2011/9/6 jerome moliere <jerome.moli...@gmail.com>: > Did you set corresponding pragmas >> page_size, cache_size, journal_mode? > page_size yes > cache_size no because we are using a higher level cache (in Java) > journal_mode is in the list of the pragmas to be tested through the > benchmark ....setting this parameter to OFF speeds up transactions for > around 20%
I think your cache size in smaller than the modified indicies size. Use more big cache or reduce indicies size. > Indicies can be very >> slow... may be the FTS4 table will be much faster for your queries. > > can you give more details about this concept ? what is the FTS4 table > ? I am sorry don't know what is is !!! See http://www.sqlite.org/fts3.html The full-text index is very fast and scalable. You can use it instead of a lot of btree-indicies when you can rewrite your search queries as FTS. >> Did you try to use in-memory database as temp storage and >> copy set of records into main database in single transaction? > > it's one of the goal of the benchmark but this solution has av ery > high level of database corruption isn't it ..so I prefer to keep it as > a joker... Database corruption can't be the result of the coping from temp in-memory database to main disk-database! -- Best regards, Alexey Pechnikov. http://pechnikov.tel/ _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
