►
From YouTube: Scalability team demo 2023-09-14
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
Yes,
he's
not
online
at
the
moment,
it's
been
some
enough.
This
morning,
yeah
they've
been
so
many
changes
in
town
land.
It's
it's
really
it.
It's
really
coming
a
long
way
quite
quickly
and
it's
quite
encouraging
to
see
yeah
just
to
see
all
the
things
that
are
changing
and
we're
moving
towards
getting
this
ready
for
dedicated
and
there's
a
blueprint
of
people
want
to
contribute
to
that
for
how
we
will
put
that
in
yeah.
A
What's
really
interesting
about
the
getting
it
ready
for
dedicated,
is
it
this
balance
between
what's
good
from
a
technical
perspective,
and
how
do
we
get
this
to
work
from
a
process
perspective,
because
the
dedicated
team,
just
don't
really
have
the
capacity
to
be
reviewing
all
of
the
reports?
So
how
do
we
make
a
choice
that
actually
is
a
good
balance?
A
C
I
heard
quiet
Rumblings
about
a
router
that
was
used
to
ask
access
some
of
the
public
git
lab
instances
that
are
out
there.
I,
don't
know
if
you
have
a
demo
of
that
Andrew
or
anything
you
can
show,
but
that
seemed
fascinating
and
the
kind
of
next-gen
router
stuff.
D
That
was
ages
ago,
and
it
hasn't
really
got
any
traction.
It
was
more
like
just
showing
I
I,
don't
know,
I,
don't
know
what
the
state
of
all
that
is.
I.
Think
we
really
okay,
I
mean
what
it
is,
is
I'll
just
find
what
is
it
called
again:
stateless,
router,
stateless
route?
Sorry
I've
got
to
find
it
quick.
D
There
was
this
discussion
around
like
how
we're
gonna
root
traffic
between
different
orgs
and
like
whether
you
know
what
what
this
so
so,
basically
for
cells.
There's
kind
of
this
idea
of
of
this
big,
stateless,
router
I
think
some
people
have
called
it
the
Sorting
Hat
and
like
when
you
go
to
gitlab.com
a
request
comes
in
and
the
request
can
go
to
like
one
of
many
cells
and
there's
a
little
bit
of
logic
in
the
stateless
router.
D
That
kind
of
determines,
which
cell
to
send
your
request
to
based
on
this
concept
of
an
organization
which
is
sort
of
one
of
the
so
effectively
an
organization
sort
of
sits
between
like
a
git
lab
instance,
and
a
namespace.
So
there's
you
know
to
one
gitlab
instance
or
one
cell.
D
There
are
many
organizations
and
then
from
an
organization
there
are
many
namespaces,
so
like
gitlab
organ
gitlab
com
and,
like
all
the
other
sort
of
top
level,
gitlab
namespaces
would
all
be
inside
a
single
organization
and
and
they
would
be
owned
by
a
single
cell
I.
Don't
know
if
that's
kind
of
like
general
knowledge
yet,
but
that
that's
kind
of
my
understanding
of
it
and
then
like
when
a
request
comes
in.
We
recognize
it
through
some
magic
process
which
I
I
don't
fully
understand
yet
as
being
owned
by
a
particular
cell.
D
Like
my
suspicion
is
that
there
will
be
like
a
query
that
goes
out,
and
then
the
stateless
router
will
do
a
lot
of
caching,
so
it
will
kind
of
if
it
doesn't
know
where
to
send
something.
It'll
go
to
a
service
and
it'll
say
you
know,
which
git
lab
instance
should
I
send
this
to
and
which
cell
should
I.
Send
this
to
it'll
get
a
response,
and
then
you
know
that
can
be
cached
for
a
very
long
time,
and
so
we
can
kind
of
just
have
this.
D
This
long-term
cash
here,
that's
kind
of
sending
requests
to
to
different
places
depending
on
on
situations
and
so
kind
of
we
were.
In
that
conversation,
there
was
a
conversation
about
how
that
stateless
router
would
be
implemented,
and
one
idea
that
was
was
discussed
was
like
well.
Why
don't?
We
use
Cloud
format
and
a
cloud
formation
cloudflare
workers,
because
it's
pretty
lightweight
it's
pretty
cost
effective.
