One of the big challenges that enterprises face with monitoring tools relates to “false alerts”. Administrators often get too many alerts when nothing is wrong. So very often, they end up turning the thresholds up in such a way that they don’t get many alerts. If this is done for all the metrics collected, this may end up defeating the purpose of having a monitoring system!
Setting thresholds for metrics being collected by the monitoring system is often a challenge. Many a times this is done manually and requires a lot of domain knowledge and expertise. Often, the threshold settings also have to be done differently from one server to another (a bigger server can take more load than a smaller one).
Many new age monitoring systems have used analytical techniques that use historical performance data to intelligently set thresholds for metrics. As we have discussed in a new whitepaper, even such approaches have challenges. The new whitepaper “Make IT Service Monitoring Simple and Proactive with Intelligent Thresholding and Alerting” touches upon these and many other topics that have to be considered when determining how thresholds are computed for metrics and how alerting is done by the monitoring system.
All whitepapers from eG Innovations: http://www.eginnovations.com/whitepaper/whitepaper-request.htm