►
From YouTube: 2021 06 30 EMEA Sharding Group Sync
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
A
B
Fabian
is
on
pdo
right
already
started.
B
Yeah,
okay,
let's
just
get
started,
commute
left
to
his
desk,
but
anyways.
So
quickly
summarize
what
I
just
wrote
there
right.
You
know
in
that
dog
pain,
distribution
and
the
geo
team.
Their
feedback
comes
back,
came
back
and
I
got
a
feeling
that
those
distribution,
the
geo,
can
become
a
long
tails.
B
This
is
not
good
news,
so
I
was
thinking
whether
we
can
drop
the
support
of
self-managed,
which
eliminates
the
distribution,
work
and
then
drop
the
work
drop.
The
requirements
of
geo
support,
then
the
geo
is
no
longer
no
a
dependency
for
this
product.
So
that's
where
I
came
from
because
I'm
worried
that
both
will
become
long
tail.
B
They
cannot
start
until
we
say
you
get.
We
have
basic
pieces
ready
for
you
to
start
your
work,
so
I
guess
our
best
scenario
is
august.
That's
the
very,
very
optimized
optimistic
the
best
best
case.
So
that's
when
I
can
start.
B
Yes,
that's
my
yeah!
That's
my
understanding!
August!
Probably
that's
the
best
case.
They
they
can
start
start
thinking
about.
No
the
solution,
I'm
I'm
even
more
worried
about
distribution.
Like
dj
said
it
can
take
a
few
several
milestones.
I
don't
know
how
many
those
are
several.
B
I
know
how
long
prefect
took
what
is
basically
six
months.
B
C
I
believe
that
this
is
exactly
what
we
are
scoping
right
now.
We
are
focusing
on
github.com,
not
general
self-managed.
There
is
a
separate
issue
for
the
self-managed,
which
actually
gonna
be
a
ton
of
the
work
anyway,
but
the
configuration
is
needed
to
be
in
place
so
like.
We
cannot
shrink
this
beyond
what
we
have
now,
because
we
are
doing
already
something
super
minimal,
which
is
like
exposed
ability
to
configure
that,
and
this
is
something
that
infrastructure
team
needs
to
perform
on
github.com.
C
B
Okay,
so
let
me
just
confirm
that
we
don't
need
distribution,
work
to
configure
the
database
or
provisioning
database
configure
the
so
that
can
be
delayed
to
the
second
iteration.
We
are
focusing
on
the
gitlab.com.
We
are
asking
the
infrastructure
team
to
to
configure
the
gitlab.com
is
that
it's
not.
C
Fully
true,
we
need
a
way
to
configure
that,
so
this
configuration
needs
to
be
set
in
omnibus
and
cng
to
some
extent,
we
can
do
it
on
our
own,
but
it
will
always
in
any
case
require
review
match
requests
somehow
from
the
distribution.
We
are
not
really
intending
to
do
anything
beyond
that.
We
just
like
things
that
you
mentioned
john,
but
we
like.
C
So
this
is
actually
like
the
focus
right
now
right
configuration
it's,
it's
still
not
super
well
described,
because
we
are
still
like
evaluating
a
way
how
this
configuration
should
look
like,
but
I
I
guess
like
in
one
week
from
now.
We
will
know
exactly
how
we
want
to
design
that,
like
don't
get
me
wrong,
like
it's
not
really
like
a
lot
of
work
to
do
it,
but
at
any
rate,
it's
gonna
require
some
help
from
the
distribution
team.
C
And,
like
the
question
ask
was
like,
of
course,
we
can
do
it-
probably
probably
twice
slower
than
distribution
team,
and
this
is
like
prequest
and
for
someone
from
the
distribution
team.
It's
maybe
like
at
most
one
week
of
the
work
with
other
testing.
C
If
we
give
them
like
a
design,
how
it
should
be
configured
how
the
settings
should
look
like.
So
it's
about
the
efficiency
like
we
can
probably
do
it.
We
don't
need
review,
but
this
discussion
was
like.
Can
we
kind
of
offload
that
aspect,
so
we
could
focus
on
more
important
aspects,
because
it's
purely
like
pass-through
type
of
the
setting
you
go
to
github
rb
set
the
database
name
and
it
goes
past
through
into
database
on
your
meta.
C
B
Yeah-
let's
don't
worry
about
who
does
it?
I
my
plan
is
to
have
distribution
to
do
the
distribution
work.
