DB Performance Optimization

Performance problems arise from many sources and manifest themselves in many ways. Unfortunately, this duality can make for difficult resolutions. Most of the time we will find the problem lurking in application code. Therefore our early work is to analyze applications, especially the parts that add, update, and retrieve data from a database. We have developed detailed structured techniques to perform this analysis, essentially creating a performance data warehouse (in effect metadata about application execution) that provides keen insight into what program units are running, what data is being used, and how much tim eis consumed. With this information we quickly identify what to fix; in fact, finding the problem is usually the time consuming part, fixing it is usually relatively easy.

Certainly, not every performance problem is the fault of suboptimal database code. In one of our favorite incidents, the fault was a dead battery on a SAN cache. Between sql optimization and new batteries, there are a broad range of possibilities, and we have encountered scenarios where many of them are appropriate. For example, database memory allocation and partition vs. indexing strategies can be arcane pursuits, but we have experts who have mastered them. The point is, if performance is a concern, its likely that we can help - and that's true whether or not the problem lies at the database level.