►
From YouTube: GitLab 15.8 Kickoff - Enablement:Database
Description
The Kickoff Video for the database group's 15.8 milestone
A
B
A
We
really
want
to
use
these
final
stages
to
deliver
specifically
parallel
batches,
so
that
we
and
customers
can
run
these
migrations
more
quickly,
but
especially
for
customers.
That'll
help
with
jump
upgrades
when
they
need
to
finalize
a
bunch
of
migrations
all
at
the
same
time,
instead
of
having
to
run
them
sequentially
like
they
do
now,
they
ideally
can
run
more
than
one
which
would
be
really
nice
and
then
we're
also
finalizing
some
throttling
mechanisms
which
we
hope
will
help
us
to
continue
to
use
the
framework
to
prevent
some
of
the
incidents.
A
We
saw
with
background
migrations
previously
on
gitlab.com,
which
is
very
exciting,
and
then
we
have
a
a
bug
we
encountered
and
we
kind
of
added
into
the
plan
last
Milestone
at
the
end
of
the
Milestone,
but
it's
not
quite
fixed
yet,
but
we're
trying
to
attribute
migrations
to
their
specific
feature
categories.
That
way
we
can
make
sure
teams
are
aware
when
their
migrations
are
failing.
So
that's
exciting
for
CI
partitioning
support.
We're
really
excited
to
help
the
CI
verify
groups
in
their
efforts
to
partition.
The
CI
builds
table
I'll.
A
Let
them
talk
about
all
those
teams.
Talk
a
little
bit
more
about
why
they're
doing
that,
but
from
our
side
we're
providing
helpers
and
and
assistants
with
their
blueprints
in
order
to
figure
out
how
to
get
their
tables
partitioned
a
little
better.
Well,
not
a
little
better
at
all
and
we'll
be
able
to
reuse
those.
We
want
to
be
able
to
reuse
These
Blueprints
for
other
tables
within
gitlab,
so
that
they
can
also
trim
down
their
table
sizes.
This
is
all
the
long-term
goal.
A
B
Think
you're
good
so
far,
I'm
super
excited
to
see
that
this
work
has
been
ongoing
for
quite
some
time
and
is
looking
forward
to
the
ending
line
soon
so
very
exciting
and
I
know.
Migrations
and
partitioning
are
both
things
that
impact
a
lot
of
other
teams.
So
it's
cool
to
see
how
the
impacts
here
can
really
spread
out
across
the
company.
A
Yeah,
the
other
places
we're
looking
to
really
continue
making
an
impact
is,
with
our
database,
query,
analysis
and
review
improvements.
So
these
these
two
efforts
kind
of
have
a
lot
of
overlap
with
each
other,
because
they're
both
trying
to
improve
our
internal
review
processes
around
when
someone
adds
new
queries
or
when
they
add
other
things.
That
would
necessitate
a
review
from
someone
who
is
a
designated
database
reviewer
and
so
from
the
query
analysis
side.
A
It
might
just
be
doing
something
completely
different,
and
so
we're
hoping
to
take
some
of
the
guesswork
out
of
out
of
that
part
of
review
and
identify
immediately
for
authors.
What
queries
that
they
need
to
perform
analysis
on
to
make
sure
that
they,
our
attic,
only
queries
that
meet
our
performance
expectations.
A
The
long
term
for
this
is
we
want
to
run
that
analysis
automatically,
but
in
the
short
term,
we're
we
think
identification
will
get
us
a
lot
of
the
way.
On
the
similar
end,
we
have
a
few
ways
for
trying
to
improve
database
review
in
general,
and
part
of
that
is
gathering
Gathering
feedback
from
both
authors
and
existing
reviewers
and
maintainers,
in
order
to
try
and
figure
out
where
all
the
pain
points
are
and
and
kind
of
make
that
an
easier
transition.
A
A
lot
of
people
get
surprised
by
database
review
and
we
don't
want
that.
We
want
them
to
have
good
good
experiences
where
we
all
understand
that
at
the
end
we
just
want
a
good
performant
getlab.com,
so
we're
database
changes
are
a
bottleneck
for
a
lot
of
different
teams,
and
we
don't
want
that,
and
they
don't
want
that.
So
hopefully
we
can
get
that
going.
A
A
A
So
the
the
load
balancing
is,
is
I'm
I'm
not
totally
up
to
speed
on
this,
but
this
is
this
was
identified.
There
are
several
problems
with
load
balancing
that
were
identified
a
few
months
ago
and
in
a
recent
incident
this
was
targeted
as
a
related
corrective
action
that
we
could
take
so
we're
going
to
pursue
it.
Awesome
sounds
good
and
I'll
fill
all
that
in
there
and
then
the
last
priority
kind
of
topic
is
removing
migrations
in
version
15..
A
Typically
we'd
like
to
do
this
in
15,
0
or
15
1,
but
we've
been
busy
so
and
under
capacity,
but
John
gotta
start
on
this
in
15-7
and
I.
Think
he'll
probably
still
be
working
on
it
in
15
8,
because
it's
going
to
be
a
lot
of
merge,
requests
to
squash
all
those,
but
he's
done
a
great
job
so
far,
automating
that
so
yeah
Roger.
Do
you
have
anything
else?
You
want
to
add.