Monitoring NNM Systems With ITO and MeasureWare

Managing large networks with many management domains will probably require 15 or so collection stations and two management stations. It is a considerable task to manage that many systems, one that requires automated tools.

If you load and configure HP MeasureWare agents on all NNM systems from a central system you can then use HP PerfView to monitor all critical system resources, some of which include:

• RAM utilization

• virtual memory (VM) utilization

• VM paging rate

• CPU utilization for all CPUs

• disk I/O rate

• queue depth of the disk controllers

• network interface error rate

There is a fair amount of opinion about threshold values for these and other metrics, and there are preset values for them. If these presets generate too many alerts when the NNM system is otherwise working well, then by all means increase the threshold values.

IT Operations (ITO) agents should also be loaded and configured on all NNM systems. The agent may be tailored to monitor specific conditions using customized scripts. During the NNM pilot project (you did do a pilot, right?), abnormal conditions may be discovered manually at first. Write down the commands you used to detect or troubleshoot the problem for later automation. Some examples are:

• monitor load factor rather than CPU utilization

• check management station and collection station communications

• verify that the read/write ovw session is active

• check that all NNM daemons are active

• watch log files for well-known error conditions

• monitor processes that use excessive CPU time

• detect hung ovw sessions belonging to logged off users

The ITO application can be loaded onto the same monitoring system that PerfView resides. ITO can then detect MeasureWare alarms. You have now created a MOM (manager of managers). Locate this system at the corporate network management center.