The case against real-time everything
Real-time sounds like a universal upgrade. It's often expensive complexity solving a problem you don't have.
There's a reflex to make everything real-time — live dashboards, instant syncs, streaming updates. It feels modern. But real-time carries real cost in complexity and reliability, and most decisions don't move at real-time speed.
Match latency to the decision
If a number informs a weekly decision, updating it every second buys nothing and costs plenty. Ask how fresh the data actually needs to be for the decision it serves. Often the honest answer is 'hourly' or even 'daily,' which is dramatically simpler to build and keep running.
The hidden tax
Real-time systems fail in more interesting ways, are harder to debug, and demand constant attention. Batch is boring and boring is reliable. Reserve real-time for the few places where the freshness genuinely changes an outcome.
Real-time is a cost, not a feature. Pay it only where the decision actually moves that fast.