Numion homepage Measure your speed Measure your website Measure the time it takes to download any page on the internet. A collection of bandwidth calculators. SiteSpeed statistics for this website Copyright 2005 J.C. Kessels Goto Numion homepage Frequently asked questions Interesting Links Contact page, to report a problem or ask a question Copyright 2005 J.C. Kessels


SiteSpeed FAQ

What is measured?
The "Basic" version will measure the time it takes for the visitor's browser to load all the images on the page. The "Frames" version measures all the html and all the images in all the frames, except for the html of the framestart page. The "Links" version will measure all html and all images, and even the time it takes for the server to generate the html.

What is a "unique"?
A "unique" is a measurement by a first-time visitor. If he returns to the same page a little while later then the page will load significantly faster, because the html and pictures will still be in the cache of his browser or proxy-caching server. SiteSpeed therefore differentiates between "unique" (first-time) and "reload" measurements. A measurement is counted as a "unique" if the same visitor has not visited the same TimerID less than 1 hour ago.

What is a "reload"?
A "reload" is a visitor that already has seen the same page less than 1 hour ago. The html and pictures will still be in the cache of his browser or proxy-caching server, and the page will therefore load significantly faster.

What is an "impression"?
An "impression" is a visitor looking at a page. An impression can be a "unique" (first-time visit) or can be a "reload". The "impression" statistics are a join of the "unique" statistics and the "reload" statistics.

Can the SiteSpeed icon be made invisible?
Yes, but I'm not telling. The icon is a little bit of advertising, really, and it would not be in my best interest to tell you how to make it invisible. I would very much appreciate it if you didn't try, the program is free and the very least you can do is give me a little credit by showing my icon.

Can Numion see who is visiting my website?
No. All data is stored on your server. SiteSpeed is not a remote counter that calls the Numion server, but a local counter running on your server.

Can I measure more than one page?
Yes. You can add the SiteSpeed lines (see step 3 on the *  download and install page) to as many pages as you like. The scripts will automatically detect where a measurement is coming from and keep separate statistics for every page.

Can I merge the statistics of multiple pages?
Yes, by explicitly specifying a name in the SiteSpeedIcon() call. This will override the default behaviour where SiteSpeed automatically detects the name of the page. For example, to merge the statistics of "http://www.yourdomain.com/" and "http://www.yourdomain.com/index.html" (which are the same page but have different URL's) you could do this:
Original line:
<script language="JavaScript1.2">SiteSpeedIcon();</script>
With explicit name:
<script language="JavaScript1.2">SiteSpeedIcon('homepage');</script>

What does the 'error' icon mean?
If the icon starts flashing 'error' in red then the SiteSpeed program was unable to append the measurement to the logfile. The most likely cause is a permissions problem. See the *  download and install page for more information.

Does SiteSpeed make my page slower?
Yes, a little bit. The script that stores the measurements is very small and lightweight, but it's a script, and it needs a little CPU time for processing. The overhead of the SiteSpeed javascript module is about 2.7Kb, the first icon is 285 bytes, and the second icon is 1000 bytes. All this makes your page a little larger and therefore a little slower. It's the price you pay for getting interesting statistics.

How compatible is SiteSpeed?
SiteSpeed is compatible with all browsers that support JavaScript v1.2 and newer. Older browsers (for example Netscape v3.0) will not see the SiteSpeed icon and will not be measured.

Is SiteSpeed a benchmark?
No, it's a measurement. It will measure the time it takes to download a webpage. Influences are not exluded, on purpose, because I wanted to know the actual time visitors have to wait for pages. The statistics will only be meaningful and show patterns after hundreds of measurements. It's like measuring temperatures, a single measurement is interesting but trends are only visible in a graph.

How accurate is SiteSpeed?
It depends on the operating system of the user. Linux can measure times with an accuracy of 1 millisecond, Windows NT/XP/2003 is 10 to 16 milliseconds, Windows 98/ME is 50 to 60 milliseconds.

Why do measurements vary so much?
Some visitors have fast equipment and are close to your website, others may be using an old analog modem over a noisy phoneline at the other end of the world. SiteSpeed measures the speed of your webserver exactly as your visitors experience it.

SiteSpeed says slow but ping/traceroute says fast?
Ping and Traceroute measure network latency, the minimum time a data packet needs to travel through a network. They skip lots of internet aspects covered by SiteSpeed, such as download time (throughput), DNS delay, server processing, and browser efficieny. Please see the article *  The Ping Fallacy for more information.

How fast should my pages be?
Ah! The 10-million dollar question! A generally accepted number is between 10 and 15 seconds, including all graphics. Due to the average speed of the internet this translates to a total size of about 50Kb to 70Kb. Also see the *  Pagesize Calculator

How do I make my pages faster?
Remove external counters and banners. Optimize your graphics. Syntax-check and optimize your html (remove comments, whitespace, and other bloat). Translate frames into frameless. Translate style sheets into flat html. Optimize javascripts. Optimize ASP/PHP/Perl scripts and cgi-applications. Translate database-driven pages into static pages. Spread your pages amongst multiple webservers. Move from a virtual to a dedicated or co-location server. Move to a better provider.

How are uniques recognised?
SiteSpeed tries to differentiate between uniques (first-time visitors) and reloads by looking for the ip-address of the user in the last 1000 measurements. If the ip-address is found in a measurement less than 4 hours ago then the visitor is counted as a reload, and otherwise as a unique. This method is usually quite adequate, but will break in certain cases.

Can my SiteSpeed counter be hacked?
Yes. With a little hacking it's quite easy to send bogus measurements to your server. I don't see why anyone would want to falsify your statistics, though.

Can I exclude myself from the statistics?
Only if you make some changes to the programs (you will need some PHP programming knowledge). Hint: use the REMOTE_HOST environment variable in the "save.php" script to ignore measurements coming from a certain ip-address.

Can I password-protect the graphs?
Only if you make some changes to the programs (you will need some PHP programming knowledge). Hint: add a bit of code to the "graph.php" script that checks for a custom parameter. If the parameter is incorrect just "exit()".

Can I get the source of the graphics Java applet?
No.