Every mid-size organisation I work with has at least one piece of middleware that’s been running quietly in the background for years. An integration platform from 2015. A message broker that predates the current CTO. An ESB that nobody fully understands but everything depends on.

It works. Sort of. It moves data between systems, handles transformations, and keeps processes running. Nobody wants to touch it because the last person who understood it deeply left two years ago. So it stays. Year after year, it stays.

And it’s costing you far more than the licence fee.

The Licence Fee Is the Smallest Cost

Let’s start with the obvious number. Legacy middleware licences are often surprisingly cheap — either because the vendor hasn’t updated pricing in years or because the organisation locked in a favourable renewal rate. I’ve seen annual licence costs of $20,000 to $80,000 for middleware platforms that support critical business processes.

That number feels manageable. It’s a rounding error in a $2 million IT budget. But it’s also misleading, because it represents maybe 15 to 20 percent of the actual cost of running that middleware. The rest is hidden.

The Real Costs

Specialist maintenance labour. Legacy middleware needs care. Patches, configuration changes, performance tuning, certificate renewals, connector updates when downstream systems change. These tasks require knowledge of the specific platform — knowledge that’s increasingly hard to find.

SEEK job listings for specialists in older middleware platforms like TIBCO BusinessWorks, IBM Integration Bus, or Oracle SOA Suite have shrunk dramatically over the past three years. The talent pool is contracting as specialists either retire or move to modern platforms. The ones who remain charge premium contract rates — $1,500 to $2,500 per day — and availability is limited.

If your middleware depends on a single internal expert, your exposure is even worse. That person is a single point of failure for your integration layer. When they leave — and they will — you’re paying emergency contractor rates to replace institutional knowledge.

Opportunity cost of constrained integration. This is the cost nobody puts in a spreadsheet, but it’s often the largest. Legacy middleware platforms are rigid. Adding a new integration — connecting a new SaaS tool, building a new data pipeline, supporting a new business process — takes weeks or months on legacy middleware that would take days on a modern platform.

Every time a business unit waits for an integration to be built, that’s delayed value. Every time IT says “we can’t do that easily because of our integration layer,” that’s a business opportunity missed. Over the course of a year, these delays compound into significant competitive disadvantage.

Security exposure. Older middleware platforms receive infrequent security updates, if any. Some are running on operating systems that are themselves end-of-life. The Australian Signals Directorate’s Essential Eight framework specifically calls out patching applications and operating systems as foundational controls. Legacy middleware that can’t be patched to current standards is a compliance gap and a genuine security risk.

I’ve seen middleware platforms running on Windows Server 2012 R2 because upgrading the OS would break the middleware. That’s not an acceptable risk profile in 2026, but it’s more common than anyone wants to admit.

Testing overhead. Every change to any system that touches the middleware requires regression testing of the middleware integrations. This creates a testing burden that slows down development across the entire organisation. Teams can’t deploy independently because everything flows through a shared integration layer that might break if upstream or downstream systems change.

Modern integration approaches — API-based, event-driven, loosely coupled — allow independent deployment and testing. Legacy middleware creates tight coupling that makes every change risky and expensive to validate.

Why Organisations Don’t Migrate

If the costs are this significant, why do organisations keep running legacy middleware? Three reasons, in my experience.

Fear of disruption. Migration means touching every integration. If the middleware handles fifty integrations, that’s fifty things that could break during transition. The perceived risk of migration often exceeds the perceived risk of staying put — even when the math doesn’t support that assessment.

Sunk cost thinking. “We’ve invested so much in this platform already.” This is classic sunk cost fallacy. The money spent on the legacy platform is gone regardless of what you do next. The question is whether continuing to invest in it produces better returns than investing in a replacement.

Lack of visibility into true costs. When the costs are hidden across different budget lines — contractor invoices, delayed projects, security remediation, productivity losses — it’s hard to build a complete picture. Nobody owns the total cost of legacy middleware, so nobody sees it.

The Migration Conversation

I’m not suggesting you rip out your middleware tomorrow. But I am suggesting you calculate the real cost — all of it — and use that to inform a deliberate decision about whether and when to migrate.

Start by documenting every integration the middleware handles. Map the dependencies. Identify which integrations are critical and which are candidates for simplification or elimination. Some integrations exist because they always have, not because they’re still needed.

Then assess modern alternatives. MuleSoft, Boomi, Workato, and similar platforms offer migration paths from legacy middleware with significantly lower ongoing operational costs. Cloud-native integration services from AWS, Azure, and GCP are even leaner for organisations with smaller integration footprints.

The migration won’t be painless. But the compounding costs of inaction are real and growing. Every year you delay, the talent pool shrinks further, the security exposure widens, and the gap between what your integration layer can do and what the business needs it to do gets larger.

That quiet, reliable middleware that nobody wants to think about? It’s time to think about it.