I have a little experience with a somewhat similar setup: a typical "real" file is 200MB...2GB. I am making a "snapshot" of the data structure (few hundred K) into an attached in-memory-database.
I've seen that the timre required to create that snapshot depends largely on the size of the entire table, even if only selected columns go into the snapshot. I.e. ATTACH ':memory:' AS mem; INSERT INTO mem.Snapshot SELECT col1, col2 FROM Data; is much slower if 'Data' contains an additional column with large data. Moving my item meta data (small) into a table separate from the possibly large blobs helped immensely. Note: I haven't investigated that much, as separating the large data column into a separate table makes sense for other reasons. It could be that the significant difference - even though very consistent to observe with multiple files - was more due to OS/disk caching than SQLite itself. -- View this message in context: http://sqlite.1065341.n5.nabble.com/index-has-higher-cache-priority-than-data-tp64393p64397.html Sent from the SQLite mailing list archive at Nabble.com. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users