If you’ve spent any time in the integration world, you’ll know that legacy migration is a bit like renovating an old villa: charming from the outside, but full of surprises once you start pulling up floorboards.
Adaptiv have been helping organisations untangle Oracle SOA, BizTalk, TIBCO, and other integration “heritage platforms” for years. We’ve learned a few things—sometimes the hard way—about what makes legacy code migration both rewarding and occasionally quite challenging. So, here’s my take on the good, the bad, and the ugly of modernising integration workloads for Azure.
The Good: Modern Platforms, New Possibilities
There are several reasons so many organisations are retiring their ageing middleware, one of which is the cloud opens doors that legacy platforms simply never could.
- Azure provides organisations with new capabilities. Unlock elastic scale, true serverless patterns, modern API gateways, and event‑driven architectures. Logic Apps, Service Bus, APIM, and Functions come together as a cohesive, modern integration backbone.
- There’s often more that can be reused than expected. Despite what’s sometimes assumed, a full rewrite isn’t always necessary. BizTalk maps, schemas, and even some custom assemblies can usually be reviewed and ported with far less effort than people expect.
- AI is finally useful for the messy stuff. Tools like GitHub Copilot and Azure OpenAI are surprisingly good at assessing legacy codebases: extracting orchestrations, summarising massive XSLT/XQRY, documenting undocumented pipelines, and even suggesting migration patterns. I’ve had AI summarise a ten-year-old SOA BPEL workflow in minutes.
The Bad: Finding People Who Speak “Legacy”
Let’s face it: deep expertise in BizTalk 2004, Oracle SOA Suite 10g, or long‑abandoned MQ‑based custom ESBs is getting harder to find—and more expensive when it does turn up.
We’ve had challenges in migration projects not because Azure was hard, but because no one remembered why a particular orchestration forked into three branches in 2011. It is also common for key personnel in organisations to retain system knowledge in their heads. These people are gold mines for understanding an old process, but this institutional knowledge is lost when those people retire or move on.
On our side, Adaptiv specialists bring deep expertise to knowing when and how to reverse‑engineer oddities—pipelines referencing outdated DLLs, schemas with missing comments, or obscure functoids that haven’t been touched in a decade. This isn’t incompetence; it’s the reality of long‑lived platforms accumulating sedimentary layers of quick fixes and emergency patches.
The Ugly: The Knowledge Gap Nobody Wants to Own
The hardest part of legacy migration isn’t code. It’s people, history, and organisational memory.
Most migration projects start with a variation of the same conversation:
“Do you have documentation?”
“Yes, absolutely.”
(Three weeks later) “We found a Visio diagram from 2014. It might be out of date.”
Lost client knowledge leads to incorrect assumptions, missing business rules, and significant rediscovery work. I’ve seen integrations where the production behaviour didn’t match the source code because of silent hotfixes pushed years earlier. This isn’t anyone’s fault—it’s the nature of systems that have quietly “just worked” for too long.
Testing: The Perennial Tug-of-War
Legacy migration amplifies a recurring tension: Who owns testing?
Testing can represent a significant portion of the overall effort in a migration project, often accounting for more than half of the total budget. This is because, during the modernisation of an integration platform, even the smallest behavioural changes can have a big impact.
Clients sometimes expect Adaptiv to “just migrate it,” while we expect clients to validate business functionality they uniquely understand. Neither side is wrong—yet neither can do it alone.
The best migrations succeed when:
- As much regression testing as possible is automated
- Clients provide SMEs who have deep knowledge of the business behaviour
- Everyone accepts that test effort will always be larger than anticipated
If your organisation is planning a migration and hasn’t thought deeply about testing, that’s a red flag.
AI as a Force Multiplier (Not a Silver Bullet)
AI helps—but doesn’t replace expertise. It can:
- Document a legacy solution in minutes (more or less)
- Identify unused flows/orchestrations
- Flag schema incompatibilities
- Summarise WSDL/XSD relationships
- Suggest equivalent patterns in Azure
- Generate C# sample code to replace legacy language functions
But it cannot:
- Understand business intent
- Infer undocumented rules
- Reconstruct institutional memory
- Make architectural trade-offs
Think of AI as the enthusiastic junior who reads everything fast but still needs senior oversight.
What Success Looks Like
When legacy migration goes well, it’s beautiful. Project objectives are reached and often exceeded:
- Lower annual costs
- Event-driven patterns
- Higher availability
- Lower operational overhead
- Automated deployment pipelines
- Out-of-support applications removed
- Better, and more modern security
That transformation is very satisfying. It leaves organisations more agile, more secure, and ready for years of innovation instead of technical debt firefighting.
Adaptiv Migration Examples
Energy Company (Platform Migration): The migration to Azure Integration Services is cited as delivering approximately 40% cost savings, alongside a pay‑as‑you‑go consumption model rather than fixed licensing costs. The migration explicitly included streamlining DevOps pipelines, moving away from low DevOps maturity toward modern CI/CD practices.
Infrastructure Company (Legacy Modernisation): Migrated from 2013 BizTalk to Boomi iPaaS to reduce TCO. Adaptiv reverse-engineered existing logic and integrated Boomi with JD Edwards, Workday, and MS Dynamics 365, improving integration delivery speed.
Water Infrastructure Company (HR System Integration): Enhanced HR processes by implementing Boomi, allowing for better data integration and workflow.
Trade Service Business (Cloud Transformation): Improved work management by leveraging Adaptiv’s expertise in Microsoft Azure.
Data Warehouse Optimisation: Provided cloud data warehouse solutions that improved performance for clients by up to 50%
Closing Thoughts: Leadership in a Transitional Era
Legacy code migration isn’t glamorous. It involves archaeology, diplomacy, modern engineering, and occasionally luck. But in a landscape where integration is the backbone of digital transformation, modernising legacy platforms is foundational to organisational resilience. Azure gives us the tools, Adaptiv gives you the expertise, and together we can turn even the ugliest legacy estate into something clean, scalable, and future-ready.
If you’re facing a legacy integration challenge—BizTalk, Oracle SOA, or something even scarier—just remember: the good, the bad, and the ugly are all manageable with the right approach and the right partners.
