On Tue, Aug 9, 2011 at 4:46 PM, Drew Kozicki <drewkozi...@gmail.com> wrote:
> We have an application that does not scale well – especially 8 and 16 > threads we are getting about 10% at best to -50% at worst (2 and 4 are okay > aprox. 40-50%). > Open a separate database connection for each thread. Don't try to use the same database connection on all threads because access to the database connection is serialized. > > > > We have narrowed the problem down to the SQLite code. > > > > Looking for best practices from anyone that overcame scaling issues. > > > First we generate 35 different SQLite databases, this is done in a separate > program and performance is not an issue. > > > > The application is a C++ app. that accesses all the databases in a read > only style. Size of database range from 1MB to 10GB. > > > Each thread does the following in a loop > > 1. Grab a record from an external system (not SQLite) > > 2. Runs several SQLite queries to process the record > > > > > > We are currently using 3.6.14 (has this been improved in new versions?) > > > Is there any compile time options or pragma's that are specifically > engineered at optimizing for read only and/or scaling? > > > Thank you in advance, > > Drew > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