The
geo
team
do
the
dish
to
your
work.
So
don't
worry,
I'm
not
asking
this
team
to
do
it.
I
right
now
we
just
need
to
figure
out
are
the
necessary
match
to
have
for
the
first
iteration,
because
right
now
we
are
we're
being
asked
how
we
can
parallelize
and
accelerate
or
add
more
confidence
to
the
january
timeline.
So
that's
all
I'm
I'm
working
on!
I'm
I'm
willing
to
involve
more.
B
You
know
other
teams
when
it's
necessary.
So
that's
that's
where
I
am.
E
There
there's
some
work,
that
is
in
the
critical
path
that
can
be
outlined,
and
then
there
are
other
things
that
can
can
and
have
to
happen
later,
very
likely
also
with
regards
to
self-managed
customers.
I
think
this
is
also
true
for
deal.
You
know
it's
like
if,
if
we
want
self-managed
customers
handled
appropriately,
we
need
to
handle
these
things
as
well,
and
this
is
probably
true
for
backup
and
restore
as
well.
C
C
B
Okay,
joe
is
not,
is
not
absolutely
required
for
the
first
iteration,
but
configure
omnibus
and
cng
configuration
that's
required
in
the
first
iteration.
B
Yes,
okay,
I
work
with
credit
offline
about
the
timeline
protection
for
distribution
when
we
can
start
when
we
expect.
To
finish
I
mean
this
is
all
this
will
be
a
discussion
between
distribution
and
and
our
team.
C
C
B
That
will
definitely
help
please.
Once
you
have
that
issue
up,
I
will
ping
the
distribution
team
to
start
thinking
about
it.
B
Okay,
I
guess
it's
clear
to
me
this
first
bullet
we
are
cleared
here.
Any
other
questions
about
my
question.
B
No,
okay,
oh
by
the
way,
is
a
august,
reasonable
timeline
that
I
I
tell
a
distribution
that
may
be
august
sometime
in
august.
You
will
be
able
to
start
working
on
that.
Is
that
too
optimistic
or
configuration
aspect,
yeah
or
start
distribution
work?
They
can
actually
start
working
on
it.
C
C
So
like,
let's,
let's-
let's
maybe
look
at
this
issue-
let's
see
like
how
much
work
of
that,
and
maybe
we
could
like
they
could
like
because,
like
we
have
like
initial
database,
wi-fi
configuration
as
soon
as
they
like
contribute
a
way
to
configure
that
we're
gonna
be
well
equipped
on
iterating
on
that,
our
own.
C
B
C
B
C
B
Yeah,
thank
you.
Okay,
bullet
2
is
also
mine.
Basically
repeat
the
question:
that's
cpop
in
the
working
group
meeting
this
monday.
I
see
there
was
a
lot
of
discussions
already,
but
basically
the
last
bullet
summary
was
we
don't
know
what
kind
of
sharing
strategy
will
choose
following.
So
this
question
is
basically
we
don't
know
right.
E
B
E
I
think
the
summary
for
me
is
the
following:
is
I
think
it's
important
that
it's
clear
that
we
are
not
pursuing
a
horizontal
sharding
strategy
right
now
we're
doing
decomposition
it
that's
important.
We
have
not
decided
on
a
future
sharding
strategy
if
it's
application
level,
horizontal
charting
or
something
else,
so
the
future
is
is
uncertain
in
that
regard.
E
But
in
any
case,
even
if
we
were
to
choose
horizontal
application
level
charting
this
was
would
not
make
it
more
difficult
or
impossible.
I
think,
on
top
of
it
from
a
database
perspective,
it
is
generally
desirable
to
have
smaller
tables,
so
we
I
think,
janis
has
on
his
like
wish
list
below
100
gigabytes.
E
So
this
drops
500
gigabytes
of
data,
which
is
really
good
and
also
I
think
there
was.
I
don't
know
if
that's
true,
but
you
know
we
have
relations
in
the
database,
so
we
have
to
normalize
at
some
point.
So
I
think
the
concern
is
not.
There
shouldn't
be
a
concern
really.
At
this
point,
I
think.
D
Now,
if
these
tables
are
very
small,
like
we're
talking
away,
10
20
rows,
something
like
that,
let's
say
application
settings
which
has
one
row
that
you
could
replicate
through
all
these
shards
if
you
wanted
to,
but
for
tables
like
users
that
gets
a
bit
annoying,
especially
if
you
have
a
couple
of
million,
but
that's
totally
fine.
