G'day, I'm currently evaluating sapdb as a solution to a problem I have. The application I'm trying to build it basically a historical data store that records value changes for a large number of double precision-typed inputs over time. Some statistical analysis will be required on the data but queries are actually going to be few and far between. The main source of distress to the database will be the number of inserts.
The database is very simple. Essentially it requires only one table which stores the entries as they come in. There is only one client inserting records into the database. There may be multiple querying clients at any one time but whereas in the final system there will be around 300 inserts per second-long transaction with spikes perhaps an order of magnitute greater than that there will only be a small query on the database a few times per minute and a large query on the database once in a blue moon. The hardware I'm testing on is not the best to show off sabdb's abilities. I've been testing on a single-processor machine with a single hard drive being used for both log and data spaces. The final machine will probably still have both log and data spaces on the same virtual drive, but it will be striped. There will be two cpus on the final machine, but the database server has to coreside with a runtime (ram-only) database system that will be chewing most of one of those cpus. What I'm looking for is a set of database parameters and tweaks I can use to maximise the insertion rate in my test setup so that I can determine whether or not the performance will match my requirements on my hardware. Any ideas? Benjamin. _______________________________________________ sapdb.general mailing list [EMAIL PROTECTED] http://listserv.sap.com/mailman/listinfo/sapdb.general
