On Thu, 2007-07-12 at 11:19 -0700, Benjamin Arai wrote: > I am running the latest 8.2.4. I am using GIN. The data is static. I do > a batch upload every week of about 500GB and the ata is never touched > again, it is always add and never delete or update. Partitioning will certainly help the index build times, even if it doesn't specifically help with the queries. > >From your slides you state: > > GIN_FUZZY_SEARCH_LIMIT - maximum number of > returned rows > – GIN_FUZZY_SEARCH_LIMIT=0, disabled on > default > > When I do a search with say LIMIT 100 isn't this essentially the same thing? Both restrict the number of matches found, but they're not the same thing. One is for the query as a whole, the other is for one index scan on a GIN index. -- Simon Riggs EnterpriseDB http://www.enterprisedb.com