like
what
can
I
do
when
things
are
not
normal
right,
so
how
do
I
invite
solve
problems
and
having
that
information
is
available
in
some
form
of
interface,
is
quite
helpful
to
two
people,
but
I
would
say
in
a
vast
majority
of
cases.
B
B
Where
it's
like,
okay,
you
know
like
what
are
like,
like
the
health
checks
right.
If
something
is
something
like
wrong
right,
there
are
command-line
tools
here,
like
geo
check
right
that
you
can
run,
but
this
is
not
available
in
in
the
interface,
for
example.
Right
you
can
then
like
get
like
a
status
like
something
similar
here
right
in
the
command
line
that
you
can
see
on
the
on
the
web
interface
and
then
like
there's,
all
sorts
of
like
hey.
C
A
B
When
I
think
the
the
state
that
we
are
in
here
with
this
is
we
we
know
that
there's
a
lot
of
like
things
that
we
need
to
improve
here
and
I
think
we
have
a
good
customer
base
to
talk
to,
but
we
haven't
yet
sort
of
focused
more
holistically
on
solving
this
as
a
as
a
like
overall
sort
of
problem
of
how
this
could
be
I.
Think
we've
been
quite
opportunistic
and
small
and,
like
I,
said
that
oh
we
do
this,
and
we
do
this
here
and
that's
been
good,
but
there
is
no.
B
They
are
for
some
I
know,
experience
guidelines
or
baselines
for
our
our
users,
yet
right
and
I
think
there's
a
lot
of
interesting
work
to
be
done
to
say,
like
hey,
what
are
the
top
five
tasks
that
people
like
systems,
administrators
going
to
the
web
interface?
Actually
right
and
I-
think
that's
that's
going
to
be
a
really
huge
and
really
interesting
opportunity
to
to
figure
out.
A
B
C
I
really
hope
I
can
reach
to
my
goal
and
can
I
keep
you
the
very
quick
feedback.
I
mean
you
actualize,
because
I
could
find
something
like
in
general,
like,
for
example,
like
I.
Don't
think,
is
a
good
user
experience
to
provide
something
which
could
not
work
like,
for
example,
for
the
read-only
Mo's,
like
I
thought
that
I
ok
project
were
create
a
new
issue
on
the
second
note
is
not
possible,
but
users
people
can
see
the
buttons
there,
which
is
enabled
state
yes,
meaning
like
yeah.
C
Ok,
this
is
clickable,
so
we
believe
me
to
the
another
webpage.
So
if
they
click
on
it
still
there's
no
other
like
visual
cues
like
ok,
you
cannot
do
it
here,
of
course,
there's
a
like
manner
like
okay.
This
is
read-only
mode,
but
I
think
that's
like
you
know,
but
I
think
in
this
case,
I
need
to
do
more
and
user
research
or
yes,
I
can.
B
B
B
B
We
will
need
to
support
systems,
administrators
logging
on
to
a
user
interface
on
the
secondary,
but
maybe
in
a
large,
like
proportion
of
the
use
cases,
the
users,
the
actual
sort
of
get
lab
users
will
go
to
the
primary
first
because
they
can
do
other
things
they
need
to
do
and
the
user
experience
of
that
is
acceptable
to
them.
Right,
maybe
that's
the
case
right
and
if
that's
the
case,
then
we
don't
need
to
expend
a
lot
of
time
optimizing.
B
The
experience
on
the
secondary
but
I
don't
know
that
right,
and
this
is
a
very
sort
of
important
decision.
I
think
because
we
could
also-
and
people
have
tried
to
do
it
on
a
technical
level.
Try
to
make
the
secondary
web
interface
appear
as
if
it
is
right
about
right
and
maybe
do
some
trickery
where
you
you
every
time
somebody
clicks
a
button,
you
send
them
to
the
primary.
So
it
looks
very
similar.
But
it's
it's
really
not
right
and
there's.
You
know
there
are
some
things
that
you
can.
B
You
could
do,
but
maybe
like
on
a
fundamental
level.
I
would
like
to
know
what
is
actually
happening
because,
like
and
I
think
this
is,
this
is
something
where
ya
at
the
absence
of
evidence
is
not
evidence
of
absence
right,
but
I
I
see
support
tickets
for
people
struggling
with
the
systems,
administrator
interface
and
I
see
that
they
they
say,
like.
Oh
I,
don't
understand
how
this
works,
but
if
not
seen
a
single
ticket
in
the
last
six
months
from
somebody
complaining
about
the
read-only
interface
right,
so
I,
you
know
I'm
a
little
bit
like.
C
B
C
B
A
A
B
Cool
cool
yeah,
yeah,
that's
Geo
in
a
nutshell
and
yeah.
We
can
talk
about
the
road
map
at
at
a
different
point
in
time,
but
on
on
a
very
high
level,
I
think
we
are
going
to
focus
a
little
bit
more
on
disaster
recovery,
an
abling
kyo-ahn
get
lap
comm,
we
working
how
we
like
add
and
verify
data,
and
that's
a
engineering
heavy
effort,
but
I
also-
and
this
is
where
you
come
in
I-
would
love
to
redesign.
B
What
really
sends
me.
The
wrong
word
understand
the
the
tasks
that
systems
administrators
need
to
perform
in
the
web
interface
and
then
really
think
about
how
we
can
improve
the
experience
there
because
I
think
that's
that's
a
lot
of
the
exposure
of
the
the
UI
gets
and
it
may
be
used
for
some
really
important
tasks
like
business
critical
decisions
right.
So
it
needs
to
be
robust.
Yeah.
C
B
C
A
C
B
It
is
probably
worth
while
capturing
it
issues
on
get
lab
okay,
because
that
makes
it
easy
for
the
team
to
to
see
what
you're
thinking
about
and
I
can
run
you
through
like
next
week
or
so
like
where
and
how
to
do
that,
because
then
it
makes
it
easy
for
everyone
to
contribute
to
those
discussions
and
their
public
as
well.
That's
really
I
try
I
try
to
do
that,
just
to
be
transparent.