System Performance Monitoring

Whenever a system resource, such as a processor core, disk or network interface is occupied by a request, it is unavailable for processing other requests. Pending requests must therefore wait for the resource to become available before they can be run.

Where a resource is too busy to keep up with all of its requests it can become a bottleneck to the system with the higher the percentage time each resource is occupied resulting in an increased delay on subsequent requests. Such a resource would be thought of as critical to the performance of the system.

Performance Monitoring available in Agent 6.5 onwards for servers and from Release Candidate for workstations, may be used as a form of preventive maintenance to identify any bottlenecks or critical resources on the monitored servers – so they system can be fine tuned or upgraded – before they cause increased delays and other problems.

Through Performance Monitoring, data is obtained that may prove useful when diagnosing any problems as well as allow you to build a performance profile of the system which can be used to identify abnormal load.

 

Performance Monitoring consists of five checks:

Processor Utilization
Processor Queue Length
Memory Usage
Network Interface Utilization
Physical Disk

The Dashboard displays the average and maximum values of the performance counter over the last Agent cycle and where the average exceeds the specified threshold, configurable in the Agent, an Alert is generated.

The Performance and Bandwidth Monitoring History Reports show the results of Performance Monitoring or Bandwidth Monitoring checks, on an individual server, over the last 24 hours and 8 day.

The number of consecutive failures before an alert is sent is configured in the Dashboard and this suppression may be applied globally across the client base as well as for each individual server.