|
> I don't think anyone knows enough to make general > statements. That's why I ran tests. As usual, I didn't make my point very clearly. Here's the scenario. A pair of absolutely identical iSeries systems. CPU model, feature code, PTF levels, everything EXCEPT for the number of records in otherwise identical databases. So the ONLY difference between the two systems is the number of records in the databases. Run some timings on an unloaded system A using READ. Each READ will take a certain amount of time per record. Run the same tests on an unloaded system B and your timings will turn out practically identical. You can extrapolate that a READ for n records will take (N * READ time) and you'd be really close. Perform the same tests with SQL and the results on the emptier database will not extrapolate to the fuller one. You cannot predict what the system response time will be based on your tests. You cannot use your tests to prove anything at all except what performance is for ONE and ONLY ONE system/dataset combination. MY SQL testing means literally nothing to anyone else. This sounds extreme but fits my experience base very well. As for the talk about system performance, it is up to the database and system admin to configure a properly performing _system_ (hardware/software/database.) A programmer is unlikely to have the political motivation to spend the time required to optimise the whole shebang. I believe this to be true (admin should optimise) of both native I/O (since the introduction of OPNQRYF) and SQL I/O. The major difference between the two in my opinion is that native I/O tests can be used as a firm basis for extrapolating the performance of the application in production. SQL I/O tests cannot. Classical RPG programmers starting to use SQL often treat SQL as if it were RPG spoken with a funny accent. This is a mistake, in my opinion, and the cause of many performance problems. All my opinion, and none personal. Just a note about my experience for the archives. I'll shut up now. --buck
As an Amazon Associate we earn from qualifying purchases.
This mailing list archive is Copyright 1997-2024 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].
Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.