On 29 Jul 2018, at 11:39pm, Gerlando Falauto <gerlando.fala...@gmail.com> wrote:
> In the current use case thre's a single process. The way I see it, in > the near future it would probably increase to 3-4 processes, > each doing 10-100 writes per second or so. Each write would be around > 1KB-20KB (one single text field, I guess). > I wonder if writing data in batches would be helpful though. The critical path in SQLite write operations is your storage device. Your entire database is on one drive, your CPU has one path to it, and the processing code in SQLite is very efficient. Consequently most of the waiting time during API execution is spent waiting for hardware to say "Okay, I've written that.". The number of indexes can make a big difference since it increases the number of database pages which must be written. So you can make a good guess whether sharding, multi-processing, or other parallel shenanigans will help. Just imagine it's 10% processing and 90% hardware access. Simon. _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