D
It's
pretty
simple
and
you
know,
let's
see
if
we
can
implement
the
logic
in
that
to
do
this,
and
so
has
anyone
I
don't
know
if
anyone's
used
Cloud
play
workers
at
all.
It's
it's
quite
a
clever
API,
because
effectively
the
API
is
the
same
API
that
you
use
for
service
workers.
So
if
you're
a
front-end
engineer
you
can,
you
can
build
something
called
a
service
worker
and
basically,
when
your
browser
makes
the
request,
you
can
kind
of
manipulate
it
and
do
different
things
with
it
and
it
runs
inside
your
browser.
D
But
if
you
go
to
inspects
and
you
go
to
application,
you
can
see
I,
don't
have
any
service
workers
because
I've
disabled
them
because
it
breaks
Google
Docs,
but
anyway,
these
service
workers.
It's
like
a
browser,
API,
that's
kind
of
Well
published,
but
what
you
can
do
is
you
can
basically
push
that
to
cloudflare
and
run
it
on
their
servers.
So,
instead
of
it
running
in
your
browser,
it
runs
like
on
on
the
internet
and
it
basically
runs
at
the
edge.
D
So
wherever
the
the
user
is,
the
service
worker
will
will
instantiate
there,
and
it's
like
you,
know,
cloudflare
Global,
Network
and
the
other
Edge,
and
so
it's
pretty
cool
stuff,
and
so
just
as
a
demonstration
I
committed
to
to
building
this
little
demo,
which
I
did
and
this
project
is
it.
So
let's
go
take
a
look
at
the
source,
so
it's
not
very
exciting
but
effectively.
D
This
is
this
is
the
interface
for
a
service
worker?
It's
really
simple.
It's
like,
if
you
have
a
request
coming
in
sorry.
D
This
is
typescript
as
well,
so
service
workers
have
to
be
written
in
JavaScript
or
typescript,
because
that's
the
the
API
of
service
workers
and
also
cloudflare
workers,
and
so
you
can
say
you
know
if,
if
a
request
is
requested,
you
know
basically
return
a
response
or
an
asynchronous
promise
to
a
response,
and
then
you
can
kind
of
do
whatever
you
want
and
for
the
purposes
of
the
demo
at
basically
hard-coded
it.
You
know
so
there's
no
cash,
there's
no
lookup,
but
that
would
be
just
the
next
step.
D
It
was
more
like
just
a
really
really
basic
POC
and
what
it
did
was.
It
said
you
know
so
basically
pause
the
first
part
of
the
namespace
and
if
it's
gitlab
com
send
it
to
this
gitlab
instance,
if
it's
X5
X
org
foundation,
so
it's
kind
of
like
comments,
organizations
right
like
so
depending
on
the
namespace.
It's
going
to
Route
the
request
to
different
git
lab
instance.
It's
just
purely
to
see
like
what
the
what
this
API
is
like
and
how
it
works,
and
everything
like
that.
D
So,
basically,
you
can
just
deploy
it
to
cloudflare
and
I
wouldn't
be
surprised
if
it's
still
running
so
if
I
go
so
the
the
the
service
worker
is
deployed
to
this
URL
over
here,
stateless
router
and
if
I
click
on
this,
it
might
still
be
running.
It's
been
a
few
months
yeah
there
you
go
so
this
gitlab
com.
So
if
I
look
at
this
URL
over
here,
you
can
see
it's
you
know
imagine
this
was
imagine.
D
Gitlab.Com
was
pointing
at
this
and
it
wasn't
statelessruiter
dot
Andrew
right,
it
was
actually
just
gitlab.com
gitlab.com
when
I
when
I
look
here.
This
is
this
is
actually
staging
I,
don't
know
where
I
can
see
that
this
is
staging,
but
it
is
it
is.
This
is
our
staging
instance.
Now,
if
I
go
and
edit
the
URL
sorry
I'm,
also
using
a
weird
browser,
I,
don't
know
if
I
go
into
here
and
I
go
to.
You
know
x
foundation.
D
This
is
now
you
know
the
same
endpoint
over
here,
but
it's
rooting.
