It is still not late to submit a paper to the CMG conference. Although the abstract deadline passed, the enforced deadline is only the paper deadline – June 13.

I believe that it is the best conference about computer performance. Conference proceedings from 1997 through 2005 are available to the public – probably the largest collection of performance-related papers (and most of them are practical).

Where the myth that you can use LoadRunner with a regular user came from? I got a client who refused to give me admin rights and I should research this question again. I did it before – but, in my practice, I usually return to the same question after such period of time that the only thing I remember about it is that I researched it before.

Well, it took some time. Doesn't look like the new Mercury knowledgebase is a major search improvement.

The Computer Measurement Group (CMG) is making its conference proceedings from 1997 through 2005 available to the public. I believe that CMG holds the best practical conference in performance analysis, capacity planning, and related areas. In addition to the areas listed below, I'd definitely add performance testing. Here is the official mail CMG sent:

A few thoughts inspired by a question about qualitative and quantitative information reporting performance test results in Yahoo's LoadRunner group.

My understanding of qualitative and quantitative in the context of reporting performance test results is:

Quantitative is what we directly measure or straightforwardly calculate from direct measurements (like response times or resource utilization)

From time to time I see questions on different forums asking what skills are necessary for performance testers. There were pretty interesting discussions. Looks like most experts agree that performance testing requires more skills than just knowledge about how to create a script for a particular load testing tool. While it is still possible to imagine a performance tester in a large corporation with deep specialization who only creates scripts and mechanically runs them while other performance experts monitor the system and analyze results, I don't see many perspectives neither for this person, nor for the approach. Systems become so complicated now that the sum of specialized expert views doesn't give the whole performance picture.

Read more and register: http://www.performance-workshop.org/content/view/44/86/

Theme: Pushing the boundaries of performance testing tools
or when the performance test tool won't let you do…
Conference Location and Time

Microsoft, Redmond, WA USA
September 27 - 29, 2007

A Microsoft organized pre-WOPR event will take place on September 26. WOPR participants will also be invited to attend this event.

Remote user experience may differ significantly from the experience of users using local network. It is not a new thing for web developers, but often missed by developers of corporate applications. Many clients have remote offices with limited bandwidth and high latency that may cause performance problems. If it is a concern, there are ways to verify remote user experience. If we speak about LoadRunner it is possible to mention the following ways to simulate remote/WAN users:

As some of you may know, I've been involved with the OpenSTA project (although not much for a long time). I still have an emotional bond to the project and still read the developers mailing list... Following Radview's recent announcement, the developers mailing list has had some discussion on the topic. Despite Radview's claims of being the first, OpenSTA began life in much the same way. The only difference is that OpenSTA was released almost immediately as open source. OpenSTA was destined to be a commercial load testing tool. Cyrano (the company that was creating it) decided to release it as opensource with a plugin API so that they could sell commercial plugins that extended it's abilities. People would also be free to write open source plugins of course. OpenSTA was first released... errrm... well so long ago, I can't remember, around the year 2000?. Dan Sutcliffe posted this enlightening analysis of Radview's recent move. In his analysis, he highlights a key limitation in the open source version:

Here's where you can find out what those "additional productivity features" are: "http://www.radview.com/product/Editions-Comparison.aspx":http://www.radview.com/product/Editions-Comparison.aspx The biggy is that you need to go to the commercial version if you want to do distributed load generation.

No distributed load generation? OK, well they need to make their money somehow... From the preliminary checks that Bernie Velivis has done, it doesn't seem that it will take much to get you close to needing a second load generator. I recommend that you check out Dan's full post. It's an enlightening read: