Mike Bromilow

If you’ve been managing Web operations, then chances are that you’ve been using performance monitoring for several years, and you know the value of regularly monitoring the end-to-end performance of your Web content from the network perspective.

A couple of years ago, the network perspective would have been enough. Today, however, the pervasive use of Rich Internet Application technologies (RIAs), such as Ajax, Microsoft Silverlight and Flash, on Web sites introduces a new level of complexity in monitoring content performance. RIAs require the browser client to take an active part in the execution of Web application code, as opposed to earlier applications that were executed fully at the Web or application server.

Now, if you don’t understand what happens when a real browser interacts with embedded RIA applications, you won’t know, despite regular monitoring, what your end users are actually experiencing.

Real browser vs. emulated browser monitoring

When the most a browser was called upon to do was display content that was either statically or dynamically processed on a Web server, it was possible to plug a credible imitation browser into a Web monitoring tool. Now, however, how the browser interacts with RIAs is a critical component in the rendering of Web pages. For example, the browser might be called upon to render code utilising say, a JavaScript engine built into the browser, or a Flash player installed as a Web plug-in. This level of sophisticated browser interaction with Web pages is far beyond what was required of browser emulators under Web 1.0 technology. Furthermore, since browsers such as Internet Explorer and Firefox represent millions of dollars of investment and years of development work, it would be near impossible to create comprehensive browser emulators.

Yet, most Web monitoring tools are still attempting to do the job with older browser emulators. These still have their place when the goal is simply monitoring for availability over the network, but it takes a true browser to accurately monitor and measure the wide variety of interactions between the Web application and its final rendering of RIA-based content for the end user.

When an imitation browser is “not good enough”

There are at least three instances where the actual end-user experience can only be captured with Web performance monitoring technologies that use a real browser. First, capturing a snapshot of a Web page for debugging; second, measuring the impact of executing JavaScript; and finally, detecting Flash plug-ins or browser-specific content requirements.

Web page rendering and debugging

When troubleshooting a site that uses RIA technologies, you can save a huge amount of time and get the page back to acceptable performance levels much more quickly when you can review a true high-fidelity bitmap of the page at the time an error occurred. However, this is only available if your monitoring system takes a screenshot of the page exactly as it is drawn by the browser.

For example, when a failure occurs because the DHTML is not displaying correctly, a real browser monitoring solution will show an actual screen shot taken at the time of failure. However, an imitation browser that captures individual objects can only try to reconstruct the image by assembling those objects via HTML, a process that is prone to errors and can therefore result in misleading analysis.

Total user experience time

The end-user’s experience can be reflected by capturing total user experience time, which includes both network time and client time. Network time is the total time needed to download a Web page from the network’s point of view, that is, from the beginning of the base file’s DNS lookup to the delivery of the last packet of the last page element. Both imitation browsers and real browsers can capture network time, however, only a real browser can capture the all-important client time. Client time includes the time during which page element execution is taking place—such as the time the browser takes to execute JavaScript or execute and display Flash objects.

Using real browser testing and monitoring, it is possible to see the difference between reassuring network time and client time, as you can see all the objects on a page and the time they took to execute on the client computer. Because an imitation browser does not capture activities such as JavaScript client execution time, when a performance slowdown occurs, it is impossible to distinguish whether the problem lies in network time or client execution time.

Detecting Flash plug-ins or browser-specific content requirements

Because RIAs make such a huge difference in how the end user experiences a Web page, sophisticated page designs often try to detect whether the browser being used can accept a Flash object to display rich content. Some Web pages also attempt to detect which version of a browser the end user is running in order to optimise user experience, for example by ensuring different coding will be downloaded to provide the best user experience.

In a situation where a Web page checks for a Flash-enabled browser, only a real browser will reflect how the end user will experience the interaction between the code download and the browser on his or her computer.

An imitation browser is at a disadvantage as it may not allow the Flash plug-in to be loaded, so it may miss certain Flash objects, and even if it can load the Flash object, it may not wait for it to render, resulting in potentially misleading Web performance measurements.

More and more Web sites are incorporating RIAs to enhance the end-user experience and encourage both stickiness and sales. Therefore real browser monitoring is becoming increasingly important. Accurate measurement of the interaction of the browser and RIA elements is critical both to Web site owners and to content providers in order to ensure that service-level agreements (SLAs) are being met, and that the end-user is receiving the most positive experience possible.

About the author

Mike Bromilow is UK country manager for Keynote Systems. He has been in sales and marketing for almost 30 years, most recently as managing director EMEA for Gomez. Prior to this, Mike held a number of other leadership positions including managing director for the UK and Northern Europe of MKS, country manager for Starbase Corporation, and senior IT sales and management positions at Amdahl, Digital, Prime and AT&T.

About the company

Keynote Systems (Nasdaq “KEYN”) is the global leader in on-demand test & measurement solutions for continuously improving the online experience. For over a decade, Keynote has been providing measurement data and testing capabilities that allow companies to understand and improve their customers’ online and mobile experience. Keynote has four test and measurement businesses: Web performance, mobile quality, streaming & VoIP, and customer experience/UX.

Known as The Mobile and Internet Performance Authority™, Keynote has a market-leading infrastructure of over 3,600 measurement computers and mobile devices in over 240 locations around the world. Keynote’s 2,800 customers represent top Internet and mobile companies including American Express, Dell, Disney, eBay, E*TRADE, Expedia, Microsoft, SonyEricsson, Sprint, T-Mobile, Verizon, Vodafone and YouTube.

Keynote Systems, Inc. is headquartered in San Mateo, California and can be reached at www.keynote.com.

Relevant links

For more technical information on the interactions between real and emulated browsers in real end user situations, click here to read the white paper, “Why Read Browsers Matter” or click here to view case studies from financial, retail and media sites.