The case against a microservices-first architecture
For most teams under fifty engineers, a well-structured monolith ships faster and costs less. Here is where we draw the line.
Microservices solve an organisational problem, not a technical one. If you have fewer than a handful of teams shipping independently, you are likely paying coordination costs without getting the benefit.
What a monolith actually gives you
- Database transactions across your domain, without sagas
- One deployment, one log stream, one place to look when something breaks
- Refactoring across boundaries while they are still cheap to move
When to split
Split when a specific part of the system has a genuinely different scaling profile, a different compliance boundary, or a team that needs to deploy several times a day without waiting.
We recently moved a client from eleven services back to a single Laravel application with a queue worker. Infrastructure cost fell 61% and deploy time went from fourteen minutes to ninety seconds.
Related articles
Securing file uploads in a CMS
CVs, artwork and documents are the most commonly exploited part of any content-managed site. A practical checklist.
Read articleLegacy modernisation without a big-bang rewrite
Strangler-fig patterns let you replace a legacy system incrementally, with value delivered every few weeks.
Read articleWhat we look for when hiring engineers
Not leetcode. We care about how you reason about trade-offs, communicate uncertainty and handle being wrong.
Read article