Do your mainframe applications have tens of millions of lines of code? Are you building and managing critical applications that are on par with the complexity of the mainframe's very own z/OS operating system? Are you searching for a way to simplify your development lifecycle and accelerate time-to-market? Look no further than IBM's very own Dependency Based Build (DBB) tool for automated build and dependency management.
Like many other companies in the mainframe space, building the z/OS platform and many of its components was a process in IBM that involved legacy tools and a fair amount of custom-built code to move from our developer's machines to make them publicly available to you. As you can imagine, these products have been touched by 1000+ developers, encompassing tens of millions of lines of code. Managing such a large body of work is challenging, especially considering what the existing processes looked like operationally. Modernization became mandatory for the IBM team managing builds for critical software products such as z/OS and Geographically Dispersed Parallel Sysplex® (GDPS).
John Bouzakis, Chief Product Owner of IBM's z/OS DevOps and Release Engineering Team, shares his team's experience from starting a project to redesigning how IBM Z builds, packages and delivers z/OS products, and successfully delivering via modern CI/CD pipelines. Today, this team stands as a model of DevOps excellence, powered by innovation, collaboration and a relentless drive for continuous improvements.
Business Challenges
This team's previous processes used many complex and custom tools, some of them duplicating similar functions. This resulted in lower productivity, frustration, and higher cost of development and tool support. A portion of the processes were manual, disconnected and error prone. It took a lot of time and effort to get even small fixes out of the door. As time went on, it became apparent that it was time to improve and modernize their IBM Z software development processes.
A subset of the development team was tasked with the challenge of migrating over 50 product elements, with tens of millions of lines of code, to modern tools and processes. This included large IBM product teams such as z/OS, GDPS and IBM Z Platform for Apache Spark. The previous process used a build tool which was created by the team. While it served its purpose, it was difficult to maintain, hard to scale and created bottlenecks. The steep learning curve also made onboarding new developers a challenge, impacting retention and productivity. This is when they realized that IBM Dependency Based Build (DBB) would be the perfect solution to enhancing and automating their build processes due to its functionality, adaptability, scalability, and maintainability.
Automating Builds
This team became one of the earliest adopters of DBB 1.0. While DBB provided the toolkit and APIs, the team had to add scripts and perform integrations themselves. DBB later recognized this gap and incorporated similar capabilities directly into the product. One of these enhancements was the adoption of YAML configuration files for the zBuilder build framework, stemming from the inputs provided by this team.
This collaboration also helped shape DBB's evolution, providing input for simplification and usability. Working with DBB development benefited both teams, adding functionality to the product and helping the team overcome obstacles. As an early adopter operating at customer-like scale, this team surfaced issues early — problems that other IBM clients would have encountered had they not been identified and addressed first. This feedback drove DBB's development direction while yielding strong product support in return. The initial complexity of upgrading from DBB 1.0 to DBB 2.0 was a challenge due to the number of migration actions. In contrast, the transition from DBB 2.0 to DBB 3.0 was much smoother — the processes significantly more refined and fewer upgrade actions required.
Today the team runs DBB 3.0.5.1 across nearly all elements, integrated with Jenkins and GitHub. The team also built internal automation tools for installs which reduced deployment times from 24 hours to just 1 hour.
Below is an architecture overview showcasing the toolchain used by John's team.
Aside from tools, the team also focused on DevOps culture – breaking silos, encouraging innovation and empowering developers.
As a result of this transformation, the team continues to raise the bar — improving developer experience while driving greater reliability, performance, and security.
What's Next?
The team is already leveraging AI in their development work and is actively exploring how to bring AI capabilities into their DevOps pipeline. As that journey evolves, their story is far from over.
This story is one of vision, grit and evolution. From a legacy toolchain to a modern DevOps powerhouse, they have redefined what is possible. Walking through how IBM uses its own products to modernize mainframe development, it shows that with planning and the right toolset it is possible to achieve dramatic improvements in speed, reliability, and developer experience. DBB has empowered this team to build smarter, faster and with greater confidence.
As John Bouzakis shared:
DBB was the cornerstone of our build modernization. Automating builds on the mainframe at our scale — tens of millions of lines of code across over 1,000 developers — would not have been possible without it. It gave us the foundation for our broader mainframe DevOps modernization.