* WebMark 2004
The traffic on corporate networks has changed remarkably over the last decade. Not only have e-mail and Web browsing traffic levels gone through the roof but all of our hot Web applications are driving new levels and patterns of communication.
Considering your heap of infrastructure and your collection of Web applications – how do you test the whole system? You not only have Internet access to consider, you also have intranet, extranet, inter-office WAN connections, remote workers, and wireless access to consider as well.
To determine whether your Web applications will deliver optimal performance in this complex evolving environment you need to have a benchmark. You need to be able to profile the multiple dimensions of Web-related network activity so that you have a baseline from which you can assess whether your Web applications will work as you intend.
Moreover, a benchmark can also provide an independent check of overall system conditions that shows you whether your application or the infrastructure is at fault.
If this is the kind of issue you’re dealing with (or have suddenly realized you should be dealing with) then you might want to take a look at WebMark 2004 from the Business Applications Performance Corporation (BAPCo).
BAPCo is a non-profit consortium chartered with the development and distribution of “objective performance benchmarks for personal computers based on popular software applications and operating systems.” Current BAPCo members include, among others, AMD, Apple, Atheros Communications, ATi Technologies, Dell, HP, Intel, Lenovo, Microsoft, Nvidia, Seagate, Toshiba and Transmeta.
WebMark has both client and server components. In offline mode only the client component is needed and uses locally cached content so that you can benchmark the client-side Web performance without network connections being involved. In online mode, both the server and client components are used.
WebMark 2004 creates two scenarios: Information Processing and Commercial Transactions. The Information Processing scenario includes content typical of a portal site, a research site, and a training site while the commercial transactions scenario simulates finance, marketplace and purchasing content.
The results of the various tests are evaluated as follows: “Each of the Web sites above have a rating that is calculated by taking the sum of duration of operations that constitute that Web site and then comparing it with the equivalent sum of the duration of operations on the Calibration System” (BAPco’s reference system specification).
The final rating value “is calculated by dividing the sum of the duration of operations in the Web site on the calibration system by the sum of the duration of operations on the test system and multiplying by 100. The result is then rounded to the nearest integer. At the end of a benchmark run, the Web site rating for each of the 6 Web sites is calculated.”
WebMark also provides Scenario Ratings for each of the two scenarios. This is calculated by “taking the Geometric mean of the Web site ratings of all the three Web sites within each scenario.” Finally, an overall rating is calculated by “taking the geometric mean of the two Scenario ratings.” See “WebMark’s Scoring Methodology” for more detail.
A much more thorough discussion of WebMark’s features and architecture can be found in the WebMark white paper (PDF).
If you’re not benchmarking your infrastructure alongside your Web applications you are missing out on a critical aspect of building robust, effective Web services and products. WebMark is an excellent way of removing the uncertainty of where performance problems might lie.
WebMark is priced at $300 for a single-user license and $600 for a site license.
