I have read that multiple indexes on one table cannot be used in combination
on most RDBMS products. If there are multiple indexes, then the optimizer
uses the best one, does a disk I/O to retrieve those records, and then goes
through a full scan of those returned to match the remaining WHERE criteria.
I've read elsewhere that the matched index pointers are compared, and THEN
the records are retrieved. Which is correct? I'm looking at a 90 million
record DB which will only be updated once every 3 months, and query
performance is my key issue. Are there any good resources out there to find
out exactly what methodology is used by Oracle, or other RDBMS's? Also, I've
seen a product by DISC (Dynamic Information Systems Corp.) called Omnidex
which uses "Inverted List Indexes" and "Dynamic Aggregation Indexes". Are
these really better than typical B tree indexes? Surely there must be a
trade off if they do perform better.
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum
Received on Wed Aug 26 1998 - 12:06:10 CDT
