Aimed at Modernizing IBM Z Mainframes, Watsonx Tackles COBOL-to-Java Conversion Amid Declining COBOL Expertise
IBM is making strides in modernizing mainframe applications for its Z systems with the latest addition to its Watsonx platform: a generative AI tool that can translate COBOL code into Java. This new feature is part of IBM’s broader strategy to support clients in overcoming the challenges of maintaining COBOL-based systems as the pool of experienced COBOL developers continues to dwindle. With this new capability, IBM aims to simplify the modernization process, enabling businesses to adapt faster to current technological needs.
The tool, called Watsonx Code Assistant for Z, builds on the foundation of IBM’s Watson Code Assistant, first introduced in May 2023. It leverages Watsonx.ai’s code foundational model to automatically generate Java code that is optimized to run on IBM’s Z operating system. This AI-powered code transformation is expected to drastically reduce the time and effort needed to transition from legacy COBOL applications to more modern, maintainable Java code, making mainframe systems more adaptable and future-proof.
One of the standout features of Watsonx Code Assistant for Z is its ability to automatically refactor COBOL code into modular, Java-based business services. Refactoring, a process where code is translated into a new language without altering its functionality, ensures that the core behavior of the application remains intact while making it easier to manage, update, and extend in the future. IBM demonstrated this capability in a recent video, emphasizing how the tool can extract key business logic from monolithic COBOL applications, transforming them into more scalable and efficient Java services.
To initiate this transformation, IBM recommends using its Application Discovery and Delivery Intelligence (ADDI) tool, which is currently under development. ADDI analyzes the target COBOL application to understand its dependencies and creates a metadata repository. This repository serves as the foundation for the refactoring process, ensuring that the code translation is accurate and complete. IBM notes that once the analysis is complete, developers can select specific business services for extraction and conversion into Java.
After COBOL code is extracted using a separate IBM tool, Watsonx Code Assistant for Z performs a deep analysis of the code and its associated data structures. Based on this analysis, it generates corresponding Java classes that replicate the functionality of the original COBOL business service. This method ensures that the new Java code integrates seamlessly with the existing system while preserving the operational logic of the COBOL application.
Once the translation process is complete, developers can compare the output of the Java code with that of the original COBOL code to ensure functional parity. IBM stresses that at any point during the refactoring process, developers have the option to enhance or modify the generated Java code, offering a level of flexibility and control that allows for further optimization and customization before the application goes into production. This iterative approach ensures that businesses can tailor the output to their specific requirements while gradually modernizing their legacy systems.