Parallel Data Query - When Most Beneficial ?

From: Daniel Druker <ddruker_at_netcom.com>
Date: Wed, 13 Jul 1994 02:47:09 GMT
Message-ID: <ddrukerCsuyEM.I8J_at_netcom.com>


We had a debate at work today that hopefully some net.genius can solve.

The question was whether parallel data query would speed up retrieval from a particular data model. The data model in question is a star model, such as referred to in a recent article on using RDBMS for decision support in Datamation. It looks like this.

  1 facts table with millions of rows, with a compound primary index, which   uniquely defines every row in the table. For this example, the primary   index has 5 elements. The actual data being selected is a single small   column.

  5 entity tables each with 10 to 1000 or so rows, each of which contains a   single column which contains one piece of the foreign key to the facts   table. The 5 entity tables joined together form the entire foreign key   to the facts table.

An easy sample query would join all 6 tables, uniquely specifying a single row in each of all 5 entity tables to generate a single foreign key to the facts table, which would then return one and only one row.

A harder query might specify a range of rows in the entity tables, returning multiple rows from the facts table. In all cases all lookups would be indexed.

The question is, would parallel data query speed up these kinds of queries ? More generally, what is the best/most effective use of parallel query ? Large table scans ? Aggregations ? Many way joins ?

Any and all input is appreciated.

Regards,

  • Dan

Contact Information:


Home:						Office:
Daniel Druker					Daniel Druker
14190 Wild Plum Lane 				Arbor Software Corporation
Los Altos Hills, CA, 94022			1325 Chesapeake Terrace
 						Sunnyvale, CA, 94089
ddruker_at_netcom.com				(408) 727-5800
						(408) 727-7140 Fax
						ddruker_at_arborsoft.com
-----------------------------------------------------------------------------
August 1, I will be moving to Boston for 6 months. You'll still be able to reach me at the above addresses and phones.
-- 


- Dan
Contact Information: ------------------------------------------------------------------------------ Home: Office: Daniel Druker Daniel Druker 14190 Wild Plum Lane Arbor Software Corporation Los Altos Hills, CA, 94022 1325 Chesapeake Terrace Sunnyvale, CA, 94089 ddruker_at_netcom.com (408) 727-5800 (408) 727-7140 Fax ddruker_at_arborsoft.com ----------------------------------------------------------------------------- August 1, I will be moving to Boston for 6 months. You'll still be able to reach me at the above addresses and phones.
Received on Wed Jul 13 1994 - 04:47:09 CEST

Original text of this message