News

Compuware Adds Machine Learning DevOps Muscle to Mainframe CI Solution

• By John K. Waters
• July 2, 2019

Compuware is advancing its ongoing mission to mainstream the mainframe for developers by integrating its machine-learning-driven zAdviser analytics tool with is ISPW mainframe continuous integration/continuous delivery (CI/CD) solution.

zAdviser is a free service for Compuware customers designed to leverage machine learning "to continuously improve mainframe software development and delivery outcomes," the company explained. It uses machine learning to find correlations among developer behaviors and key performance indicators (KPIs) based on DevOps data and Compuware product usage data. The combination of ISPW usage data with software delivery KPI data from zAdviser provides deeper insights into how an organization's dev teams work and where future resources might be applied most effectively, explained Compuware product manager Steven Kansa.

"We work with a variety of different customers across the globe," Kansa told ADTmag, "and one of the most common challenges we see among those organizations is the need to understand developer productivity. They're asking, how are our developers operating today, and what can we do to help improve their productivity?"

They're also facing the unique challenge of senior developer attrition, Kansa said, as a growing number of existing mainframe mavens approach retirement and new developers with considerably less experience in these environments take their place. Add to that the ongoing pressure everyone is feeling to increase the quality, velocity, and efficiency to meet changing business demands and new competitors.

The combination of ISPW with zAdviser addresses these challenges by providing different dashboards that show companies how their development organizations are operating, and to give them a way to look for efficiencies that they can gain in refactoring code that might be causing the most problems and bottlenecks within their delivery pipeline.

"This is about finding the patterns that make you more productive," Kansa said. "What are the things that people are doing that help improve throughput efficiency and quality. Where is the technical debt within your system, and potentially obsolete code?"

"The truth is, in large organizations, they often don't know much about the quality and effectiveness of their developers," said Compuware CEO Chris O'Malley. "Are they awesome? Are they average? We've [at Compuware] have accumulated all this data from working with customers over decades—huge amounts of it across many customers—and patterns emerge in that data around how the work gets done. So, we can use machine learning and pattern recognition to ask, what are the ways of working that are associated with high performance."

The zAdviser dashboards provide dev teams with visualizations of the constraints in changing source code, so they can focus their development efforts to better meet business and customer requirements. The idea is to provide a means of answering several key questions, O'Malley explained, including:

• What modules are contributing most to our technical debt?
• Where do we need to focus our code revitalizing and refactoring efforts?
• Where are the constraints that, if improved, could have the biggest positive impact on improving future flow?

He also said the zAdviser/ISPW combo will provide more insightful developer productivity data to help teams answer such questions as:

• How much time is spent developing and testing code?
• How long does production-ready code wait before being promoted into production?
• What's the time difference between code first being checked out to it running in production?
• How are regressions and fallbacks affecting development flow?

"This release provides capabilities that allow companies to focus on the right things, understand as they're tuning dials, if you will, how what they're doing is actually impacting their productivity," Kansa said. "And we're using machine learning combined with the data to give them those capabilities."

Compuware is also announcing expanded integration of ISPW and Git, the version control software-of-choice for many distributed development teams. This integration is meant to help developers who don't have much mainframe experience work more confidently with mainframe source code. Changes in Git are now automatically synchronized back into the mainframe, where ISPW's automated build, deploy, and fallback capabilities can be leveraged in a CI/CD pipeline. By employing consistent methods for all code across all platforms, the company claims, "customers will gain consistent visibility into their diverse codebases and ease the process of managing code throughout the enterprise."

"Reducing technical debt inherent in mainframe applications is essential for any organization looking to incorporate mainframe applications into their CI/CD pipelines," said Jason Bloomberg, president of industry analyst firm Intellyx, in a statement. "With the new analytics from zAdviser ,along with modern mainframe CI/CD tools like ISPW and its integration with Git, it's possible to break down the silos separating mainframe development from the rest of the DevOps effort, thus reducing technical debt and extending the speed, quality, and collaboration benefits of CI/CD to the mainframe."

"In large enterprises, developers matter in a big way," O'Malley said. "And they really should be thought of as teams of performance athletes. In fact, I'd argue they're the most important people in the organization," adding in a statement: "Companies can transform their most critical back-end platforms by enabling a preferred developer experience, automating manual-process constraints, integrating with best-in-class tools, and inspiring a culture of continuous improvement inspired and guided by software delivery KPIs of velocity, quality and efficiency."