As
long
as
we
don't
join
on
them,
use
separate
queries.
It's
totally
fine
to
get
your
customer
data
from
chart
a
and
then
say
hey
for
these.
D
A
Yeah,
the
simple
answer
to
it
is
no.
This
does
not
make
future
sharding
whenever
we
get
to
that
point
more
difficult.
E
E
C
Actually,
like
our
practice
from
last
month,
shows
that,
like
we
actually
picking
right,
iteration
like
this
iteration
alone
is
pretty
complex
to
like
to
perform
and
like
doing
the
horizontal
sharding
could
be
really
disastrous
to
do
so.
Whatever
we
do
later,
it's
just
gonna
be
significantly
simpler
to
do
because
of
like
breaking
a
lot
of
these
monolith
structures
already.
B
Cool
okay
third
is
also
my
chest
forward.
Looking
that
that
one
doesn't
make
sense
to
compile
a
list
of
the
the
work
that
can
be
found
out
in
the
future,
sometimes
down
the
road,
I
will
engage
with
other
teams
to
get
them
started
once
they
are
unblocked,
so
fabian.
E
My
understanding
of
this
was
that
we
kind
of
talked
about
what
we
know
already,
where
we
can
find
out
work
like
the
distribution,
piece
geo
and
backup
and
restore
which
maybe
will
be
fanned
out
but
are
not
necessary.
I
think
other
than
that.
E
If
I
understood
correctly,
some
things
may
become
more
clear
once
we
have
the
the
poc
green
and
we
can
find
it
out
at
that
point,
but
I'm
not
sure
we
have
enough
knowledge
beyond
what
we
just
discussed
to
say,
because
the
I-team
will
have
to
do
all
of
those
15
things
and
do
them
now.
But
camille
keep
me
honest
here.
C
B
C
Like
to
answer
these
questions,
but
I
think
it's
too
early,
I
I
think
we
are
still
working
on
these
fundamental
blocks,
so
turn
like.
I
think
we
will
have
this
answer
with
like
more
items
like
that
are
better
defined
and
the
scope
like
like
in
the
next
few
weeks
until
the
end
of
this
milestone
so
like
from
my
perspective,
not
yet,
but
I
I
hope
that
we're
gonna
have
this
like
a
stream
of
the
things
to
fix
like
in
the
next
weeks,
gotcha,
okay,
so.
B
B
Yeah
understood
so
my
point
is
just
basically
bring
this
to
everybody.
Is
you
know
over
make
everybody
aware
that
we
can
leverage
other
teams
once
we
identify
those
works
that
can
be
find
out,
identify
that
articulate
and
quick,
and
I
we
can
help
to
find
that
out
to
get
other
teams
committed
to
those
those
work
so
always
think
about
it's,
not
only
in
the
starting
team.
We
have
the
whole
engineering
team
available
to
work
on
this
charity.
B
Do
we
need
to
any
more
discussion
needed
for
this
number
four.
E
I
don't
believe
so
at
this
moment.
I
think
this
is
something
that
was
put
into
the
appropriate
issues
and
they're
still
being
discussed.
That's
my
understanding,
but
that's
a.
I
think
this
is
also
maybe
an
example
of
there
are
difficult
decisions
to
make
and
we
need
to
discuss
them,
and
sometimes
that
involves
also
the
database
team
and
janice
and
andreas,
and
so
we
are
not.
E
B
A
You
is
this
blocking
anything,
or
is
this
a
future
discussion
like?
Are
we
waiting
on
a
decision
here
before
we
can
move
forward.
C
No,
I
think
we
are
at
the
point
that
this
is
not
working.
Then
we
can
continue
our
order.
It's
feel
like
it
still.
That
means
that,
like
we
need
to
figure
out
how
to
deal,
but
the
main
problem
right
now
is
like
we
are
dealing
with
a
lot
of
answer
to
anything
around
different
aspects,
so
it's
very
hard
to
like
provide
a
golden
solution,
yeah
this
amount
of
the
arsenity
so
yeah,
that's.
C
Why,
like
I,
I
guess,
like
speed
and
conquer,
or
something
like
that,
to
figure
out
like
how
to
tackle
these
smaller
items.
