Boomerang: How fast do users think your site is?

Introduction
How does it work?
Using boomerang
Data

Boomerang: How fast do users think your site
is?

Philip Tellis / philip@bluesmoon.info

ConFoo.ca/2011 – 2011.03.09-11

ConFoo.ca/2011 – 2011.03.09-11 Boomerang: How fast do users think your site is?

Introduction
How does it work?
Using boomerang
Data

$ ﬁnger philip

Philip Tellis
philip@bluesmoon.info
@bluesmoon
geek - paranoid - speedfreak
yahoo
http://bluesmoon.info/


Introduction Time
How does it work? The adversary
Using boomerang Measure twice
Data boomerang


Introduction Time
Data boomerang

Only 10–20% of page load time is spent on systems we control


Introduction Time
Data boomerang

It’s what we can’t control that users notice

http://twitter.com/stoyanstefanov/status/43747252317077504


Introduction Time
Data boomerang

We can’t control

browsers


Introduction Time
Data boomerang

We can’t control

plugins


Introduction Time
Data boomerang

We can’t control

OSes


Introduction Time
Data boomerang

We can’t control

viruses


Introduction Time
Data boomerang

We can’t control

antiviruses


Introduction Time
Data boomerang

We can’t control

microwaves


Introduction Time
Data boomerang

We can’t control

baby monitors


Introduction Time
Data boomerang

We can’t control

naughty neighbours


Introduction Time
Data boomerang

We can’t control

ﬁle shares


Introduction Time
Data boomerang

We can’t control

governments


Introduction Time
Data boomerang

We can’t control

rodents


Introduction Time
Data boomerang

We can’t control

Try simulating all that in the lab!


Introduction Time
Data boomerang

We need to measure real end-user performance


Introduction Time
Data boomerang

We need to measure real end-user performance from the real
end-user’s perspective


Introduction Time
Data boomerang

While this might work, it isn’t necessarily representative


Introduction Time
Data boomerang

What about JavaScript?


Introduction Time
Data boomerang

boomerang is...
A piece of javascript that you add to your web page where it
measures and beacons back to you, the end user’s perceived
performance of your page


Introduction Time
Data boomerang

How?

<script src="boomerang.js" type="text/javascript">
</script>
<script type="text/javascript">
BOOMR.init({
user_ip: "<network ident>",
beacon_url: "http://mysite.com/beacon.php"
});
</script>


Introduction Time
Data boomerang

What does it do?

Measures user’s network throughput and latency to your
server
Measures the current page’s load time
Beacons these results back to your server


Introduction Latency
How does it work? Bandwidth/Throughput
Using boomerang Load time
Data Accuracy

How does boomerang work?


Data Accuracy

Let’s take that one at a time


Data Accuracy

Measuring latency

Download a 32 byte gif 10 times in sequence
Measure the time to download each
Discard the ﬁrst measurement because it’s overpriced9
Calculate the arithmetic mean, standard deviation and
margin of error of the rest


Data Accuracy

Measuring throughput

After the latency test is done, we download progressively
larger images
Stop at the ﬁrst image that times out
Redownload that image a few more times
Calculate the median, standard deviation and margin of
error of the largest image


Data Accuracy

Measuring latency before throughput helps here

Those 10 latency images do a lot to widen the TCP
window size
The bandwidth images make much better use of available
bandwidth
The image we end with makes the best use of bandwidth


Data Accuracy

How do we measure page load time?

In the onbeforeunload event, measure the time and
store it in a cookie
In the onload event, check the cookie, and measure the
difference with the current time
We also make sure that the page that set the cookie is the
referrer of the current page


Data Accuracy

What? Two pages?

Yes, this needs two pages and cookies. If those aren’t
supported, we try to use the WebTiming API10 .


Data Accuracy

How accurate is it?

