Hi, I am building a libevent based application that must be able to handle tens of thousands requests per second.
Each request needs multiple database lookups. Almost all requests do the lookups on the primary key of the tables only. So far I have been using Hash Tables from the glib2 library. But now I need SQL access to the same data and I don't want to write my own SQL engine, so I'm thinking of adding sqlite to the mix (in memory mode only). I've added a virtual table driver and it does seem to work but it's not an ideal solution, because C structures are pretty inflexible (can't add columns without recompiling), and most of the in-memory data is loaded from an external SQL database, which routinely changes the database layout. So I am wondering if I can drop the glib Hash Tables, and go sqlite all the way. But I'm afraid the process of constructing SQL queries / parsing them by sqlite, and interpreting the results in my app, multiple times per event will be too slow. So now my question: - can I implement/add a hash index in sqlite? - can i access the table directly without the overhead of constructing an SQL query/decomposing the result? Thanks for any hints and tips, Ron _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users