single high-dimension index versus several low-dimension.
Date: 30 Jun 2003 09:15:19 -0700
Message-ID: <a4e6b37.0306300815.6aef509b_at_posting.google.com>
Hello,
I have a rather nebulous question on a single high-dimension index versus several low-dimension.
Say I have a large amount of data (but small enough to be kept in memory) and I expect many queries that specify say 4 or 5 required values. Does anyone have any insight into using a high-dimension index, such as an X-tree (say I am allowing selection based on 10 of 20 attributes) versus several low-dimension indexes (b+tree or even plain old binary tree) and joining the results from each index lookup?
I'm not using a database, but in many ways the application quite resembles one - though there is no need for transaction support etc.
All suggestions and/or paper pointers welcome.
Best regards,
Niall
Received on Mon Jun 30 2003 - 18:15:19 CEST