Software Engineering
IBM team outlines framework for converting legacy mainframe systems to modern APIs
A new approach helps organizations expose critical mainframe functionality through APIs while maintaining performance and reliability.
Legacy mainframe systems continue to handle mission-critical workloads in banking, retail, and healthcare, but their capabilities often remain locked behind outdated interfaces. A team of researchers has developed a framework to help organizations modernize these systems by automatically identifying and implementing Application Programming Interfaces (APIs) that expose mainframe functionality to newer applications.
The Challenge of Modernizing Legacy Code#
Transforming mainframe applications to support API access is far from straightforward. Organizations must grapple with understanding convoluted program logic, isolating interconnected components, introducing new architectural elements, and ensuring that existing functionality and performance guarantees—such as strict turnaround time requirements—remain intact.
How the Framework Works#
The proposed approach identifies candidate APIs by analyzing various artifacts within the codebase: transactions, user screens, control-flow logic, inter-service communications, business rules, and data operations. Once candidates are identified, static analysis techniques—including liveness analysis and reaching definitions—automatically compute the API signatures by determining which fields belong in the request and response payloads.
Real-World Validation#
The framework was tested through interviews with nine mainframe developers, each averaging 15 years of experience. Researchers evaluated the methodology using GENAPP, a publicly available test application, as well as two proprietary industrial mainframe systems. The findings demonstrated that the framework successfully uncovers additional candidate APIs and meaningfully reduces the implementation effort required for APIfication.
The approach has already proven its practical value: its API-signature computation capabilities have been integrated into IBM watsonx Code Assistant for Z Refactoring Assistant. Researchers further validated the identified APIs by deploying and executing them on actual IBM Z mainframe hardware, confirming the method's real-world applicability.
Implications for Organizations#
This framework offers a systematic path for legacy mainframe custodians to unlock the value of their systems in cloud-native and microservices-based architectures without the risks and expenses of wholesale rewrites.