Book a call

Choose who to talk to

Pick who you’d like to speak with and book a time that works for you.

Dima Blyshchyk, CEO & Founder at Mitrix

Dima Blyshchyk

CEO & Founder

Max Fedaraka, Head of Business Development at Mitrix

Max Fedaraka

Head of Business Development

By Dmitri Koteshov · Mitrix Technology
HealthcareSoftware development

Legacy healthcare platform modernization on .NET and Azure

Mitrix's engineering team explains how we modernize legacy healthcare platforms on .NET and Azure without a big-bang rewrite: boundaries, parallel run, cutover and rollback.

A .NET Framework app with a WinForms client can run a DME network for a decade because it encodes the business. Replacing it is less about code and more about preserving the workflow while changing the plumbing.

Healthcare legacy systems carry clinical rules, payer quirks, integrations and years of undocumented decisions. A rewrite that ignores that history produces a modern system that fails in the first week of real operations.

Why legacy persists

The business logic is the product. Custom rules for eligibility, pricing, suppliers, authorizations and exceptions live in code that nobody wants to touch.

The integrations are fragile. Flat files, HL7 interfaces, stored procedures and scheduled jobs connect systems that still work. Each one is a reason to postpone a rewrite.

The workflow is validated. Clinicians and operations teams have adapted to the system. Changing the interface can change the process, and the process is often the reason the business runs.

Downtime is expensive. A DME network or a claims pipeline cannot stop for a migration weekend and hope for the best.

Documentation is thin. The people who know why a rule exists may have left. The code is the specification.

Modernization options

Choose a path by risk and business value:

  • Rehost. Move to newer infrastructure with minimal code change. Fast, low risk, limited benefit.
  • Replatform. Replace the runtime, database or hosting while keeping the application structure. Useful for .NET Framework to .NET 8 or SQL Server to Azure SQL.
  • Refactor. Improve the architecture inside a boundary: API layer, background jobs, data access, observability.
  • Rewrite. Rebuild a module or the whole product. Highest cost and risk, sometimes the only path.
  • Strangle. Put a facade in front of the legacy system and move capabilities out one at a time. Usually the safest route for a live healthcare platform.

Constraints

Healthcare modernization has constraints that a generic migration does not:

  • PHI boundaries, access control and audit must survive every cutover.
  • Interfaces must continue to work during and after the migration.
  • Clinical and financial data must reconcile after they move.
  • Downtime needs a clinical fallback as well as a technical rollback.
  • Every change needs validation with the people who run the workflow.
  • Regulated records may need retention and legal hold support.

A concrete legacy example

A DME platform runs on a WinForms client, a .NET Framework API and a SQL Server database. Orders, suppliers, deliveries and billing all work. The business wants a browser client, better reporting and Azure hosting, but cannot stop operations for a rewrite.

The safe path is to put an API boundary in front of the existing database, build the new order intake screen against that API, and run it for one order type and one supplier group. The legacy client stays available. Once the new path reconciles with the old one, the next capability moves. The database is migrated later, behind the API, when the application no longer depends on its schema.

Data migration

Data migration is a project of its own. It includes identity mapping, deduplication, document migration, history preservation and reconciliation. Decide which records are active, which are historical and which are legally required to remain immutable. Test the migration with a full business cycle, not a sample of ten rows.

Keep a rollback copy and a documented way to compare old and new. If a claim or a delivery cannot be traced back, the migration is not complete.

When not to modernize

Sometimes the right decision is to stabilise, not replace. If the system is stable, the business is not changing, the integrations work and the team can still support it, a costly rewrite may destroy more value than it creates. Put observability, security patching and documentation around the legacy system instead.

Modernize when the platform blocks a business change, the risk of staying is measurable, or the cost of maintenance is rising faster than the cost of migration. “The technology is old” is not a business case on its own.

A safe sequence

  1. Inventory and dependency map. Systems, interfaces, jobs, databases, reports and owners. Include the undocumented ones.
  2. Define boundaries and contracts. Choose a module with clear inputs and outputs. Put an API or event boundary around it.
  3. Extract a thin slice. Move one capability at a time. Keep the legacy path available behind a feature flag.
  4. Parallel run. Process real data through both paths and reconcile. Compare counts and key fields, including the UI.
  5. Cut over. Freeze changes, agree on the final delta, run the rollback plan and keep the old path warm.
  6. Decommission. Remove the legacy path only after the new one has run clean through a full business cycle.

Team and operating model

Modernization needs a product owner who understands the clinical and financial workflow, an architect who can define boundaries, engineers who can work in the legacy stack, and QA who can test both paths. It also needs an operations representative who can approve a cutover and a support owner for the first month.

Do not hand the legacy system to a team that has never run it. The people who know the exceptions are part of the migration, even if they are not writing code.

Architecture on .NET and Azure

A typical path is .NET Framework to .NET 8, a thin ASP.NET Core API in front of the legacy database, Azure Functions for background jobs, Service Bus or RabbitMQ for queues, Azure SQL for the operational store and Blob Storage for documents. Infrastructure as code moves from Bicep or scripts to Terraform, with CI/CD and environment separation.

The integration layer should stay explicit. Do not let the new UI talk directly to the legacy database. Put a contract in between, version it and monitor it. That contract is what lets you replace the legacy module later without another rewrite.

Testing and cutover

Regression tests should cover the clinical and financial paths, including the new features. Data reconciliation should compare source and target at the record and field level. Performance tests should include the batch jobs and interfaces, because those often fail first under a new runtime.

Run a failover exercise and a rollback exercise. An unexecuted rollback plan is a document. Keep the legacy system readable for an agreed period after cutover, and give operations a clear path to report discrepancies.

How Mitrix fits

Mitrix works with .NET Framework estates and .NET 8, Azure Functions, Azure SQL and Terraform in healthcare delivery, and builds API layers around existing EMR and ERP systems. Our DME network case included a WinForms application, C#, Angular, Azure Functions, Azure SQL, Service Bus and Blob Storage.

If you are planning a rewrite, start with a boundary and a parallel run, not a big-bang cutover. Read EMR integration or healthcare QA, or talk to us about the first slice.



You might also like

FAQ

Frequently asked questions

Should we rewrite or refactor a legacy healthcare platform?

Refactor or strangle first when the platform is live and the business rules are not documented. A full rewrite is justified when the architecture blocks the care model and the organization can support a multi-year program.

How do we modernize without downtime?

Use a strangler path, feature flags and a parallel run. Keep the legacy system available, move one capability at a time, reconcile data and test rollback before each cutover.

How long does a .NET healthcare migration take?

A focused module can move in months. A full platform with interfaces, data migration and clinical validation is a multi-quarter program. The boundary and reconciliation work usually takes longer than the code migration.

What about PHI during migration?

Keep the same access, audit and retention controls in the new path. Avoid copying PHI into lower environments, and validate that the new system logs access and changes at least as well as the old one.

Can you migrate the database separately from the application?

Sometimes. It is safer when the schema is stable and the application can tolerate the new platform. Otherwise, put an API boundary in front of the data and migrate the application and data in smaller steps.