IBM Tivoli Monitoring > Version 6.3 Fix Pack 2 > Administrator's Guide > Manage historical data > Performance impact of historical data requests
IBM Tivoli Monitoring, Version 6.3 Fix Pack 2
Use a data mart to improve long or complex queries
This section describes the how a data mart can be used to increase the performance of your primary datastore.
Within the Tivoli Management Services infrastructure, the warehouse proxy regularly inserts new data from the short-term history files into the data warehouse tables. This detailed data is derived by queries from historical views to report this information and can be derived by queries from an external reporting tool. Any active datastore needs to balance read and write activity to maximize performance of the datastore. The data warehouse has periodic write activity balanced with frequent read activity for formatting and creating reports. Under some circumstances (especially formatting reports over long durations or executing complex queries), the database read and write activity can become unbalanced and result in abnormal wait times. Under these circumstances, you can significantly improve performance by adding a secondary datastore, commonly called a data mart, for reports from causing long or complex data queries.
Depending upon the reporting requirements, there are two mechanisms that can be used, exploiting the open interfaces that are included with the warehouse:
- If the complete database is required, use the Database Replication Facilities of the Tivoli Data Warehouse RDBMS.
- Write and schedule SQL extract scripts, similar to ETL Scripts in Tivoli Data Warehouse V1.x, to extract desired data elements at a scheduled interval from the Tivoli Data Warehouse and populate a reporting database. This reporting database can be optimized for use by an external reporting tool, just like data marts were used in Tivoli Data Warehouse V1.x. These scripts can be SQL Scripts, shell scripts, or PERL scripts.
- Sample data mart SQL script for IBM Tivoli Monitoring
The following SQL script is an sample script of how you can create and populate a data mart. Your actual script needs to be revised to reflect your environment.
Parent topic:
Performance impact of historical data requests