Managed.IT - issue 65 29 OPINION Petra Goude non-existent. That’s a problem when the time comes to modernise IT systems to enable new functions and work with cloudenabled platforms. n Identifying dependencies. Sorting out the tangled control structure of COBOL systems can pose serious challenges and cause costly delays. GenAI can help developers ‘decompose’ or reverse engineer legacy systems to identify dependencies between different modules and codebases and determine approaches to modernisation more accurately and easily. n Recommending actionable solutions. Generative AI can provide recommendations for updating and improving code quality and architecture. By applying GenAI algorithms to legacy code, programmers can save time and control costs while maintaining, refactoring and re-engineering COBOL-based systems. n Using AI to unlock more of the mainframe’s potential. By combining human expertise with AI technologies, organisations will accelerate their mainframe modernisation initiatives and make fuller use of their data. COBOL expertise will be the key to understanding a firm’s business logic, while AI will accelerate coding tasks with fewer errors. To maintain business continuity and protect their entrenched IT investments, governments and critical-infrastructure industries must find new ways to get the most out of their mainframes. Lacking their own skilled personnel, these organisations will need trusted partners who bring both technical and business expertise to their projects. COBOL skills are becoming extinct. That’s bad news for governments, financial services, travel & transportation, telecommunications and other organisations that run their core IT infrastructures on the mainframe. The mainframe’s bulletproof reliability and security runs on billions of lines of decades-old COBOL code (or other legacy code such as PL/I and RPG for IBM i systems). And many of the people who can write, repair and update that code are approaching retirement. The foundational relationship of COBOL and other legacy programming languages to newer languages such as Java or C++ is similar to the relationship of Latin to modern Roman languages. The newer languages have their roots in the legacy language. But just because you can speak French doesn’t mean you can decipher the classical texts of Ancient Rome. There needs to be translation, both in text and context. Here are five ways Generative AI can accelerate this process: n Translating COBOL into more modern programming languages. A new generation of COBOL experts can use Generative AI to translate COBOL code into languages like Java or C++. In the right hands, GenAI can help organisations to automate the translation process – bringing more consistency and reliability to this otherwise lengthy, complex and labour-intensive endeavour. n Generating technical documentation. Experts can also direct GenAI to create technical documentation for legacy COBOL programs. This is critical to mainframe operations, as documentation of essential older code is often scarce or 5 ways Generative AI can address the COBOL skills crisis By Petra Goude, Kyndryl Global Practice Leader for Core Enterprise & zCloud Detailed planning Robert’s confidence in Node4’s capabilities was vindicated by the execution and planning they put into the data centre migration, which required servers to be moved from the train operator’s HQ to Node4’s data centre in a small time-window between 1am, when the last West Midland Trains service enters the depot, and 4am, when the first train of the day departs. In the days before the move, Node4 got the network infrastructure and perimeter firewall up and running and tested the failover. They also worked behind the scenes to keep the two internet providers on track, ensuring links between the primary and secondary data centres were installed as scheduled. “On the actual night, we shut down non-critical servers at 11 pm and were given the approval from rail control to shut down the critical servers at midnight. All the servers were wrapped and transported from the Birmingham location to the Node 4 data centre where they were unloaded and racked with connected power cables by 2 am. Our business-critical servers were back online at 4 am and non-critical ones by 6 am. I don’t think I’ll ever forget the pressure we were under that night, but Node4’s team were professional, calm and exacting in their approach at each stage. As a result, everything went according to plan.” Roberts adds that Node4’s approach on the night and in the support it has provided since are typical of that can-do attitude which had so impressed him and his team on that initial site visit. “Instead of putting up barriers, causing delays and becoming a bottleneck, Node4 finds solutions, solves problems and generally makes my life so much easier. It’s an absolute pleasure to work together,” he said.