C
C
E
So
sounds
excellent
sounds
very,
very
fun
to
focus
on
something
that
has
nothing
to
do
with
which
side
you
know,
yeah.
E
Yes,
cool
cool
on
the
note
of
thing
people
being
away,
I'm
flying
to
the
us
tomorrow,
so
I
will
be
likely
mostly
out
on
friday,
but
then
I'll
be
on
us
time
zones.
So
it's
going
to
be
well,
you
know,
like
the
perspective,
may
vary
I'll,
have
more
time
with
overlap
with
youtube
and
and
craig
and
not
as
much
with
janus
york
and
and
camille.
B
Oh,
by
the
way
before
everybody
goes
so
there
will
be
a
meeting
at
8.
00
am
today
to
review
my
proposal
with
eric.
Basically,
I
proposed
that
we
want
more
sres,
but
really
the
infrastructure
pushed
back,
and
they
cannot
do
that.
So
a
compromise
to
compromise
the
plan
is
that
they
have
that
they
have
reserved
the
capacity
in
q3
and
we
have
nick
and
jose
to
work
together
with
us
to
break
down
the
sre
work
and
also
generate
requirements.
B
That's
the
requirements
right!
That's
the
plan
right
now.
If
anyone
wants
to
join
that
media,
let
me
know
I
will
invite
you
it's
basically
review
with
eric.
I
to
be
honest,
I'm
not
too
confident
on
this
compromise.
The
plan
to
you
know
either
boost
the
confidence
level
of
the
january
timeline
or
even
better
pull
the
timeline
forward.
I'm
not
confident
on
this
plan,
because
the
all
the
sres
are
part
part-time
on
this,
so
they
are
not
committed
to
charting,
but
we
do
need
I
from
what
I
heard.
A
So
I
asked
dylan
to
create
the
readiness
review
issue
with
sre
and
I'm
going
to
talk
to
brent
here
shortly
about
it.
So
it'll
give
them
more
context.
We
may
have
used
the
wrong
template
because
they
have
two:
we
picked
one,
but
it's
okay.
If
we
can
get
the
we'll,
have
more
information
to
sre
shortly
and
yeah,
we'll
just
keep
iterating
with
them
and
get
the
confidence
level
higher.
E
But
correct
me:
if
I'm
wrong,
my
my
understanding
is
that
the
team
actually
has
what
they
need
right
now
with
the
support
given
the
state
of
the
of
the
project.
I
think
we
know
that
we
will
need
more
support
at
a
future
point
in
time,
but
I
I
think
the
concern
or
the
main
question
is:
will
we
be
able
to
do
this
by
october?
E
B
Okay,
so
right
now
the
goal
is
add
more
confidence
of
the
january
end
of
january
timeline.
We
projected
roughly
that
I
described
it
as
50
confidence
to
deliver
the
first
iteration
by
end
of
january
and
now
so
we're
no
longer
talking
about
october
yeah,
no
longer
talking
about
october.
B
A
A
I
don't
think
we're
blocked
on
anything
on
sre
and
nick
has
the
ability
to
provision
environments
for
us
if
we
need
to,
and
he
is
dedicated
to
the
sharding
group
to
a
certain
degree,
so
we're
in
a
better
spot
but
to
fabian's
point.
I
agree.
I
don't
think
we're
waiting
on
anything
for
sre
right
now,
but
we
do
need.
A
C
Oh,
I
I
think
you
are
correct,
like
it
would
be
welcome
to
half
person
but
like
if
we
not
have
like
for
another
month.
I
think
nothing
terrible
will
happen
because
of
that
just
there
is
like
a
lot
of
planning
ahead
on
the
sre
work
and
the
sooner
they
they
start
doing
that
the
better
and
better
organized
it
will
be
no,
but,
like
I,
I
think,
like
it's
still
like.
We
still
have
this
problem
with
the
high
uncertainty
on
a
lot
of
things.
C
B
How
about
this
we
are
at
the
time
I'll,
add
camille
and
robin
to
that
meeting.
If
you
cannot
join,
that's
fine
totally
fine,
but
I
will
add
you
to
the
meeting.
If
you
can
join,
please
join
so
we
can
get
a
full
context
to
eric,
because
right
now
eric
doesn't
have
the
full
context.
I
I
don't
think
he
has
that,
so
we
want
to
present
the
full
picture
to
him.
I
will
try
my
best
to
describe,
but
if
you
can
join
that
meeting,
welcome.