The
requests
to
a
different
gitlab
instance.
Obviously,
I'm
using
these
public
gitlab
instances
in
lieu
of
cells
because
we
don't
have
cells
yet.
But
you
know
if
I
edit
this
URL
and
go
back
to
gitlab.com
and
if
you
know
it's
the
same,
the
same
endpoint,
but
it's
now
routing
it
to
staging.getlab.com
and
then
I
forgot,
which
other
ones
I
had
in
there.
D
Let's
see
if
it
works,
as
a
video
line
looks
like
you
need
to
be
signed
in
which
it
wasn't
at
the
time.
Maybe
they
got
sick
of
people
sending
random
weird
requests
to
their
to
the
gitlab
instance
and
then
Debian.
So
if
I
go,
you
know
again
forward
slash
Debian
over
here
and
you
know
so.
D
But
it
was
it's
a
very
straightforward
demo.
It's
really
small,
really
simple,
where
it
gets
complicated
is,
is
apis
and
because,
when
I
was
when
you
make
a
request
here,
if
I
go,
look
at
this
you'll
see
under
Network
I,
just
reload.
This.
D
To
a
project,
page
I
think
there's
quite
a
few.
So
if
I
go
into
here,
here's
here's
a
classic
example.
So
this
URL
is,
you
know
API
graphql,
so
which
endpoint
sorry
I
can
see.
You're
struggling
a
bit
there
Bob
here
we
go,
you
know
so
this
over
here
has
got
no
routing
information
like
the
the
the
the
router
doesn't
know
where
to
send
it
to,
and
so
I
just
did
a
real
hack.
You
can
see
it's
not
even
working
anyway,
so
clearly,
it's
not
as
good
as
I
hoped,
but
wait.
D
D
It
says
where
it's
coming
from,
and
so
we
know
that
you
know
this
API
graphql
is
going
is
coming
from
this
over
here
and
I
can
so
I
can
extract
that
from
the
referrer
and
then
I
know
which
gitlab
instance
to
send
that
graphql
request
to
if
I
look
here
is
this
and
then
the
same
with
this
API
request?
D
So
here
you
can
see
it's
got
the
tags
Theory,
let's
see
if
it's
got
yeah,
so
it's
got
the
branches
and
the
tags
that
actually
that
worked
I,
don't
know
why
this
isn't
working
but
doesn't
seem
to
be
working
and
that's
basically
the
demo
I
don't
know.
If
anyone
has
any
questions.
How
did
you
hear
about
that?
Sam.
C
We
were
talking
the
the
delivery
strategy
and
the
what
level
delivery
can
commit
to,
and
especially
in
terms
of
the
router
does
that
evolve
into
a
separate.
You
know
router
team,
who
cover
the
the
entry
point
for
gitlab.com
and
have
to
read
from
some
kind
of
manifest,
which
says
here
are
the
cells.
D
Yeah
I
think
I
think
I
think
that
that's
gonna,
that's
gonna,
ultimately
be
a
team.
That's
you
know
at
least
initially
like
it
would
probably
be
the
sales
team
that
build
that
out.
But
you
know
over
the
longer
term,
I
mean
I,
don't
know.
Maybe
it
won't
change
that
much.
You
know
once
who
knows
it's
difficult
to
say
at
this
point,
but,
but
you
know,
I
would
probably
presume
that
that
starts
off
as
the
sales
team
and
or
that
organization
with
support
from
from
us.
D
It's
obviously
not
something
you
can
run
in
Runway,
it's
a
very
specific
specialized
thing,
but
like
it
also
was
really
nice
to
see
how
these
Cloud
workers
work
and
they
they
they're
pretty
cool.
They
do
what
they
say
on
the
tin
and
they
are
yes
four
months
ago.
So
it's
been
a
while
since
I
looked
at
this,
but
they
they
work.
D
They
work
really
really
nicely,
and
you
know
when,
when
I'm
clicking
around
on
this
now
it's
going
through,
like
the
local
Cape
Town
pop,
so
you
know
there's
it
really
helps
reduce
some
of
the
some
of
the
extra
latencies.
