core Estimated learning time: 4 h

10.3 Monoliths, microservices and the honest trade

You can say which one a given team should build, and it will not always be the fashionable answer.

Before:08. Observability & Reliability

Most successful systems started as monoliths, and this topic tells the trade honestly: microservices buy independent deployment and independent failure at a real operational tax, and the right choice depends on the team more than the technology. Service boundaries drawn on data ownership — not the org chart — is the design rule that survives contact. The warning at the centre is the distributed monolith: services that must deploy together, sharing a database, combining the costs of both shapes with the benefits of neither. Serverless closes the topic as a third point on the same spectrum, not an escape from it.

Work through these

  • What a monolith actually is, and why most successful systems started as one

    A monolith is one deployable unit containing the whole application, and most successful systems began as one because it is the fastest thing to build and reason about. The word is used as an insult more often than it is used accurately.

  • Microservices: independent deployment, independent failure, and the operational tax

    Splitting into independently deployable services buys independent release and independent failure, and charges an operational tax in networking, monitoring and debugging. The tax is paid whether or not the benefits arrive.

  • Service boundaries drawn on data ownership rather than on org chart

    The useful place to draw a boundary is around data one service owns, not around a team on an organisation chart. Boundaries drawn the second way tend to need constant renegotiation.

  • The distributed monolith — the worst of both, and how you end up there

    Services that must be deployed together, in order, have all the costs of distribution and none of the independence. Recognising this state is the point, because it is arrived at gradually and rarely noticed.

  • Serverless as a third point on the same spectrum

    Functions are a third position on the same spectrum, with the smallest unit and the most platform involvement. Seeing all three as one spectrum makes the choice less tribal.

Sign in to keep your progress.

Free resources

Links last checked 29 Aug 2026.

Stuck here?

Ask a mentor. A real person answers, and they can see exactly which topic you're on. Usually within a couple of working days.

Checking your session…

Topics shown in module order.