Distributed joins, are they REALLY NEEDED?

Hi all,

Forgive me for the cross-posting. I am currently working on my doctoral thesis on Distributed Join Processing algorithms and their performance issues (e.g., how to efficiently join N relations residing in M distributed sites so that the network cost and join processing cost are minimized.) Trying to avoid doing something purely academic, I would like to get some insights from hands-on practioners regarding the necessity of distributed join handling. In particular the following questions are of great interest to me.

• Before you think about performing ad-hoc distributed join queries, won't replicating the referenced relations locally solve your problem better?
• What's the nature of your application requiring distributed join queries? Are the queries highly restrictive in their join result size?
• How large is the data volume involved?

Please reply by email. I would post a summary if enough interests are shown by the netters.

Regards,

• Jay Li

Department of Computer Science
Columbia University
New York, NY 10027 Received on Fri Feb 10 1995 - 05:33:42 CET