Leveraging Predictive Analytics in IT

Being in analytics I feel sort of bad for IT. They do a ton of work to deploy BI for their business users, but don’t typically benefit since they’re not the end-user.

But we’ve seen a shift where IT is starting to better leverage the analytical capabilities used by the business.

SAP IT Operations Analytics (SAP ITOA) is an analytic tool for IT that gives a single view of all devices within IT from applications, to network devices, to servers. Anything that generates a syslog.

 

While realtime monitoring and alerting are cool, no one likes having their lunch interrupted because of a text message saying there’s a performance issue. Avoiding degradation or outages in advance could have material benefit to your organization.  So, we’ve taken the expertise SAP has built around Predictive Analytics in other areas of the business and applied it to the datacenter. Here are some practical examples…

 

Notifications about Deviations from Expected Behavior

A simple example of Predictive Analytics in the data center is comparing expected behavior to actual.

 

Let’s take CPU usage…if it spikes too high that’s bad, you’ll start to see poor performance of your applications. Similarly, if it drops that could be indicative of an issue, perhaps an application has crashed. SAP ITOA predicts performance as expected and sets maximum and minimum error bars based on deviations we’ve seen previously. If actual performance deviates from predicted performance beyond the max and min deviations you can take a proactive action to remedy the situation before it becomes an issue.

 

ITOA predictive.jpg

 

Predictive Maintenance

Monitoring will alert you when a system or component needs reactive repair.  MTBF data will allow you to move toward preventive maintenance. To operate more efficiently and to mitigate downtime you will need early alerts that will allow you to do predictive maintenance.  Companies that have implemented predictive maintenance have seen cost improvements of up to 40% over their preventive expenditures.

 

Reducing Log Noise

Data centers generate a lot of logs (one of ours is seeing half a billion events per day!). But not every log entry is a relevant event.

 

Take the example of URL filtering in firewall logs. The administrator may set logs to capture when a specific URL or category of website is accessed, but if the access is within the corporate policy (maybe only accessing shopping websites outside work hours) there is no need to ingest and do analytics on the event.

Predictive Analytics can help IT administrators identify what is an event, like someone accessing a website outside permitted conditions, and non-events, which can be deleted or archived.

 

Drawing out Correlations

Predictive automation will also allow your operations center professionals to improve their knowledge base. By interacting directly with the log data they can determine which variables had the most impact on degraded performance or outages.  Your team can conduct their own assessments to determine root cause analysis.

 

Combining the power of SAP ITOA with Predictive Analytics will allow you to:

  • Have advanced notification of incidents before they occur
  • Improve the thresholds set by monitoring tools
  • Forecast when applications, systems and networks will be at utilization rates that will impact performance
  • Increase the knowledge base of their operations staff.

 

 

To report this post you need to login first.

2 Comments

You must be Logged on to comment or reply to a post.

Leave a Reply