Almost every organization has an application everyone dislikes and no one dares touch. It runs on an old server, depends on one person’s knowledge, and sits under a process the business cannot pause. “Modernize everything” is not a plan. A decision for each application is.
This guide helps an IT director, CIO or operations leader choose among six options, in a sensible order, without betting the organization on a rewrite.
The Six Options
| Option | What it means | Fits when |
|---|---|---|
| Retain | Leave it as is for now, deliberately | Low risk, near end of life, or blocked by a dependency |
| Rehost | Move it to new infrastructure with minimal change | The app is stable but the hardware or data center is the problem |
| Replatform | Move with targeted changes (managed database, newer OS or runtime) | Small changes remove real operating burden |
| Refactor | Re-architect or rewrite parts of the code | The app is strategic and its design limits the business |
| Replace | Switch to a commercial or SaaS product | A standard product now does the job and your customizations add little |
| Retire | Switch it off after migrating or archiving what matters | Few users, duplicate function, or no remaining business value |
AWS’s portfolio assessment guidance frames this as a balance of business drivers, prioritization criteria and strategy. It advises reserving the most complex approaches, such as refactoring, for strategic workloads, and applying the decision to application components rather than whole applications. A three-tier app may rehost its servers while its database is replatformed.
Signals That Point to Each Choice
Collect facts before opinions. For each application, record:
- Business criticality and users: who depends on it, and what stops if it fails.
- Age and support status: operating system, runtime, database and vendor support dates.
- Security posture: known weaknesses, authentication, patching, logging and the data it holds.
- Dependencies: other systems, integrations, file shares, printers, licenses and hardware.
- Knowledge risk: whether documentation exists and who can actually change it.
- Cost and pain: support hours, outages and workarounds.
- Compliance exposure: contract or regulatory obligations attached to the data.
An application with high business value, a sound design and old infrastructure is a rehost or replatform candidate. High value with a design that blocks change is a refactor candidate. Low value with high operating cost is a retire or replace candidate. Do not refactor something simply because it is old.
The Risks Teams Underestimate
- Hidden dependencies. Microsoft’s migration planning guidance stresses discovering dependencies first and grouping workloads that share databases, APIs, authentication or network connections so they move together.
- Split operation. When some components move and others stay, the connection between environments becomes a risk. Keep that period short and documented.
- Data and integrations. Replacement projects often stall on data conversion and the reports no one remembered.
- No rollback. Define what counts as a failed cutover and test the way back before the real move.
- Compliance drift. Moving data can change who can access it and where it is stored.
Sequence the Work
Both AWS and Microsoft guidance recommend starting with simpler, lower-risk workloads, moving non-production before production, and adding one or two representative complex workloads to early waves to expose problems. A practical order:
- Inventory and grade. Build the application list and score each against the signals above. A read-only application health check can supply evidence on a single critical application, including security, data and compliance findings.
- Retire first. Switching off unused systems reduces cost and attack surface immediately.
- Stabilize. Fix backups, access control and monitoring on what you keep.
- Pilot a simple move. Prove your process, rollback and communication on a low-risk app.
- Move the critical systems after you have shown you can do it, avoiding financial close, peak seasons and similar periods.
- Refactor or replace the strategic few with funding and a named business owner.
Use App Delivery as a Bridge
Some applications cannot be rewritten soon but still need to leave a failing server room or stop living on unmanaged laptops. Virtual desktops and published apps can bridge that gap. Microsoft describes Azure Virtual Desktop as a service that can deliver full desktops or individual applications, including line-of-business apps in formats such as Win32, MSIX and Appx, from any device. Users connect through reverse connections, so no inbound ports need to be opened.
The benefit is practical: the application runs in a controlled environment, the data stays off the endpoint, and you buy time for a proper replacement. The caution is also practical: a bridge is not a destination. Set a review date, keep the application on the modernization list and confirm that licensing, performance and any special hardware or GPU needs work in the hosted environment.
Z7’s virtual desktops and apps service covers Azure Virtual Desktop and Omnissa Horizon, designed, deployed and managed end to end.
Make the Decision Record
For each application, write one page: the chosen option, the reasons, the owner, the dependencies, the cutover and rollback plan, the success measure and the review date. When the plan changes, update the record instead of the memory of the room.
Where Z7 fits
Z7’s IT solutions team designs, migrates and runs cloud, private cloud and on-premises infrastructure, and can help grade your portfolio and sequence the moves. To start with one application, book a conversation.