►
From YouTube: Application performance session - 2020-11-09
Description
Update on LCP Leaderboard + recent CLS topic of project homepage
A
B
B
It's
good
to
hear
yeah,
I'm
also
doing
good.
B
I
hope
that
this
week
will
be
less
intensive
as
last
week,
but
overall,
I'm
not
doing
well
and
happy
to
especially
kick
off
the
q4
now
this
week
in
reality,
as
we
were,
tackling
a
couple
of
other
topics
and
that's
what
I
also
wanted
to
talk
about
today.
But
I
think
we
simply
wait.
One
more
minute
and
then
we
get
started
was
mainly
about
okay,
that
we
have
set
up
for
q4
and
death,
which
is
mainly
around
taking
as
many
routes
as
possible
below
the
lcp
of
2.5
seconds
and
yeah.
B
A
A
B
B
B
B
This
is
just
seven
days
right
now
and
we
can
extend
this
to
the
last
90
days
and
there
we
can
see,
for
example,
a
nice
drop
that
we
especially
had
at
the
beginning,
but
still
we
have
some
bouncing
numbers
which
go
around
400
milliseconds
up
and
down,
especially
for
the
merge
detail
page,
but
this
can
give
us
a
better
insight
quite
fast
to
see
if
stuff
is
just
an
outlier
right
now
or
not,
or
if
there
is
actually
a
problem
or
an
actual
improvement.
B
The
other
thing
is
that
where
we
can
also
target
specifically,
work
is
the
time
to
first
bite.
B
So
we
see
right
here
just
the
time
to
first
bite
so
that
we
can
have
a
better
understanding
if
this
might
be
a
more
of
a
frontend
problem
or
is
more
of
a
backend
problem
and,
for
example,
we
can
see
here
that
the
verified
pipeline
charts
page,
for
example,
is
not
rendering
anything
before
16.4
seconds,
and
the
same
goes
for
a
couple
of
other
routes
that
are
going
very,
very
long,
and
what
we
did
now
is
we
said:
okay,
let's
do
opr
around
it,
which
is
here,
and
we
have
now
the
ocr
to
get
80
percent
of
all
the
53
routes
that
we
have
in
depth
that
we
defined
below
2.5
seconds.
B
So
this
means
we
have
our
own
dashboard
for
that
which
is
here
and
off
we
go
so
here.
What
we
can
see
here
is
all
the
roots
that
we
have
in
depth
and
we
will
basically
try
to
tackle
and
those
roots
as
soon
as
possible,
to
figure
out
what
we
can
do
about
them,
some
even
just
already
started
and
to
see
if
we
can
do
stuff
more
acing.
If
we
need
some
changes
in
the
back-end
side,
we
can
also
see
what
is
taking
right
now
quite
long.
B
A
B
Have
very
bad
queries,
or
we
are
need,
have
the
need
to
basically
make
stuff
async
and
figure
that
out,
because
everything
currently
we
have
defined,
I
think
yeah,
everything
below
500
milliseconds
on
time
to
first
bite
is
okay.
B
We
still
can
get
stuff
accomplished.
I
think
we
take
currently
around
a
second
depending
on
the
page
on
the
front
end
and
so
up
to
like
800
900
milliseconds.
It
still
can
work
that
we
have
an
lcp
below
2.5
seconds,
but,
of
course
the
higher
the
back
end
is
and
the
worse
it
will
get.
Also,
of
course,
on
the
front,
end
yeah
and
that's
how
we
are
going
to
go
into
q4
and
the
q4
just
started
last
week
as
we
are
having
a
fiscal
year,
starting
always
with
february.
A
Just
a
comment:
it's
it
looks
really
useful
to
me
because
I'm
also
a
database
maintainer
and
I
don't
have
to
analyze
complex
queries
and
and
some
of
the
pages.
I
actually
know
that
we
have
actually
backup
problems
because
we
are
either
querying
too
much
data
from
the
database
or
we
execute
way
too
many
queries.
