Overview.
Capabilities.
Challenges.
Client’s outdated systems were becoming increasingly difficult and expensive to maintain, while also hindering the integration of new functionalities which are crucial for staying competitive in the rapidly evolving travel industry.
Recognizing the need for modernization, the client had been working on migrating their codebase from Assembler to Java over the past decade. However, their manual approach to this transformation proved to be highly labour-intensive and time-consuming. The process required business analysts to meticulously review existing code and create detailed Product Technical Specification Documents (PTSDs), which developers then used as a basis for rewriting the code in Java.
Given the inefficiencies of this approach, the client sought to explore alternative solutions that could accelerate their Mainframe Off-Load (MFOL) efforts, reduce overall costs, and achieve significant Shift Left savings.
They were looking for innovative strategies to streamline the modernization process, improve system maintainability, and enhance their ability to implement new features quickly and efficiently.
Solution.
Coforge presented an innovative solution leveraging Gen AI technology to automate the process. To demonstrate the efficacy of this approach and evaluate the effectiveness and benefits, Coforge implemented GitHub CoPilot as per below approach.
High level approach for Mainframe Modernization Program
Depicted below is the approach taken for the program:
- Reverse Engineering: Analysts review the TPF programs to generate the PTSD documents
- Code generation: Co-Pilot is used to automatically write the Java code based on the prompts
- Manual Review: Developers then review and validate the generated code snippets and manually integrate into the main program.
- AI-Driven Optimization: Co-Pilot is leveraged again to further optimize the code, assist in creating unit test cases, and integrate the result into the DevOps pipeline.
- Deployment Preparation: Following these steps, the target programs are finalized and ready for deployment, testing, and migration.
- Continuous Improvement:
This streamlined process combines human expertise, and Gen AI-driven conversion & optimization to efficiently modernize legacy systems while continuously improving the modernization methodology.
The Impact.
- Code gets generated in Java (Spring Boot) automatically with least manual intervention
- Functional & Integration Test script can be generated automatically
- Leveraging this approach, Code generation and testing effort can be reduced as much as 80 to 90 percent leading to overall savings of ~30%
33%
Reduction in time and effort during the development & testing phase
Zero
Repeated errors / code review defects
>90%
Code coverage from the auto-generated Unit test case
Related reads.
