DB2 10.5 for Linux, UNIX, and Windows
DB2 10.5 for Linux, UNIX, and Windows
This parameter specifies a collect and reset interval, in minutes, for workload management (WLM) statistics.
Every x minutes, (where x is the value of the wlm_collect_int parameter) all workload management statistics are collected and sent to any active statistics event monitor; then the statistics are reset. If an active statistics event monitor exists, depending on how it was created, the statistics are written to a file, to a pipe, or to a table. If it does not exist, the statistics are only reset and not collected.
Collections occur at the specified interval times as measured relative to Sunday at 00:00:00. When the catalog member becomes active, the next collection occurs at the start of the next scheduled interval relative to this fixed time. The scheduled interval is not relative to the catalog member activation time. If a member is not active at the time of collection, no statistics are gathered for that member. For example, if the interval value was set to 60 and the catalog member was activated on 9:24 AM on Sunday, then the collections would be scheduled to occur each hour on the hour. Therefore, the next collection will occur at 10:00 AM. If the member is not active at 10:00 AM, then no statistics are gathered for that member.
The collect and reset process is initiated from the catalog member. The wlm_collect_int parameter must be specified on the catalog member. It is not used on other members.
The workload management statistics collected by a statistics event monitor can be used to monitor both short term and long term system behavior. A small interval can be used to obtain both short term and long term system behavior because the results can be merged together to obtain long term behavior. However, having to manually merge the results from different intervals complicates the analysis. If it is not required, a small interval unnecessarily increases the processing time. Therefore, reduce the interval to capture shorter term behavior, and increase the interval to reduce processing time when only analysis of long term behavior is sufficient.
The interval needs to be customized per database, not for each SQL request, or command invocation, or application. There are no other configuration parameters that need to be considered.