Keynote KITE 2.0: Check Web Site Performance

Keynote’s Keynote Internet Testing Environment, or KITE, is a desktop application for testing the performance of Web sites. While KITE is aimed primarily at Web application developers, it caught my interest as a tool for measuring the performance of an organisation’s current or prospective software-as-a-service-based applications.

I tested KITE Version 2.0, which began shipping in August. I used KITE to create scripts of typical operations with a few Web-based SAAS applications, and then ran those scripts both from my local machine and from a collection of Keynote’s global monitoring sites. I was impressed with the granular performance information the product yielded after a short period of use.

To examine and measure rich application functionality – such as through AJAX – I had to make some script modifications as described in Keynote’s online product documentation.

KITE is freely downloadable, and KITE scripts can be run either from the desktop and network on which the software is installed, or through Keynote points of presence in San Francisco, New York, Frankfurt, Germany, London and Hong Kong. For San Francisco, there’s a second point of presence that measures Web application performance over a DSL link to provide a reading on the impact of a user’s last-mile connection on the test application.

For an additional fee, users can expand their tests to additional locations, as well as configure tests to run remotely at scheduled intervals.

KITE runs on Windows XP or Vista, and uses Internet Explorer 7 for script recording and for local test sessions. According to Keynote, the company has no plans to support other operating systems. However, I was able to perform remote tests using Web browsers other than Internet Explorer by modifying my script settings using KITE’s Script Properties Editor.

KITE in Action

At its most basic level, KITE enables users to test a particular URL using Keynote’s Web site—the service accesses a given URL from the six points of presence I mentioned above—and provides a breakdown of all the elements on the specified page and how much network time and browser rendering time it took to load those elements.

For scripted site interaction, the KITE desktop application comes into play. The KITE client enables users to create two types of script, those that involve scripting of a real Web browser, called TxP scripts, and those that involve scripting of an emulated browser, called ApP scripts. The ApP scripts incur fewer costs when run on Keynote’s remote points of presence (beyond the six free points), but the TxP scripts are required to test rich, asynchronous Web applications.

Conclusion

The free software from Keynote offers strong granular performance data, but developers will need to conduct some trial-and-error work with KITE’s scripting tools to get it just right for them.