Self-Learning Predictive Analytics for MS Perfmon

Every day, thousands of IT support teams turn to Microsoft's Performance Monitor and Windows® event logs to diagnose IT problems. For simple issues, these tools can be invaluable, on their own. When problems start to get a little more complicated, however, you can be stuck looking at thousands of data points and hundreds of events trying to determine which are related and which are just coincidence.

Prelert applies the industry's most advanced Artificial Intelligence algorithms to Perfmon and event log data to self-learn the behavior and metric relationships of even the most complex environments. We provide an 'out-of-the-box' intuitive, real-time dashboard that identifies problems and provides their root cause analysis as they develop and often before users are impacted.

 

Faster IT Troubleshooting & Problem Analysis

• Analyzes time series and event data in real-time
• Auto baselines individual metrics
• Self-learns relationships between data points
• Discovers patterns of activity that define behavior
• Identifies and often predicts problems

• Distills tens of thousands of data points that are 'coincidental' to the few that are related to the problem
• Provides the 'root cause analysis' narrative and supportive detail

 

Prelert reduces MTTR as much as 90% by helping users identify the cause of problems as they develop. It eliminates the need to define thresholds and alert rules and significantly reduces the workload associated with defining and maintaining dashboards.

Users get a highly intuitive dashboard out-of-the-box that distills mountains of data to the information they need to support critical environments.

 

Outages in Microsoft Exchange®

A large global bank experiences intermittent incidents where the Exchange STORE process fails and must be restarted manually before mail can be delivered. For months, the Exchange support team manually reviewed Perfmon and Windows event data for any clues as to the cause.

After eyeballing millions of performance metrics and tens of thousands of event logs, all they know is that there are a large numbers of significant spikes in network, disk and CPU metrics on the Exchange servers coincidental to the process failure. They can't tell which are a cause, result or merely a random coincidence.

 

Prelert - Self-Learning Analytics

With our built-in adapter for Perfmon and Event Logs, Prelert is easily installed on the relevant servers. Prelert's highly advanced analytics processes millions of performance metrics and event logs in real time. Prelert self-learns the baseline behavior of the system as well as how all the components interact to deliver the Exchange service.

When the problem reoccurs, Prelert identifies it and provides and isolates the related set of metrics that establish the root cause. One of the domain controllers misbehaves prior to the STORE process failure consistently.

When a problem occurs, Prelert has already isolated the set of monitoring data related to the issue. The result is not only the proactive identification of a problem, often before users experience disruption, but the real-time isolation of the problem's root cause.