B
Yeah
now
that
would
be
awesome,
especially
to
take
a
look
already
at
the
ones
that
you
know
already
that
are
doing
simply
and
very
heavy
database
queries.
Sometimes
we
even
might
take
a
look
if
we
need
to
get
data
async
or
if
we
need
to
even
remove
data.
I
still
remember
we
had
this
on
the
project
home
page,
where
we
had
one
field
that
took
almost
80
90
of
all
the
time
we
removed
it.
B
No
one
complained:
no
one
was
missing
it,
but
improved
the
performance
quite
nicely
on
the
packet
already,
and
so
I
think
we
have
a
couple
of
options
there
to
get
everything
into
a
nicer
range
cool.
The
other
thing
that
I
wanted
to
talk
about
today
is
the
so-called
cumulative
layout
shift.
B
B
When
does
the
biggest
largest
content
element
on
the
page
is
actually
rendered
and
has
finished
rendering,
and
but
what
is
the
main
focus
overall
on
performance
data
right
now
is
called
so-called
web
vitals
and
there
are
a
couple
of
them
and
another.
One
of
those
which
is
also
quite
important,
especially
to
the
perceived
performance,
is
the
so-called
cumulative
layout
shift,
which
simply
means.
How
much
does
this
page
jump
around
while
it's
loading
and
a
very
good
example
of
that?
It's
jumping
too
much?
B
Is
the
project
homepage
in
that
sense,
and-
and
what
we
can
see
here
is
that
we
have,
we
can
even
make
it
slower,
and
so
what
we
can
see
here.
We
have
the
first
render
the
first
visual
change,
and
now
we
even
have
a
jump
in
there
due
to
some
font
rendering
later
now
we
have
some
fields,
then
something
pops
up.
Another
thing
pops
up-
and
the
last
thing
that
you
can
even
see
down
here,
is
that
the
columns
between
the
loading
state
and
the
actual
render
state
is
different.
B
So
we
have
another
jump,
and
this
basically
makes
the
page
look
very
nervous,
and
this,
of
course
also
means
that
it
feels
much
less
good
performing.
And
what
we
are
looking
at
right
now
is
to
have
a
look
that
we
are
pre-rendering
already
on
the
hammer
pages
most
of
those
elements
and
simply
have
them
inactive
or
empty,
or
just
already
with
a
loading
state
and
then
the
view
application
just
renders
over
it,
but
that
there
is
no
visual
change
between
the
loading
state
of
the
hammer
page.
B
The
loading
state
of
the
viewer
application
application
up
to
the
actual
finally
rendered
thing,
and
this
should
also
reduce
how
fast
it
feels
nicely,
because
it
simply
feels
the
data
is
popping
up
and
then
something
that
that
the
eye
is
not
recognizing
as
good
as
stuff
that
is
moving
around
and
on
other
hand.
Every
jump
that
we
make,
and
that
is
happening
on
the
page-
is
also
decreasing
performance,
because
this
means
that
layout
needs
to
be
re-rendered
recalculated
by
the
browser.
B
And
this
is
something
we
definitely
want
to
prevent
and
reduce
as
much
as
we
can.
So
that
will
be
something
that
the
main
work,
the
main
tool
that
we
have
for,
that
is
simply
render
in
in
html
already
the
loading
state
and
just
replace
it,
then
with
the
application,
and
that
should
be
already
of
nice,
help
there
yeah.
B
Those
were
the
two
topics
that
I
wanted
to
talk
about.
Is
there
anything
that
you
have
adam.
B
Good-
and
I
think
we
have
so
much
work
in
front
of
us
for
q4,
with
the
new
lcp
leaderboard
and
all
the
routes
that
we
have
in
front
of
us
that
we
simply
end
the
session
early.
Today
I
will
post
this.
People
can
take
a
look
at
it
async
and
also
can
leave
questions
async
and
for
the
next
time
we
will
do
some
more
promotion
around
for
the
next
session
good.
Then
I
wish
you
a
really
nice
day
thanks
a
lot
for
joining
and
take
care
and
see
you
soon
in
another
call.