Making the transition from waterfall to an Agile-DevOps culture poses numerous problems for any size company. When a large corporation undertakes the transition, though, the issues have the potential to climb exponentially unless they have Software Intelligence to understand structural quality and gain insight into the overall performance of their application portfolios.
In 2015, amidst a rapidly evolving home mortgage market, Fannie Mae recognized it needed to transition to Agile-DevOps to keep up with the progress of its competitors. With only 10 teams using Agile-DevOps, releases were taking between nine and 18 months, and they lacked intelligence around software quality.
Initial efforts to transition were difficult, though, thanks in large part to the morass that was Fannie Mae’s network – a complex, technical-polyglot ecosystem composed of 461 applications, several hundred utilities, and almost 18,000 open source components. Major updates to the platform were taking up to nine months when the company needed them in less than one. It became apparent that those overseeing the transition would also need to track improvements to prove to management that the gained efficiencies warranted the transition.
In the most recent issue of IEEE Software Magazine, Bill Curtis, senior vice president and chief scientist at CAST, and Barry Snyder, product manager for the DevOps release pipeline, discussed how implementing an automated analysis and measurement solution facilitated Fannie Mae’s transition from a traditional waterfall approach to one designed around an Agile-DevOps methodology. They broke down the process into six segments:
By implementing an “out of the box,” automated application assessment and measurement solution like CAST’s Application Intelligence Platform, the company performed a language-agnostic analysis of the entire, multi-tiered, system. AIP evaluated code against 1,200 rules of good architecture to develop a Total Quality Index (TQI), which provided a starting point for improvements, and by incorporating a self-service interface, teams could perform automated analysis at their own pace, which increased platform usage by 481 percent.
- Productivity Analysis
Per executive management, there needed to be a “periodic empirical evaluation of progress in quality, productivity, and delivery speed.” Fannie Mae opted to measure functional size for this using Automated Function Point (AFP) analysis. The automatically generated AFPs provided an outcome-based productivity evaluation, which identified what processes worked and which did not.
- Structural Quality Analysis
Using automated analysis during sprints enabled teams to detect flaws in application quality. This early detection allowed teams to address issues within a day or two rather than waiting to find and fix them after the release. The net result was a vast improvement in the five application quality health factors – robustness, performance, security, changeability, and transferability – and a higher TQI than with waterfall.
- Aligning Metrics
At first, Fannie Mae needed one-off solutions to collate data across the enterprise. The data gleaned through automated analysis and measurement provided the information the company needed to eliminate those silos. With measurements and analytics aligned across the enterprise, Fannie Mae had holistic insight into the transition to Agile-DevOps.
- Analyzing Improvement
Automatically monitoring and measuring productivity in terms of quality and functionality delivered data that made the company aware when teams achieved increased functionality without a decrease in quality. Curtis and Snyder placed productivity gains following adoption of Agile-DevOps and automated analysis at an average of 28 percent across teams.
- Baselines and Self-Evaluation
Analytics collected through the automated assessment solution established a baseline for evaluating the effectiveness as each team adopted the practices. Fannie Mae took that one step further and established an enterprise-level relationship between structural quality and the completeness of adopted practices, which gave them insight into the effectiveness of Agile-DevOps practices.
Fannie Mae’s adoption of automated application analysis and measurement contributed significantly to improvements in productivity and cycle-time gains, much of which resulted from early detection of structural flaws in the early stages of the project. Eventually, with analytics aligned across the enterprise, executive management had the empiric data it needed to justify the move to Agile-DevOps.