Legolas Woodland wrote:

> Please can some on share his/her knowledge about this item.
> It will help me so much to complete my job.

The best advice is usually to test your own application, or code that
mimics your application. Ensure you perform the database operations
fully and don't take short-cuts. I've seen very bad benchmarks over the
years, for example executing rs = ps.executeQuery() but never processing
the ResultSet, ie. not fetching the rows or columns. Also it's good to
ensure the test performs what you think it meant to do, usually by
running test sql after the performance runs, e.g. if I expected 100,000
orders to be added, check they actually there in the database.

This presentation shows Derby faster than MySQL and MySQL faster than
Derby. :-)

Derby was faster when the database was larger than the memory
MySQL was faster when the database was smaller than the memory

http://www3.java.no/JavaZone/2005/presentasjoner/BerntJohnsen/Bernt_Johnsen-DerbyJavaZone.pdf

I know since those numbers there have been some fixes and interesting
discoveries by the Sun team in Norway that improves Derby's performance.

Dan.

Reply via email to