Scaling up YouTube

Do: I'm gonna give you guys
a little kind of peek

behind the scenes of what it's
taken to scale YouTube.

Since this is only
a half an hour talk,

I can't really go
into a whole lot of detail.

But this should give you
some idea

of what it takes to provide

one of the largest sites
in the world.

Okay, so the general kind
of layout here

is that I'm gonna talk about
a little historical context,

set things up.

Talk about a few pieces
of our scalability.

Some of the key challenges
we've faced,

lessons we've learned,
and go into Q and A.

This graph is kind of--

this shape of graph appears
very often.

I guess in the context
of this presentation,

it could be a number of things.

It's probably--

it probably corresponds
to the price of a gallon of gas,

but I'm not gonna talk
about that.

Google has great food, so...

I think that was true,

but I'm not gonna talk about it
either.

I'm still scarred by that.

And--but the thing that it does
actually represent

is the daily video views
over our lifespan.

So it's pretty amazing
the growth that we've had.

It's kind of interesting
when you see

at certain points when
we've added machines or whatnot.

We added capacity,
made things faster.

Added some great features.

Then the graph definitely
spikes up a bit.

Okay, so quick timeline
of YouTube.

February 2005,
we were founded.

We received our first round
of VC funding

in October 2005,
eight months later,

so we got our first office.

And one of the more interesting
scalability challenges we had,

or one of the more interesting
engineering challenges we had

was trying to keep our monitors
dry in our new office.

So the first day that
we moved in it started raining,

which is normally fine
when you're in a building.

Except that there were about,
I don't know,

seven or eight different leaks
in various places.

And we had just gotten
our new, nice,

big 24-inch monitors.

And we didn't really want them

to have any kind of issues
with the water, so we--

After--after maybe
20, 30 minutes

of trying various things,

we took the plastic wrapping

from our IKEA furniture.

We fashioned a funnel
out of that,

and we used the rubber trash can
to catch the water.

We had just enough trash cans
and just enough plastic

that we were able to divert
large--large pools of water

into the trash cans.

And I think I--

I think I drove back once
just to make sure

that the trash cans weren't
over-full.

So definitely interesting times

already during that time.

Just about five months later,

we actually hit 30 million
video playbacks per day.

That was only one year
after we were founded.

So pretty amazing numbers.

And I'll go into a little bit
of details,

you know, what it took
to get there.

A few months later,

we more than tripled
that number.

So pretty amazing.

It was kind of interesting.

We'd add, like, 5%--

you know, 5% or 10% capacity

and we'd get as much if not more
traffic gain from that,

so it's kind of interesting.

A few months later
we got acquired by Google.

And my personal favorite part

is we got a lot of free food.

So that made me quite happy.

The crazy thing is that
by the time we were acquired,

we were already a top ten site

by various external measures.

Like, just top ten period
across all websites.

So pretty amazing considering

that we were only maybe about
1 1/2 years old at that time.

We were number four on Alexa
by Q3 2007

and by other measures as well.

And today I think there's
a new updated number now

but last I heard there was
10 hours of video

uploaded per minute.

It might be more like 13 now.

And hundreds of millions
of video playbacks per day.

It's certainly a lot
of playbacks per day.

I can't give the exact number,
but it's--

It's pretty mind-boggling.

So real quickly,
the pre-acquisition team.

We had two sys admins.

Two software guys,
myself included,

making sure that the site's
scaled.