Offers
Monolith modernisation
Relieve existing systems step by step, without stopping operations or rewriting everything.
“Every change to the legacy system costs weeks and nobody dares to touch it.”
- Format
- Assessment in 2 weeks, delivery in iterations
- Who it is for
- Organisations with a grown system and overdue modernisation
How you notice it
- Small domain changes take weeks because nobody knows the side effects.
- There are areas of the code nobody dares to touch any more.
- Tests are missing or so slow that people route around them.
- The people with the decisive knowledge left the company long ago.
- A rewrite has been attempted once — and abandoned.
Outcome
- Analysis of coupling, domain boundaries and risk hotspots
- Migration path with strangler steps instead of a big bang
- Test coverage and refactoring of the critical paths
- Handover to the internal team with documented decisions
How we work
01
Weeks 1–2 — assessment
We read code, deployment and data, talk to the team and cut the system along the domain. The result is a map of coupling, risk hotspots and worthwhile first slices.
02
Migration path
A strangler plan: which part is extracted first, how long old and new run in parallel and how we know a step succeeded.
03
First slice in production
We implement the first part — including tests, data migration and a fallback path. Operations keep running throughout.
04
Iterate and hand over
Further slices in the same pattern, increasingly led by your team. Decisions are captured as ADRs so they stay traceable later.
Frequently asked
- Do we have to move to microservices?
- No. A well-cut monolith is often the better answer. We only extract what needs its own change rate, its own load profile or its own ownership.
- What if nobody understands the legacy system any more?
- That is the normal case. We work from behaviour: logs, data flows, characterisation tests. The system itself is the most reliable documentation.
- Can we ship features in the meantime?
- Yes, and that is intentional. Strangler-style modernisation runs alongside daily business. Freezing feature work is almost never sustainable.
- How long does the whole thing take?
- The assessment takes two weeks and gives a reliable range. After that we work in four to six week iterations, each with a visible result.
- What if the assessment argues against modernisation?
- Then we say so. Some systems run stably enough that a rebuild only creates cost. That outcome is a good outcome too.
When this does not fit
If the legacy system is being replaced soon anyway, or the domain is currently being rethought from scratch, the rebuild is not worth it. An event modeling sprint is the more honest start then.
Often combined with
Background in the article Modernising a monolith without rebuilding everything.