Users started complaining about spotty application performance; sometimes it was good, sometimes it was horrible. The first thing came to my mind was some kind of database bottleneck. In my experience, CPU and I/O contention are usually the culprits but one time (not this time) it actually was a network problem. Here's how I broke it down...

A team member thought we should add an index on a 90 million row table to improve performance. The field on which he wanted to create this index had only four possible values. To which I replied that an index on a low cardinality field wasn't really going to help anything. My boss then asked me why wouldn't it help? I sputtered around for a response but ended up telling him that I'd get back to him with a reasonable explanation.

I recently created an application that started with the appfuse stack. Everything performed admirably, until I created a page with numerous form elements (~30K)that was large (~38M when saved to disk). The page would either cause an OutOfMemory (OOM) error or would take 3+ minutes for the first byte download. Here's what I did to bring full page load times to < 15s.

Heinz writes The Java Specialists' Newsletter. In his latest issue he offers up a little program that will test CPU efficiency...

I was given a relatively easy task, to sort a file according to two different keys in ascending order. The issue was that the file was quite large and the parsers provided created a great deal of objects, so memory was going to be an issue...

At my current gig, we have been evaluating DASDs from both Hitachi and IBM. Both providers want to be my client's supplier. Yep I work for a client with some clout. In the race to be the DASD provider here, each side is one-upping the other. Here's the latest scores..

After reading this article about Young vs. Old Generation GC, I decided to run the same tests in the IBM 1.4.2 JDK.

I don't really have a good handle on how to use this information to predict what adding another CPU will do.. but it is interesting to think about. In a perfect world, a computational job that is split up among N processors would complete in 1/N time, leading to an N-fold increase in power.

This article provides several tips and techniques that are commonly used for tuning Java applications for optimum performance on AIX. A discussion of the applicability of each tip is also provided. Using these tips, you should be able to quickly optimize the Java environment to suit your application's needs.