
On Mon, Nov 26, 2007 at 08:41:31AM +0100, Markus Gritsch wrote:
On Nov 26, 2007 7:53 AM, Martin Kersten
wrote: MonetDB internally uses (hash-)indices, which are created for the duration of the session. And the holy grale for database designers is to create a self-organizing system, one that learns how to index without user interaction.
Yes, this would be the case in an ideal world :) Nevertheless the system could take some hints from the schema designer into account (via indexes), in case he exactly knows how the schema is queried and what columns can benefit greatly from having an indexed on them.
Agreed. However, if the cost of maintaining the indices under updates exceeds the (potential) benefits of exploiting the indices for queries, the investment is debatable. Or, in other words, even without "old-fashioned" indices, MonetDB/SQL appears to outperform MySQL by more than a factor 3 (in your case). Hence, we still have some margin/time to finish our hard research work on (a.o.) self-organizing query-driven adaptive indices --- that work even without "hints from the schema designer", e.g., in case it is not a priory known how a schema is queried, or which columns could bebeit from an index --- to hopefully get the very promising initial results of a prototypical implementation stable and robust enough for the product version ;-) Stefan
Markus
-- | Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl | | CWI, P.O.Box 94079 | http://www.cwi.nl/~manegold/ | | 1090 GB Amsterdam | Tel.: +31 (20) 592-4212 | | The Netherlands | Fax : +31 (20) 592-4312 |