Hitachi

JP1 Version 12 JP1/Performance Management - Remote Monitor for Virtual Machine Description, User's Guide and Reference


1.3.1 General procedure for performance data collection tasks

This subsection describes the flow of data and the general procedure for collecting performance data.

Organization of this subsection

(1) Data flow during performance data collection

The following figure shows data flow that occurs when performance data is collected.

Figure 1‒9: Data flow occurring during collection of performance data

[Figure]

The Remote Monitor Collector service and the collection process operate asynchronously.

(2) General procedure for collecting performance data

The Remote Monitor Collector service of PFM - RM for Virtual Machine collects records as specified by the Collection Interval setting of each record. The collection process operates based on the Interval setting specified during setup of the instance environment.

The following figure shows an example of the procedure for collecting performance data.

Figure 1‒10: Example of the procedure for collecting performance data

[Figure]

#1

Although performance data collection is to start at 00:05, the processing does not start and a KAVL20517-W message is output, because a temporary performance information file has not been created.

#2

If an alarm is bound in the period between collections at 00:15 and 00:20, whether to evaluate the alarm at 00:20 (first collection after the binding) depends on whether log data is collected by the record used by that alarm.

  • If log data is collected by the record that is used by the bound alarm

    The alarm is not evaluated.

  • If log data is not collected by the record that is used by the bound alarm

    The alarm is evaluated.

#3

Because the collection that started at 00:14 did not end by 00:19, the collection that was to start at 00:19 is skipped.

#4

During the collection at 00:20, in the same way as for the collection at 00:15, the temporary performance information file that was output at 00:15 is loaded.

If log data has been collected or the alarm is evaluated, no performance data is collected. At this time, a KAVL20516-W message is output.

For collection of data for a real-time report, the contents of the temporary performance information file that was output at 00:11 are displayed in the same way as for collection at 00:15.

Important

Depending on the time when the PFM - RM for Virtual Machine service starts, the message KAVL20516-W can be output repeatedly even though a problem (for example, the system is temporarily under high load or takes a long time to communicate with the monitoring target) has not occurred. In that case, you need to change the settings. For details, see Appendix N. What to Do When the Message KAVL20516-W Is Output Repeatedly.

Note
  • When performance data is stored in the Store database as log data, the data that was collected during the interval before the storage time is stored. The maximum time that is required from the time performance data collection starts until the data is stored in the Store database is approximately the Interval time specified during instance environment setup (the default is 300 seconds).

  • When a real-time report is displayed, the contents of the temporary performance information file existing at that time are displayed. When a displayed real-time report is refreshed from PFM - Web Console, the display is updated to show the current contents of the temporary performance information file.

  • Collection processing might be delayed depending on the number of monitored hosts and the load status during processing.

  • Alarm evaluation is performed with performance data that was collected during the interval before the time at which evaluation is performed. The maximum time that is required from the time performance data collection starts until alarm evaluation ends is approximately the Interval time specified during instance environment setup (the default is 300 seconds).

    Note that performance data collected during an earlier interval might be used for evaluation, depending on the alarm-binding time or collection delay, if the bound alarm does not use a record for which no log data is collected.