Maybe Stopgap Your Legacy Data Issues And Wait for What’s Next?

September 29, 2025 Alex Woodie

The temptation to move off legacy systems like IBM i and System Z is overwhelming for some companies. First, there are business issues, such as enabling data integration and real-time capabilities, that older systems legitimately struggle with. Plus, the siren call from Kubernetes and microservices is getting louder. But the longer you can hold on to current systems and wait for next-gen AI-powered systems to emerge, the better off you will be.

That’s the advice being proffered by the folks at Adeptia, a 20-year-old Chicago-based technical services firm that has worked on data integration issues with many Fortune 500 companies that are struggling to fulfill business requirements on IBM i and IBM mainframe technology.

According to Adeptia’s chief innovation officer Deepak Singh, many Adeptia customers start out planning to ditch their older systems in favor of building or buying something new that can meet their business needs. But the more they work with Adeptia and its “legacy wrapper” approach to data integration, the longer they end up sticking with the older systems.

“What we’re finding is that, in order to do legacy modernization, there are different paths,” Singh said. “The ultimate is to just completely re-platform, go into a new application. But that’s very difficult and challenging. So they tend to put a wrapper around it and fix the problems that are there, whether it’s a process that they need to improve or a data problem. They will use a product like ours or some other to provide more flexible data options.”

Deepak Singh is the chief innovation officer at Adeptia.

This is exactly what Adeptia provided for a large financial services firm that helps manage 401(k) plans for millions of Americans. The company struggled to adequately ingest the large number of documents that contained the retirement plan details sent to it by clients every two weeks. Some of its clients sent the employee records as PDFs, while others sent them as Excel spreadsheets, and still others used EDI.

“Their challenge was their systems are still mainframes, which have been running for 30 to 40 years,” Singh told IT Jungle. “These are systems that they haven’t touched since the 1990s.”

Instead of ditching the old mainframe app, which would have been expensive, painful, and risky, the company ended up offloading the first stage of data processing from the application onto Adeptia. By handling the front-end data ingest and validation in its product, Adeptia was able to provide the connectivity that clients needed without opening up the mainframe, Singh said.

The engagement stuck in Singh’s mind in part because of one piece of advice the customer’s chief architect provided about the nature of legacy problems concerning technology, data, and training.

“They said if errors happen consistently, it’s technology,” Singh said. “If errors happen sporadically, it’s the data. If they happen only during certain workflows, it’s the process that is wrong. And if they only happen when specific people are involved, it’s training.”

That led to an additional insight: Most of the problems with legacy systems are actually data problems in disguise. “They figured that if they fix the data layer, then suddenly the system that they felt was outdated was actually working fine,” Singh said.

Another common issue that Adeptia runs into is real-time visibility into legacy systems. One customer, a large pharmaceutical company, was expanding its ecommerce operation, but struggled to get order confirmation out of the legacy system. Its customers wanted order acknowledgement within 30 seconds, which the legacy system was not designed to handle.

“You need to look in the inventory, but the older system, which was just an order processing system, doesn’t handle that,” Singh said. “They needed to modernize, and they were like, oh my God, it’s going to cost us $100 million to upgrade all of our systems to be able to do this! Otherwise we’ll start losing orders.”

What comes after the mainframe?

Ultimately, the customer signed up with Adeptia, which used its data integration platform to devise a new process. Adeptia’s software now sits in front of the inventory management system and is able to determine product availability and immediately send back acknowledgements to their customers when they place the order, Singh said.

“There were like a half a dozen customers that had this the same issue, where they needed to modernize their older systems to be more real time,” he added. “The choice was either to upgrade the whole legacy system or kind of modernize it with putting a wrapper around it. That’s the approach that we are seeing a lot of companies take.”

Adeptia’s wrapper-based approach to legacy modernization jibes with the findings of a recent Kyndryl report, which found IBM i and mainframe shops are backing away from big-bang platform replacements and instead taking a more tactical approach to application modernization. Wholesale migrations off the mainframe (i.e. IBM i, System Z, and Fujitsu big iron) remain exceedingly rare, with a 0.2 percent rate in Kyndryl’s 500-person survey.

Hybrid modernization approaches, where some applications or process are moved off the mainframe (often to the cloud), is becoming quite popular, Kyndryl found. The former IBM Global Services unit said it is witnessing in mainframe modernization projects “a degree of agility not commonly associated with technology projects of such significant size, scope, and complexity.” What’s more, modernization project costs are going down, while the return on investment (ROI) is going up, it found.

Adeptia’s engagements are so successful that clients often put legacy migration plans on hold indefinitely, Singh said. As the AI revolution winds its way through the business community, there’s another factor at play: Why embark upon an expensive, time-consuming, and risky effort to replace monolithic ERP system with a modern microservices-based cloud solution running on Kubernetes when AI could potentially make those newer systems obsolete?

“I checked with the 401(k) provider four months ago, and they have pushed it off indefinitely,” Singh said. “Now their focus is all on AI. They’re thinking now the AI systems are going to come into play, so they don’t want to replace the mainframes with something that is a bridge to whatever the future platforms would look like, which are more agentic.”

Companies that would have otherwise started executing their plan to migrate legacy IBM i and mainframe systems to a modern architecture may just want to see how the whole AI thing plays out. If there’s the potential to leapfrog a generation and go straight to agentic-based application architectures while saving hundreds of millions of dollars and minimizing risk, then why not take it?

This Issue Sponsored By

Table of Contents

Content archive

Recent Posts

Subscribe

Pages

Search