How Data Analytics Can Improve Your DevOps Processes

Organizations are keen to implement DevOps processes, often with the goal of delivering value to customers more quickly. Yet when they’re asked about results, the answer is often along the lines of, “We think it’s faster.” Rarely do organizations have precise data that enables them to determine whether their DevOps processes are actually working.

This is where data analytics comes in. It allows for the creation of metrics throughout the DevOps pipeline. Often, organizations have tools in place to measure these processes, but they may not be using these tools for this purpose, or they may not be measuring the right things. 

Modern applications are made up of many pieces, each of which has a task and talks to various components. Every time this happens, transactions occur that can be monitored. Visibility into these transactions can be the critical first step toward discovering issues, understanding why one area is working better than another and ultimately finding ways to improve outcomes.

A simple analysis might determine that version 1.1 of an application had 10 bugs while version 1.2 had only five, and then identify the changes that drove the improvement. Data could be gathered to measure wait times between handoffs, product quality or the number of errors. Analytics could also be applied to production: When the customer started using the product, what were the results? 

Over time, data tracking can paint a picture of how well teams are working. It can also reveal areas to shore up DevOps best practices to fix the problems that an organization may already have observed. Suppose you measure how quickly code is tested after it’s ready, for example. If three days pass before automated testing occurs, the automation didn’t accomplish much. You may have discovered that the real problem is that your testing services aren’t where they need to be.

Whether the metrics are simple or complicated, the goal of bringing data analytics into DevOps is to deliver value to customers more quickly, with fewer errors.

The biggest improvements often come about when data analytics results are shared with the people who need them. In my experience, it’s rare for developers to receive feedback from production about performance issues. But that information can be extremely useful at the beginning of the development cycle. It can help developers understand what’s working and where they may be falling short.

While development teams take an interest in performance data, testing teams love to know how many bugs made it through their tests. Architects are interested in whether the infrastructure they designed is supporting the application in the right way. Many application monitoring solutions have a dashboard where much of this information can be visualized. Each group could have its own dashboard, or all the data could be integrated into one master dashboard.

In the absence of data, there’s a natural tendency to assume everything is fine. But precise measurement and analysis using data analytics can help organizations spot where DevOps processes are achieving the desired results and where they may be able to make improvements.

Story by Dave Roesch, a senior solution architect for CDW. With over 20 years as a solution architect for both data center and development platforms, Dave has a proven track record in helping his customers find solutions to increase speed and accuracy in their operations to deliver business value on demand. He has DevOps, SAFe Agile Program Consultant (SPC) and Design Thinking Leader certifications, and he has experience helping customers adopt a culture of continuous improvement. He has deep skills and experience in a variety of technologies, such as automation (for example, infrastructure as code) integration (such as continuous integration/continuous delivery) and cloud. His background includes experience in many industries, including computer, financial, manufacturing, retail, government (federal and state), education and healthcare in the United States and Asia.