Clustered Tables and Cluster Keys

I'm working on a physical database design at the moment and I'm considering using clusters.

Basically I want to have a cluster of

    KEY1 VARCHAR (13 chars)
    KEY2 DATE And two clustered tables: Table A...

    KEY3 NUMBER (0 .. 30)
    DATA_1 NUMBER
... ...

    DATA_n NUMBER

and Table B...

    KEY3 VARCHAR
    DATA_1 NUMBER
... ...

    DATA_n NUMBER

For each clustered key value Table A would have 30 records and Table B would have from 50 to several hundred records. 6 data fields in each table.

Access patterns will generally be bursts of inserts folowed by bursts of SELECTs with KEY1=constant and KEY2 being a date range. Table A and Table B will be read in turn, but not joined. Key fields will never be updated. Data volumes around 1 to 2 million records per day (inserts and deletes to a rolling data set).

I was wondering if anyone has any words of wisdom on the pros and cons of clusters (beyond what's in the Oracle manuals).

Regards,
Tim Bunce. Received on Mon Feb 06 1995 - 14:00:11 CET