►
From YouTube: GitLab 13.12 Kickoff - Enablement:Memory
Description
Kickoff for the 13.12 release for the Memory team.
Planning issue: https://gitlab.com/gitlab-org/memory-team/team-tasks/-/issues/93
A
A
So
we're
planning
ambitiously
regardless
of
that.
So
I
walk
you
through
all
of
those
specific
items
here.
Just
keep
that
in
mind.
Given
the
memory
team
structure,
we
can
often
provide
a
lot
of
valuable
input
to
those
challenging
projects,
so
the
first
one
is
removing
unicorn
support.
A
This
is
going
to
happen
in
14.0,
so
we're
planning
to
do
a
lot
of
work
to
remove
unicorn
completely
from
from
the
code
base.
But
as
a
quick
reminder,
why
are
we
doing
this?
So
unicorn
was
the
default
web
server
for
git
lab
that
was
replaced
by
puma
in
13.0
and
since
then
unicorn
has
been
deprecated,
and
so
the
main
reason
was
the
switch
to
pumbaa
is
puma
is
multi-threaded.
A
It
uses
less
memory
and
overall
has
a
lot
of
performance
benefits,
so
we're
we're
going
to
actually
remove
unicorn
completely
by
14.0,
so
you
must
upgrade
to
puma
and
we're
going
to
do
a
lot
of
cleanup,
making
sure
that
we
don't
ship
components
and
configuration
that
we
don't
need
anymore.
A
Another
item
that
we've
worked
on
is
reducing
the
memory.
Consumption
for
specific
puma
and
cyclic
end
points,
so
these
are
very
expensive
endpoints
in
in
gitlab.
That
are,
you
know,
expensive
to
run
on
gitlab.com
specifically,
but
any
changes
that
we
make
here
benefit
the
product
as
a
whole
and
so
to
highlight
a
few
findings
that
the
team
actually
established.
Looking
at
these
expensive
endpoints
is
that
there
are
a
fair
number
of
inefficiencies
in
the
product.
A
A
So
there's
a
lot
of
work
that
the
team
can
actually
start
doing
now,
having
identified
some
of
these
inefficiencies
and
being
able
to
reason
about
them,
so
I'm
quite
excited
because
changes
that
we
make
here
can
really
improve
the
performance
of
gitlab
at
any
scale
right.
It
may
be
most
obvious
for
larger
installations,
but
rendering
markdown
faster,
I
think,
will
generally
will
be
generally
appreciated
by
many
users
yeah.
So
we
are
going
to
do
that
and
I'll
keep.
You
posted
quite
excited
about
this
work
here.
A
The
next
item
I'm
going
to
talk
about
is
a
few
follow-up
items
from
shifting
the
database
load
from
primary
to
read
replicas,
specifically
for
sidekick.
So
this
was
part
of
a
rapid
action.
We've
already
done
a
bunch
of
work
in
this
and
we're
going
to
now
do
a
bit
of
follow-up
wrap-up
actions
in
here,
and
this
goes
sort
of
hand-in-hand
with
the
last
item
I'm
going
to
talk
about,
which
is
to
move
some
sidekick
workers
to
utilize
database
replicas.
A
And
that's
it
keep
in
mind
that,
as
I
said
earlier,
there
may
be
some
other
items
that
the
team
will
be
working
on.
But
this
is
14.0
and
I'm
actually
quite
excited.