Igor Tandetnik wrote: > "Paolo Pisati" <p.pis...@oltrelinux.com> wrote > in message news:499a954e.5020...@oltrelinux.com > >> I've a perl script that reads rows from a file, does some processing >> and then insert the resulting data in a db, one INSERT per row. >> Usually the input file is made of hundreds of thousands, thus a >> typical run consists of hundreds of thousands of INSERTions. >> >> the performance was good until i decided to build indexes: >> > > Try creating indexes at the end, after all the inserts are done. > unfortunately, that's not an option for me: in the middle of all these inserts, someone could want to retrieve some data and without indexes the operation would be painfully slow - at minimum, this db has to hold around 14 days worth of logs, thus a select without indexes could take too long and block the remaining inserts.
-- bye, P. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
