Skip to content

Engineering

In praise of boring infrastructure

The most reliable platforms we run are the least interesting to talk about. That is the point.

Chipo NdlovuHead of Platform Engineering6 min read

Every few months a new tool promises to change how infrastructure is built. Some of them are excellent. But the platforms we are proudest of are made from parts that have been in production for a decade, configured carefully and monitored well.

Choose technology your team can run at 3 a.m.

The question we ask about any component is not ‘is it powerful?’ but ‘can the person on call understand and fix it, alone, in the middle of the night?’ That question rules out a surprising amount.

Reliability is mostly the absence of surprises.

service: mobile-money-api
slo:
  availability:
    objective: 99.95
    window: 30d
  latency_p95_ms:
    objective: 300
alerting:
  burn_rate_fast: 14.4   # page
  burn_rate_slow: 6      # ticket
yaml — A service-level objective: written down, measured, reviewed monthly

Write down what ‘working’ means, measure it, and spend your innovation budget on the product rather than the plumbing.

ShareLinkedInXEmail