On Wed, Jan 09, 2008 at 11:51:16AM -0500, Josh Harrison wrote: > accessed frequently. So clustering the table according to one index will > yield poor performance to queries involving other indexes.
Maybe not poor, but certainly not optimised. > Index-only scan is a good solution for this I guess for queries involving > indexed columns (like in oracle) !!! I think I don't know what you mean by "index-only scan". Oracle can't be returning you data just by scanning an index, can it? It needs to get the tuple if you need it back. My bet is that you haven't tuned your vacuums correctly, or you aren't doing ANALYSE often enough on the affected tables, or you need to SET STATISTICS higher on some of the columns in order to get better estimates (and therefore better plans). A ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match