At IBM, we have a history of applying analytics and now cognitive computing to improve the way we do things. Recently, one of my development teams wondered if there is a means of applying analytics and cognitive computing to improve a DevOps process? Out came the idea for a new product, Application Delivery Intelligence (ADI).
Throughout the DevOps lifecycle we produce a lot of data, including code, designs, test cases, test execution records, and operational information about applications. Application Delivery Intelligence is about analyzing this information to guide us in how to work more effectively. Our intent is to incrementally deliver more capabilities, allowing teams to work smarter. While our focus is centered on DevOps, a lot of this Application Delivery Intelligence will apply to companies using other approaches to development and delivery.
One of the first focus areas for the Application Delivery Intelligence team has been what we call Test Optimization. As companies adopt a DevOps style of development, it means that they need to shift testing left.
A challenge that teams are facing with this shift is the cost of doing more frequent testing. Common questions are:
- How can you afford the labor and MIPS cost associated with frequent regression testing by focusing on the most relevant tests?
- How do you know whether you have quality exposure in the test effort?
- Are you trending in the right direction?
We think the sky is the limit for ADI. What problems can we solve by applying analytics and the power of Watson to optimize how we develop software on a mainframe?
Tell us what big problems YOU think we should focus on solving–post a comment below.