Long response times or no results returned when specifying historical collection time spans for some workspaces

For historical data collection in large datasets, the intervals set can affect performance dramatically. For example, when you select a long time span with a lot of data, the resulting query can take 60 seconds or longer to complete and use a lot of CPU.

About this task

Apply one or more of the following solutions to mitigate this problem:

Procedure

  1. Select Refresh Every in the View menu to access the submenu.
  2. Select On Demand in the submenu.
    If you set a short interval instead (60 or fewer seconds) the monitoring agent processing required might not complete within the refresh interval, causing subsequent requests to be queued. The monitoring agent then works continuously to process the query. High CPU utilization continues until the user navigates to another workspace or closes the Tivoli® Enterprise Portal.

What to do next

  • Specify longer historical collection intervals of 30 minutes or 1 hour, instead of the 15-minute defaults, for the attribute groups that generate this problem. Longer historical collection intervals reduce the number of rows per hour stored in the persistent data store.
  • Consider not collecting historical data for attribute groups that you are experiencing this problem with. Collect other data that provides the perspective on system performance or activity that you require.
  • Modify the KFW_REPORT_TERM_BREAK_POINT parameter in the KFWENV file of the Tivoli Enterprise Portal Server, which is located in the $CandleHome$\CNPS path. This parameter controls how many hours of historical data (counting back from the present time) are to be retrieved from the persistent data store (short-term history) datasets. The default is 86400 seconds (24 hours). A shorter time setting creates smaller sets of data to be searched in the persistent data store by the Tivoli Enterprise Portal. Older data (data excluded by changing this parameter) can be accessed if you are populating historical data in the Tivoli Data Warehouse. Be aware that modifying the KFW_REPORT_TERM_BREAK_POINT parameter affects all applications that are using the Tivoli Enterprise Portal Server.

    The Tivoli Enterprise Portal queries the Tivoli Data Warehouse for data older than the value of KFW_REPORT_TERM_BREAK_POINT. Configure a one-hour warehousing interval to ensure that data is available in the Tivoli Data Warehouse. A one-hour warehousing interval also improves performance of situations and real-time queries.