Seeking performance data

From: Rich Remington <richr_at_ihs.com>
Date: 1996/10/29
Message-ID: <01bbc5ec$510ed1a0$dd0f4dc0_at_remingpc.ihs.com>#1/1


If any of you are aware of performance benchmark data for the following problem please e-mail directly to me and I will post a summary of the findings (if it seems appropriate.)

We produce several technical database products and need to begin combining objects from several current databases into new products. In order to manage this we will be implementing a "global" ID server where all objects (documents, images, sound/video clips, etc.) will get a unique and unchanging identifier. We would like to know if anyone has done studies (or knows of a study) that can tell us how much penalty there is for using a character key (say 12 or 16 bytes) versus a binary key. We will need to access large amounts of data (potentially millions of items - video, documents, etc.) based on this key in a very short period of time (e.g. a few hours.) We are aware that database products can use a 4 byte binary key very efficiently, but we feel that the universe of keys this could generate is too small for the near future.

We have both Oracle and Ingres in house - hence the reason for posting to these newsgroups. I don't necessarily trust performance data from the vendors... If anyone knows of another newsgroup that I should post to, then please let me know.

Thanks! All information is appreciated. Received on Tue Oct 29 1996 - 00:00:00 CET

Original text of this message