Solution · Example scenario: Incremental Legacy Modernization

A legacy system moves to a new architecture in small, verified steps while it keeps running, where the system allows it.

Definition: What incremental legacy modernization is

Incremental migration

Incremental migration is moving a system to a new architecture or platform in small, verifiable steps while the existing system keeps running.

  • One part moves at a time; the rest of the system stays in service.
  • Incremental legacy modernization applies this method to a legacy system, as part of Legacy Software Modernization.

This page is an example scenario: it describes a typical situation and how Logevon approaches it.

Strangler fig pattern
The strangler fig pattern is an incremental migration approach in which new components gradually take over the functions of a legacy system, usually behind a routing layer, until the old system can be retired. The name comes from a metaphor described by Martin Fowler; the Microsoft Azure Architecture Center describes the façade that routes requests between the two systems.
Parallel run
A parallel run is a period in which the old and the new implementation process the same work side by side so that their outputs can be compared before the new one takes over.
Cutover
Cutover is the point at which live work switches from the old system or component to the new one.

Situation: When the system cannot stop, and cannot stay as it is

Example scenario

The business runs on it every day.

Daily operations pass through the system, so taking it offline for a long migration is not an option.

A single switch is too risky.

Replacing everything at once puts the whole operation behind one cutover.

Nobody can specify the whole system.

How it behaves is known to a few people and written down nowhere.

Change cannot wait for a full replacement.

New requirements keep arriving while the system is being replaced.

Approach: How Logevon would approach it

  1. Map boundaries and dependencies

    In Understand and Assess, we chart the system's parts, the data they share and who holds the knowledge, and check that an incremental route exists.

  2. Choose the first part

    In Design, we pick a part with clear boundaries, few dependencies and low risk, and define how its output will be verified.

  3. Route, move, verify

    In Build & Integrate, a routing layer goes in front of the legacy system; the part moves, runs in parallel and is verified before cutover.

  4. Repeat and hand over

    In Enable & Transfer and Improve, each step is documented, your team takes on more of each one, and the plan is adjusted.

Logevon delivers this work through the Forward Deployed Engineering (FDE) model.

Architecture: What the solution could look like

During an incremental migration the old and new parts run together, connected by temporary components.

What the solution could look like
ComponentRole
Routing layerSends each request to the legacy system or to the migrated part. It is usually removed when the migration ends.
Migrated partTakes over one function on the architecture the system needs.
Legacy systemKeeps serving every function that has not moved yet.
Data synchronizationKeeps data consistent while both sides read or write it.
VerificationCompares old and new outputs, in a parallel run where practical, before cutover.
Rollback pathLets work return to the legacy system until its part is retired.

Where the migrated part exchanges data with other systems, that work is described in Enterprise System Integration.

Handover: What stays with you

A system migrated in parts

Each migrated part in service and verified against the legacy behavior.

A written map of the system

Boundaries, dependencies and data flows, for the migrated and the remaining parts.

A team that can run the next step

Your people, ready to continue the method on the remaining parts.

Ownership

Your code and data, per the agreed terms.

Limits: Limits and prerequisites

What the approach depends on

  • Incremental migration is not always possible. The assessment shows whether it is.
  • Running old and new side by side adds cost and complexity, so the migration needs a planned end.
  • Data synchronization needs the most care: both sides must stay consistent.
  • Uninterrupted operation is a planning goal, not a guarantee.

FAQ: Questions about incremental migration

Which part is moved first?

One the assessment identifies: clear boundaries, few dependencies and low risk, but real work. The first step proves the method and the verification on a small area.

How long do the old and new systems run side by side?

It depends on how many parts there are, how their data depends on each other and how much time your team has. Coexistence has a cost, so the end is planned from the start.

How is the database migrated?

In parts, like the application: the data for one part is copied, kept in sync and checked for consistency before cutover. Some data cannot be split this way; the assessment shows which.

When is a single cutover the better choice?

When an incremental route does not fit. The Microsoft Azure Architecture Center notes that the pattern may not suit a system that is small and simple to replace, whose requests cannot be intercepted, or that must be retired quickly. The decision criteria are on the Legacy Software Modernization page.