By Naresh Duddu
In the world’s top companies across industries, mainframes continue to power core operations: recent research says that 61 percent of global infrastructure hardware decision-makers use mainframes, and 54 percent plan to ramp up usage over the next two years. However, at some time organizations using decades-old mainframes will need to modernize them to derive the myriad benefits of cloud and AI, such as scalability, agility, resilience, and innovation.
Understanding mainframe modernization
Mainframe modernization involves migrating legacy systems to a more advanced technology ecosystem, such as cloud native architectures. While modernization approaches can range from a focused upgrade of a specific area to complete system overhaul, the former is preferred by most organizations because of its lower risk and cost. Similarly, considerations of business objective, resource availability, and risk influence the choice of modernization method, the popular ones being Reengineering – reverse engineering and greenfield development; Refactoring – improving existing code or converting it to a modern language automatically; Rehosting – moving applications to new environments without changing their code; and Replacing – switching to a Commercial Off-The-Shelf (COTS) product.
Dealing with challenges
Since mainframes typically host critical workloads and data, modernization is a risky, complex exercise that becomes even more challenging when there are older technologies dating back to the 1970s and 1980s. Worse, piecemeal upgrades to legacy systems over the years can create an awkward blend of new and old technologies, increasing the risk and difficulty of modernization.
One of the biggest barriers to mainframe modernization is the lack of expertise and documentation. Before starting, the current generation of workers who have never been exposed to mainframe technology need to be trained to understand and modernize the system, which also entails significant cost and risk of disruption.
Similarly, system downtime – necessary during upgrade – also costs and disrupts, and unless managed carefully, can even negatively impact customer experience. Also, organizations, especially those in sectors involving sensitive data and regulation – financial services or healthcare, for example – should guard against the elevated risk of cyberattack and regulatory non-compliance that modernization can bring.
Generative AI can enhance mainframe modernization by automating key activities
In the past, a number of technology tools facilitated mainframe modernization by automating code refactoring, core generation, and other tasks. But the arrival of generative AI has taken automation to a new level.
Generative AI can analyze legacy code to extract business rules and provide code / business rules documentation of high quality, required for converting classic legacy code to a modern language, such as Java or C#. This accelerates development cycles, facilitates integration with hybrid cloud applications, and allows organizations to manage mainframe applications more efficiently.
Advanced gen AI tools can even present their code analysis in English or any other natural language and accelerate the generation of the corresponding modernized code, based on the business rules they extracted.
By breaking down complex, monolithic code into smaller and simpler modules, and creating in-line documentation for existing code, generative AI can improve its quality and maintainability.
Strategies for maximizing the impact of gen AI in mainframe modernization
Organizations can take different approaches to generative AI-enabled mainframe modernization, such as:
Prompt Engineering: Providing carefully worded prompts can improve accuracy of outputs and help achieve specific outcomes like segregating mainframe code, extracting and classifying business rules, generating user stories, test cases and flow charts.
Retrieval Augmented Generation (RAG): RAG technique can be leveraged to build knowledge repository on mainframe application documentation, provide more context to AI when extracting knowledge from code, retrieve relevant knowledge content based on user queries, and identify duplicate functionality built in the application.
AI-First Approach: For the best results, organizations should work towards adopting an AI-first approach, integrating gen AI at the core of the entire modernization process rather than using it for isolated tasks.
AI is the future of mainframe modernization
Artificial intelligence technology, particularly generative AI, is transforming mainframe modernization by automating various processes, relieving organizations of significant complexity and effort. By leveraging AI-driven automation, organizations can seamlessly transition their inflexible and cost-inefficient legacy applications to a modern cloud environment to secure a host of benefits, such as scalability, agility, efficiency, resilience, and innovation capability.
AUTHOR:
Naresh Duddu, Associate Vice President and Global Head of Modernization, Infosys
Naresh Duddu is the associate vice president and global head of the modernization practice at Infosys. The practice incubates new offerings, builds services and consulting IP and spearheads application transformation engagements across verticals and service lines. The practice encompasses the four key pillars of application modernization – Open Source, Agile/DevOps, Legacy modernization, and Cloud, and the views expressed in this article are his own