Obviously,
there's
still,
you
know
other
things
to
do
with
it,
but
the
fact
is
we're
already
cloudflare
customers,
so
it
might
make
a
lot
of
sense.
C
D
Yeah
yeah,
so
they
sorry
I,
thought.
You
said
the
cell
I
was
like
okay,
yeah
yeah,
so
yeah
I
mean
it's
it's
it's
it's
a
pretty
cool
technology
and
you
know,
cloudflare
are
also
a
very
they've,
got
much
better
Network
and
egress
and
policies
and
and
billing
policies
than
almost
anyone
else
in
the
at
the
scalar
industry,
which
is
which
is
a
positive
thing,
but
yeah.
D
But
it's
it
kind
of
like
we
haven't
spoken
about
it
much
since
then,
it
was
kind
of
like
it
was
done
as
part
of
this
off-site
and
demo
was
done
and
kind
of
waiting
for
it
to
to
come
back
into,
but
we
really
need
to
have
a
fastboot,
that's
kind
of
the.
We
all
need
to
get
in
a
room
and
talk
about
these
things
with
all
the
different
teams
that
are
involved
in
in
all
of
this
effort.
So
that's
a
different
conversation,
though.
D
Yeah
they
they
yeah.
That's
it's
it's
very
cool.
It's
pretty
impressive
stuff,
I
mean
I
still,
don't
know
what
we'll
do
with
like
API
graphql
and
those
kind
of
things
and
all
those.
B
C
D
B
D
D
No,
no
so
like
actually
knowing
that
like
API
V4
projects,
you
know
so
so,
knowing
all
the
different
like
paths
and
then
knowing
what
the
identifiers
in
those
paths
are
and.
B
D
D
Dictionary,
but
can
you
all
right,
no,
no,
no
yeah,
so
so,
but
the
fetch
to
the
because
it's
not
redirecting
to
a
particular
instance
right.
It's
it's
doing
the
fetch.
So
it's
doing
a
proxy.
You
know
so
I'm.
Just
the
the
easiest
thing
for
me
to
do
here
is
just
to
kind
of
dump
the
the
output
out
so
basically
copy
the
original
response
out
to
the
to
the
customer,
but
I
could
all
to
the
user,
but
I
could
just
as
easily
consume
that
and
then
make
another
request
right.
D
C
C
D
Don't
want
to
be
like
if,
if
you
building
your
whole
app
in
there,
then
yeah
then
like
what
we
want
to
do
is
be
it's
like
a
thin
layer
right,
so
yeah,
here's,
the
cache,
API
so
yeah
and
I
and
I
think
it
kind
of
respects
like
the
global
kind
of
I.
Don't
know
how
it
works
globally
and
stuff,
but
there's
elements
to.
B
D
I
was
also
kind
of
interested
to
see
like
the
performance
like
because
really
what
you
want
to
do
is
you
want
to
kind
of
stream
your
requests
right,
so
you
don't
want
to
you,
don't
get
a
request
and
then
go
to
like
staging.gitlab.com
kind
of
read
the
request
in
and
then
pass
the
output
back.
You
want
to
kind
of
connect
the
pipes
and
say,
okay,
and
so
that
was
something
else.
That's
I
mean
it
works
really.
D
Well,
it's
it's
good
stuff
and
it's
another
excuse
for
meteorites
some
JavaScript
and
get
that
out
of
my
system.
Every
now
and
again.
C
Well,
it's
been
dropped
from
Triple,
H's
or
dhh's
latest
project.
Doesn't
it
now
typescript.
C
D
To
see
yeah
all
the
things
that
could
be
possible
I
mean
the
other
thing.
That's
I
was
wondering
about
assets
and
stuff
like
that
that
you
know
it's,
it's
very
cool
stuff.
D
It's
just
it's
just
another
option.
Another
thing
that
we
might
like,
as
as
everything
kind
of
develops,
it
might
be
a
tool
that
we
may
use-
or
maybe
you
know
it
makes
sense
to
have
like
I,
don't
know
a
big
go
service
running
somewhere
in
a
certain
cloud
and
we've
passed
everything
through
that.
Like
you
know,
this
isn't
nothing
settled
at
the
moment,
but
there's
just
one
option.