Abstract image of neural networks connecting together

AI in mainframe modernization

October 15, 2024

AI in mainframe modernization

How generative AI provides a way out of legacy application constraint and stagnation on the mainframe.

Mainframe systems are at the heart of our economy - they manage our bank accounts, our insurance policies, put food on the shelves of our supermarkets and keep our essential infrastructure running. They are reliable, secure and scalable, and have been serving us for decades. According to a <a href="https://www.ibm.com/topics/mainframe" target="_blank">recent IBM report</a>, 45 of the top 50 banks, 4 of the top 5 airlines, 7 of the top global retailers, and 67 of the Fortune 100 companies rely on mainframes as their core platform. Mainframes have a proven capability to manage scalability, security and resilience, making them a good fit for transaction processing, financial systems and record keeping. However, legacy mainframe applications were not built to meet the needs of modern business. Most mainframe applications are built on aging technology (such as COBOL) that does not meet modern business needs for digital innovation, business agility and cost efficiency.

Key issues

Manual, inflexible business processes
System complexity
Declining pools of expertise
Lack of documentation
Niche technologies
Inefficient technology operations
Lack of integrated management visibility
Stranded data assets
Release management integration
Vendor lock-in
Difficulty accessing from digital channels
Maintaining regulatory compliance

An IBM study showed that mainframes handle almost 70% of the world’s production IT workloads.

<h4>The key challenges in mainframe modernization</h4> Ideally, we could modernize mainframe applications so that they accelerate, rather than hold back, business change. But mainframes are complex systems that have been built up progressively over many years. Modernization teams have struggled to fully understand the current state of applications. Due to gaps in understanding, teams were unable to effectively manage change impacts and transition risks. These factors resulted in high modernization costs, or sub-optimal modernizations that re-create legacy on a new tech stack. Now, generative AI is removing the barriers. <h4>How generative AI can transform the mainframe modernization process</h4> Thanks to revolutionary new capabilities, generative AI creates completely new possibilities for mainframe modernization and puts a modern core within reach. Gen AI techniques are highly applicable across the entire modernization journey.

figure explaining modernization scenario with gen AI

Figure 1

<h5> 1.    Optimizing on the mainframe</h5> Rapid results can be realized using gen AI to address productivity, risk, and complexity within the mainframe environment. Focusing on these opportunities in the short term can also free up capacity to re-invest into more transformational modernization activities. Gen AI tools now make possible: <ul> <li>Legacy understanding. Existing applications can be reverse engineered to fill documentation gaps, interactively explain specific code, and extract business rules. </li> <li>Language translation. COBOL can be translated to modern languages such as Java. Unlike previous translation techniques, generative AI creates flexible “modern” Java and not just COBOL code re-written in Java syntax. </li> <li>Modern engineering. New requirements can be implemented in Java using modern DevSecOps practices. Within this context, gen AI can accelerate new development, increasing productivity.</li> </ul> <h5>2.    Modernizing to cloud</h5> Gen AI can also de-risk and accelerate the migration of mainframe workloads onto cloud platforms. Depending on technical complexity and business value, three major modernization approaches are applicable. Gen AI tools now make possible: <ul> <li>Rebuild. Many applications can be migrated to Commercial, Off-The-Shelf (COTS) solutions. In this scenario, gen AI can understand existing business logic and generate specifications for parity in new implementations. </li> <li>Refactor. Legacy code can be re-factored to modern languages with gen AI, and then additionally re-factored to run effectively on the cloud (e.g. containerization). </li> <li>Rehost. Stable applications that need to migrate to either decommission hardware, or co-exist with other apps, can run on virtual mainframe environments. AI can assist in identifying dependencies and integration points for clean scoping.</li> </ul> <h4>Practical steps to begin your AI-powered mainframe modernization journey</h4> <ol> <li>Optimize for business benefit To maximize the value of modernization activity, different strategies should be adopted for different applications. Infrequently changing applications that are not constraining the business will benefit most from automation and platform optimization and will likely not justify more invasive modernization. Applications that are functionally fit for purpose but are creating significant business constraint should be a focus for technical refactoring. Applications that offer insufficient functionality or ineffective experiences should be a focus for business-led re-implementation. </li> <li>Sequence for self-funding In capital-constrained environments, there are opportunities to sequence modernization to subsidize development activities through savings in operations costs. Operations automation, infrastructure right-sizing and developer productivity acceleration are all high-priority early-stage activities that create headroom to re-invest in modernization. For this reason, having a partner that handles operations, modernization and new development within a single delivery organization enables much more efficient re-balancing of focus as efficiencies increase. </li> <li>Accelerate with AI and Automation</li> </ol>

All stages of modernization can be enhanced by adopting a range of gen AI, AI and automation tooling, including: <ul> <li>Analysis and understanding of legacy environments</li> <li>Translating and re-factoring legacy code into modern code</li> <li>Acceleration of new code development</li> <li>Automation of DevSecOps and production operations  </li> </ul>

Figure explaining how to determine modernization target with app

Figure 2

Case study

Building a best-in-class scalable order management platform for a North American retailer

The client had a large footprint of legacy mainframe applications that was misaligned to their vision, constraining change, and undermining customer experience. Cognizant re-engineered the order management system within Google Cloud Platform using a modern architecture and consolidating across lines of business.

Modernization of 5M lines of code was successful and now processes over $10B of annual digital sales. The modernized solution enables new marketplace and partnership capabilities. The new cloud platform greatly enhances business agility with auto-scaling, self-heal and zero-trust security capabilities.

<h4>What’s next for mainframe modernization?</h4> As more organizations embrace digital transformation, the role of generative AI in mainframe modernization will continue to grow. AI-driven automation offers a way to bridge the gap between legacy and modern systems, allowing <a href="https://www.cognizant.com/en\_us/services/documents/cognizant-ai-first-for-mainframe-modernization.pdf" target="_blank" rel="noopener noreferrer">businesses to unlock new potential</a> while maintaining the core strengths of their mainframes.