On Thu, Jun 25, 2009 at 12:36 PM, David Fletcher<fr...@tuscanyda.com> wrote: > >>>>>> "PK" == P Kishor <P> writes: > >>> As expected, as the table grows, the underlying B-tree >>> implementation for SQLite means that the number of disks accesses >>> to (a) find, and (b) add a chunk, grows larger and larger. We¹ve >>> tested up to 20 million chunks represented in the table: as >>> expected performance exponentially decreases as the number of table >>> entries grows. > > PK> Why don't you use, or at least test, with BerkeleyDB? Since you have > PK> only one table, you can hardly benefit from the SQLness of an rdb. If > PK> nothing, you will have a point of comparison with another technology, > PK> and know for sure if SQLite is the appropriate solution for you. > > Perhaps http://tokyocabinet.sf.net might be a better choice. >
heck! Do two comparisons -- SQLite v. BerkeleyDB v. Tokyo Cabinet. Nothing like thorough testing for the purpose of science. :-) -- Puneet Kishor "assertions are politics... backing up assertions with evidence is science" _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
