I didn't tell the whole story. I also tested Paradox, DBISAM (a product for Delphi) and Interbase. Interbase and MS Access were very close to each other, and Paradox was way bigger than MS Access. DBISAM was a little smaller than Paradox, but also way bigger than MS Access.
Of course, there are reasons for the big increase in storage usage. I used the test results as an argument to replace a database server with an intelligent self-made server application in a project, where speed was very important. Instead of using a database server, we simply wrote a Delphi program, that our clients could connect to, and the Delphi program processed requests serially, not parallel, preventing transactions from overlapping. Paradox files were used for storage, but all data was present in memory in optimized data structures. The performance was incredible. The solution wouldn't scale beyond one server, but one server could take a very high load, and for this application, scalability would therefore not be a problem.
The original topic, though, was that it gets more and more important to reduce memory usage because memory becomes the bottleneck of systems. And the question is, will Microsoft become siginifcantly slower than open-source systems due to bigger code and data size and RAM bandwidth problems? |
