IT operations management provider Integrien has ratcheted up the sophisticated analysis of its Alive tool in a release to be launched Jan. 22 that can help predict and head off impending outages or performance problems.
The analytics in the intelligent systems management software allow the tool to learn normal behavior in activity across an n-tiered application. When it sees abnormal behavior, only then does it generate an alert.
"For every metric you collect, we use our dynamic thresholding, so we don't alert to just an event, but we alert to an abnormal precursor to a problem," said Steve Henning, vice president of products at the Irvine, Calif., company.
Alive takes advantage of existing monitoring tools such as those from IBM/Tivoli, Hewlett-Packard's OpenView and others and runs regressive analytics on top of the events those tools collect.
The tool also retains a record of abnormal activity that allows users to go back in time and build a model of the problem that shows what the precursors are to a specific problem. This allows IT managers to predict these problems before they reoccur, enabling operators to head them off.
Alive greatly reduces the amount of manual effort required to sift through thousands of static alerts generated by existing monitoring tools and helps IT operators more quickly pinpoint problems-even before they happen.
BT Infonet, which beta tested the new release, found it could reduce the number of IT operators required to troubleshoot problems from eight down to two using Alive, according to Armand Shirikian, the company's senior IT manager.
"Alive gives you a holistic view that pulls information from everywhere," he said. "With the new version, it can see something is brewing on the application side, and there is a very good chance that will affect the database. An engineer forwards the alert to the DBA, who can then take action. That really helped me to reduce the number of false positives that happen constantly."
The updated Alive 6 builds on its existing problem fingerprinting capability by adding new metric-to-metric and alarm-to-alarm correlation to more accurately trigger alerts and predict impending problems.