Previous Topic: Data Warehouse GuidelinesNext Topic: General Tips


Data Warehouse Performance

Most of the workload for data warehouse queries is caused by scanning of large fact tables . However, these tables are relatively compact for the number of rows that they contain (for example, 10GB per 100 million rows). Therefore, the database can readily cache these tables, which further reduces the physical I/O for database queries.

Note: Reducing the I/O bottleneck generally leads to intensive CPU usage while database queries are running. Therefore, increasing the number of CPU threads will result in faster query response times.

iConsole Standard Reports May Be Slower

BusinessObjects reports for CA Data Protection are based on the CA Data Protection data warehouse. iConsole standard reports are not.

However, if your users run a mix of data warehouse reports and non-data warehouse reports, the database cache is shared between both sets of tables. In particular, non-data warehouse reports may be slower because of the tables required by these reports are in the cache. Specifically, you may see a corresponding increase in IO for standard iConsole reports. If you can increase the physical memory to increase the cache size, this will remove or mitigate this problem.