Defining data collection intervals

The collection intervals control how often real-time data is collected. Each time you collect data, the monitoring agent incurs a certain amount of CPU usage. How much processing is incurred depends on the number and type of resources you monitor.

Collection intervals are specified in minutes and have a valid range of 1 to 60. The default value for a collection interval is 5 minutes. This default value satisfies the monitoring needs of many customers who require near real time access to data.

The collection intervals are specified with the PARMGEN method, or you can modify a collection interval by executing a MODIFY procname,KN3FCCMD command.

For more information, see topic KN3FCCMD command reference in the Parameter Reference Guide.

Specify values for the following collection intervals:
  • KN3_TCP_SAMPLE_INTERVAL: the collection interval used to control most of the data collection, with the exceptions that follow.
  • KN3_SNA_VTAM_SNAC_SNACINTV: controls the Buffer Pool and VTAM® Environment data collection. This interval is only used if you enter Y for KN3_SNA_VTAM_COLLECT_DATA.

The z/OS® Communications Server network management interface (NMI) and Performance Monitoring Interfaces maintain performance information between collection intervals. For example, spikes are identified in such attributes as Maximum Authorized Private Storage Allocated. Over- and under-utilization values are identified in such attributes as Total Bytes Sent.

The network performance data displayed in the Tivoli Enterprise Portal is the latest collection interval; a user request to view data does not initiate a new data collection. Therefore, increasing the frequency of data collection (that is, decreasing the collection interval) by the monitoring agent generally results in your network operators viewing more recent data but results in higher CPU consumption by the monitoring agent. By decreasing the frequency of data collection (for example, from 5 to 15 minutes), you reduce the number of times per hour that the monitoring agent collects data (from 12 to 4 times per hour), thus reducing the total CPU consumption (by a factor of approximately 3).

You can further reduce the CPU consumption by changing the KN3_TCP_ROUTE_TBL_FREQ parameter using the PARMGEN method or on a KN3FCCMD START ROUTE command.

For more information, see topic KN3FCCMD command reference in the Parameter Reference Guide.

The KN3_TCP_ROUTE_TBL_FREQ parameter controls how often the Routing Table Statistics (Gateway table) are collected. By default, this data is collected once every 10 collection intervals (once every 50 minutes when the default collection interval is used). The frequency does not affect the amount of storage used; it affects only CPU consumption.

Set the collection intervals and routing table frequency to meet your monitoring needs, ensuring that you are promptly alerted to problems so they can be resolved quickly. But, also be aware that each data collection comes with a cost in CPU usage.

Run IBM® Z OMEGAMON® AI for Networks and use Tivoli® OMEGAMON® on z/OS or the Resource Measurement Facility (RMF) to measure CPU usage. Adjust the collection intervals until you have achieved the desired balance between currency of data and CPU usage.