Predictive Maintenance in IT
Think of any “mission-critical” business process: order processing, being able to connect with customers (phone and email), shipping orders—it’s hard to imagine any that aren’t tied to IT. IT really is mission control of any organization, large or small.
This means preventing outages before they impact your business can be transformative.
One way to do this is via predictive maintenance.
In the past organizations have done preventative maintenance…basically replacing them before they fail based on an educated guess. For example if servers normally fail around three years, you’d proactively replace them at two and a half years. There are two issues with this approach: some of those servers would have kept running for longer, so you’ve wasted money, but more importantly some may fail sooner so you still expose your organization to risk.
Predictive maintenance combines all the variables that could contribute to a failure, like the manufacturer, how many times the server has crashed, temperature, astrological sign (okay, that may be a stretch) but basically way more variables than a human can compute.
We collected all this information via logs and master data in SAP IT Operations Analytics (SAP ITOA). SAP Predictive Analytics then accessed the HANA data and ran a model for probability for failure. The result was exposed as a SQL view and pushed back into SAP ITOA.
We see that we have seven servers with a 9% probability of failure
But where it gets really interesting is on the right hand side—we are enriching the model with master data. In this case the business owner, support team, and SLA.
As we go through these seven servers with a high probability of failure we see only two have a high SLA. So we are able to prioritize replacing them first if resources are limited. We also can see other information about the servers like the current status and how heavily they have been used over the last week.
Here is a video walking through this scenario…
More information about other ways Predictive Analytics can benefit IT organizations (like reducing noise, correlating causes of issues, and forecasting expected performance) are detailed in this blog.