Data, DevOps, and the Rest of the Business

Source – informationweek.com

So, what’s a guy best known as an advocate for DevOps doing in a data and analytics track session at Interop ITX? Did Andi Mann, chief technology advocate for DevOps company Splunk take a wrong turn? Afterall, analytics has grown to be associated with data scientists and their models, AI, visualization, and algorithms, not rolling out application code updates to fields of servers.

Actually, the high-energy Mann was there to show how data is at the heart of DevOps and, as his 50-minute presentation went along, subtly show how the DevOps concept can play out in any number of business processes.

“You use data from machines to learn what is abnormal. Once you have analytics to identify what is normal, you can act on what is abnormal,” said Mann during an opening overview on analytics. He used the example of a homeowner checking on the status of their home loan, and being dissatisfied with the lender’s service. That customer is ripe to jump to another bank. Yet, if the current lender’s system could tell from the data that the customer was unhappy — that their situation was the abnormal — through visualization tools, the bank could take proactive steps to save the relationship.

That’s a fairly typical scenario for data and analytics professionals to consider. Now step back into the DevOps arena.

Mann outlined eight steps in the typical DevOps process, the software development lifecycle (SDLC) that runs from “idea” through “build” up to “deploy” and “monitor”.  At each of those stages, he listed three or four commonly used tools.

“Each tool is giving off data, like which developers are doing how much work on which system,” he said. The tools also provide data on things like the quality of the application and where issues and delays may crop up. “That data adds value across the software development lifecycle,” he noted.

As those tools feed data into the analytics system, a dashboard provides a view into the data, an at-a-glance look into the DevOps team’s progress, not just with dozens of numbers but with the green-yellow-red color coding that is easy for all of us to understand. Simply, red is bad, green is good, yellow says you need to keep an eye on things.

It’s ingrained in our brains to recognize red — as in the 24,000-millisecond response time below — as trouble, while the 1,600-millisecond result sits happily in the green.

Mann applied the same concepts to business sectors outside of DevOps. For example, a healthcare provider can map the progress of patients from an emergency room, through the in-hospital stay, to discharge. Where are the delays in the various steps? Where are the quality of care issues? Where are the costs?

Then he showed how data collected at multiple points provides a retailer or other business with a view into the customer experience as they move throughout a store or within a website, with “clickflow” data.

In the end, feeding those bits of data taken by diverse tools at each stage of a business process — whether building an application or improving customer experiences — can provide managers and staff with actionable data, enabling them to improve quality, build products faster, or sell more of those products.