What is this?
This third installment of the PMF database tuning series will not actually discuss database tuning per se; instead, this article will be a discussion about the discussion - a meta-discussion if you will. As I indicated in the previous articles, database tuning is a very well researched field; however, in spite of all this research, there are no definitive answers to many issues. The primary reason for this is the complexity of the problem domain - there are simply too many variables in the 'equation'. Factors which can affect your response times include CPU speed, memory space in RAM, latency of disk drives, # of users, size and complexity of the underlying data structures, the number and types of indexes, etc., etc, etc. As you can see from just this short list, there is no way that the PMF development team can possibly test all combinations; so when we do our performance testing we choose parameters based on our best understanding of how you, our customers, actually use PMF.

How we test - some technical details
Currently most PMF performance testing is done on a dual core Dell Latitude running at 2GHz, with 3 Gig available RAM. We test MS SQL Server, Oracle, and DB2/UDB; and we attempt to test all standard release levels. This allows us to isolate database issues from network issues.

Who is doing the testing?
All tests are run by the PMF development and QA teams. While we have expertise in database design, and while we can draw on a wealth of database expertise here at IBI, we do not know the intricacies of all databases; for this you need a trained and experienced DBA who is familiar with your local environment.

In summary
When you read any statements made about the strengths or weaknesses of particular databases, you should keep in mind that there may be factors that we are not taking into account. So if you are in the process of choosing a database, or if you are considering switching databases, please do not use these discussions to influence your decision, since you may experience very different results under seemingly identical circumstances.

Your Mileage May Vary
For those of you who are not familiar with this expression, "your mileage may vary" is an American idiomatic phrase: http://en.wiktionary.org/wiki/your_mileage_may_vary which applies to this situation.


EricH