PerformanceEngineer.com

STPCon 2007: Day Two

8:45 am - 9:45 am: Keynote: Performance Testing: It isn’t what you might think
Scott Barber

Ed Correia gave a somewhat awkward introduction (ask someone else who was there) of conference keynote speaker Scott Barber. Barber spoke for approximately a half hour about myths and realities of performance testing. He went through several “myths” or common misconceptions that he has encountered in his years as a performance testing consultant. The clips from “Princess Bride” were very entertaining and surprisingly appropriate to the subject matter, but a lot of what Barber said seemed as though he was “preaching to the choir.”

10:15 am - 11:15 am: Performance Testing for Managers
Scott Barber

I have a hard time criticizing someone who has to teach a full-day class one day, followed by 4 more 1-hour sessions the following 2 days, and furthermore, it’s all about the same general subject matter, so please don’t take this session review the wrong way. I probably should have suspected from the session title, but this talk had very little to offer after having sat through the full-day course yesterday. At least I got to catch up on my email at work.

11:30 am - 12:30 pm: Ensuring a Superior User Experience With Web 2.0 Applications
Patrick Lightbody

Lightbody works for Gomez, and other than the Gomez logo at the top of each of his slides, there was refreshingly little mention of his company’s products and services. Instead, the discussion centered around how and why web development and tools have changed since 5 years ago. Nothing too earth shattering or ground breaking here. I did learn about some new tools for testing web applications, you may want to check these out:

• Selenium
• Drosera
• Firebog (well, I had heard of this one)
• YSlow

Also reinforced during this session was the importance of incorporating testing into a continuous integration process. This has become such a recurring theme, and yet there are few tools, documentation, how-to’s, wikis or anything else that I can find that can give me guidance about how to set it up.

In general, it was a decent talk and the hour actually seemed to pass by quite quickly.

2:15 pm - 3:15 pm: Runtime Diagnostics Testing
William Louth

I actually intended to attend Steven Haines presentation about Continuous Performance Management Using Java, but after scanning through the slides beforehand (they were distributed on a CD when registering for the conference), it looked like half of the time was going to be a promotion of Quest’s JProbe and java diagnostic tools, which was not what I was looking for. So instead, I headed for the Runtime Diagnostics Testing presentation by William Louth.

Besides some explanation of his quite humorous slides (check out his blog for some examples), what I was expecting from this class was some geeky demonstrations about how to capture interesting, perhaps obscure, data from a running JVM. Louth is clearly a very, very smart individual, and watching him do his code jujitsu on his PowerBook was quite impressive.

There was a little too much time on the slides and not enough on the demos to get a good understanding of either a) what he was talking about in the slides or b) what the tool’s capabilities are.

I won’t rule out that perhaps Louth was just way over my head (I’d like to hope not), but this session didn’t quite live up to my expectations. The slides mention a few technologies I was looking forward to seeing demos of, but they only got a brief mention: AOP. DTrace, XRay. The only tool demonstration was using JXInsight, which, in my opinion, is a great tool, not to mention that the developer edition is freely available. Louth, being the CTO of JInspired, didn’t push the tool at all, which was somewhat surprising, but I will suggest that if you have an IQ over 150, the tool can do just about anything you want it to - it is very powerful. However, figuring out how to use it can be challenge.

Summary: Louth is a smart, smart guy and definitely knows his stuff, but he didn’t really succeed in presenting and transferring his knowledge to me. I’d go hear him speak again, though.

4:00 pm - 5:00 pm: Concepts in Performance-Testing SOA
Scott Barber

Barber started off saying this is his least favorite technical session, and after sitting through it, I can see why. I can summarize the entire hour in the following sentence: SOA is not a well-defined industry standard, but even so, it doesn’t mean much to a performance tester because the things we have to test don’t really change.