Latency measurements are very accurate (±1%)
Bandwidth is to an order of magnitude. For bad
connections can be ±30%
Page load time sometimes has outliers, you need
post-ﬁltering
The margin of error tells you how good your data is


Introduction Basic
How does it work? In-page timers
Using boomerang AJAX
Data

Include it on your page

<script src="boomerang.js" type="text/javascript">
</script>
<script type="text/javascript">
BOOMR.init({
beacon_url: "http://mysite.com/beacon.php"
});
</script>


Introduction Basic
Data

For most sites, that’s about it


Introduction Basic
Data

You probably want to do more


Introduction Basic
Data

Measure more than just load time

<html><head>
<script>
var t_pagestart=new Date().getTime();
</script>
...
<script>
var th=new Date().getTime();
</script>
</head>
<body>
...
<script>
var tj=new Date().getTime();
</script>
... ConFoo.ca/2011 – 2011.03.09-11 Boomerang: How fast do users think your site is?

Introduction Basic
Data

...
<script src="boomerang.js"></script>
...
var te=new Date().getTime();
BOOMR.plugins.RT.setTimer("t_head", th-t_pagestart).
setTimer("t_body", te-th).
setTimer("t_js", te-tj);
</script></body></html>


Introduction Basic
Data

This adds the t_head, t_body and t_js ﬁelds to the beacon


Introduction Basic
Data

Loading dynamic content

BOOMR.init({
beacon_url: "http://mysite.com/beacon.php",
auto_run: false
});


Introduction Basic
Data


// Just before download starts
BOOMR.plugins.RT.startTimer("t_done");


Introduction Basic
Data


// Just after download finishes
BOOMR.plugins.RT.done();


Introduction Basic
Data

Much more

http://yahoo.github.com/boomerang/doc/howtos/


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

The beacon

GET request to the beacon URL (response ignored)
All parameters passed in the query string
Extra timers are passed in as a comma separated list in
t_other
before_beacon JavaScript event ﬁred just before the
beacon is sent


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

What should we do with the data?

Sanity checking to:
Remove fake data
Remove abusive data
Maybe just rate limiting


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

What can we do with the data?

Statistical analysis to:
Remove outliers
Aggregate based on bandwidth blocks
Measure trends over time and correlate them with code
changes


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

Bandwidth blocks


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

Bandwidth blocks

Data points from some countries may require narrower bands


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

Geographic data

Looking at latency from different geographic locations can tell
you where your next mirror should be


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

ISPs

Grouping data by ISP can tell you who’s behaving badly


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

More data

Write plugins to get more performance data
We already have a DNS plugin
I’ve just completed a plugin to measure IPv6 support and
latency
What about a full WebTiming plugin?
I’m also looking at measuring connection setup time


Introduction
shoulda
How does it work?
coulda
Using boomerang
woulda
Data

You decide

Once you have the data, you can do anything with it


Introduction
How does it work?
Using boomerang
Data

Boomerang: It (almost) always comes back
http://github.com/yahoo/boomerang

http://yahoo.github.com/boomerang/doc/


Introduction
How does it work?
Using boomerang
Data

Photo credits

ﬂickr.com/photos/21233184@N02/4389412851
http://xkcd.com/445/


Introduction
How does it work?
Using boomerang
Data

Contact me

Philip Tellis
philip@bluesmoon.info
@bluesmoon
geek - paranoid - speedfreak
yahoo
http://bluesmoon.info/
slideshare.net/bluesmoon
joind.in/talk/view/2913


Introduction
How does it work?
Using boomerang
Data

References

github.com/yahoo/boomerang
More bandwidth doesn’t matter (much) – Mike Belshe
Analysing Bandwidth & Latency – YUI Blog
It’s the latency, stupid – Stuart Cheshire
The statistics of web performance


Introduction
How does it work?
Using boomerang
Data

Overpriced images

The ﬁrst image might require a DNS lookup and TCP
handshake
Slow start is not an issue since 32 bytes ﬁt in 1 packet


Introduction
How does it work?
Using boomerang
Data

WebTiming API

JavaScript interface to network timing information including
DNS, TCP connect, download, etc. Supported by:
Microsoft Internet Explorer 9
Google Chrome (WebKit)
W3C speciﬁcation


Boomerang: How fast do users think your site is?

Related slideshows

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

Recommended for you

More Related Content

More from Philip Tellis

More from Philip Tellis (20)

Recently uploaded

Recently uploaded (20)

Boomerang: How fast do users think your site is?