Sometimes the most natural path forward is the least expected. One example is predictive analytics, a powerful IT-based tool that is finding a new home … right at home, in IT. As reported at InformationWeek, predictive analytics is being applied to data from a network and systems management (NSM) to predict and head off data traffic slowdowns.
Predictive analytics is the branch of data mining that analyzes current and historical facts to make predictions about future events. And such is the mystique of predicting the future that people tend to think about predictive analytics on the grandest scale. Thus a notice for an upcoming conference on predictive analytics talks about the ongoing recession and how predictive analytics might conceivably have foreseen and thus forestalled it.
Somewhere in the meta-verse Isaac Asimov, whose Foundation Trilogy imagined the future-shaping science of psychohistory, must be smiling. But predictive analytics can be applied to domains closer to home. The results will be more immediate – and perhaps more reliable as well.
Watching Data Silo Patterns
Using predictive analytics to manage IT infrastructure begins with providing the analytics software with a rich accumulation of data from the NSM system, data that provides a baseline of system operating patterns.
Once the data has been provided, the predictive analytics software is set loose to chew on it. The software does not “know” what any of the source data means, but it is able to find statistical correlations and patterns in that data. These patterns in turn can identify points at which data flows lead to congestion.
Jake McTigue, author of the InformationWeek piece, offers a brief but vivid example of the sort of information that predictive analytics can provide: For the first two weeks of analysis we don’t get any alerts. However, at 9:30 a.m. Wednesday on week three we get an alert: “Building 2 performance will be degraded in 30 minutes, confidence: 85%.”
This is the sort of actionable information that any IT administrator would dearly love to have.
The Virtues of Reliable Data
The idea of using predictive analytics to help manage IT operations is not entirely new; one account of this process dates to 2008. But the recent spread of IT automation is helping to make this initiation process simpler and more reliable.
One basic challenge in predictive analytics – or, indeed, all statistical analysis – is the reliability of the data. In particular, survey or polling data that rely on human self-reporting are “notoriously unreliable.” (Ask people how much they exercise. Do you really believe their answers?)
But the sort of data that can be mined from NSM systems tends to be highly reliable. And if the system itself is substantially automated the reliability – as well as sheer data volume and rapid availability – will be even better.
All of which comes together to make the IT environment and data center operations one of the most attractive targets for applying predictive analytics in the enterprise.
