On 04/08/2011 04:42 PM, Jaco Breitenbach wrote: > Dear experts, > > My application makes use of a SQLite table with a unique index. During > normal processing, the unique index is used to ensure no duplicate records > are entered into the table. However, at regular intervals large numbers of > records (millions of records) that are already known to be unique, are > inserted into the SQLite table from another source. In order to speed up > these bulk inserts, I first drop the index on the SQLite table, do the bulk > insert, and then recreate the index. > > The problem I'm encountering is that dropping of the index can potentially > take much longer than recreating it. In a test I've run this morning, it > took 53 minutes to complete the "DROP INDEX my_unique_index", but only 9 > minutes to recreate it.
Is your database an auto-vacuum database? If so, it might be faster if you issue a "PRAGMA auto_vacuum = 2" before dropping the index. Dan. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
