← BackReference (opens in a new tab)

The scaling proposal

More services are proposed before the bottleneck is measured. · PM Drills · 14 · 30 sec–2 min

The scaling proposal · 30 sec–2 min

Situation

A report slows as customer data grows.

The team suggests splitting the application into microservices. Nobody has measured where request time is spent.

Think before scrolling

What evidence should precede the architecture decision?

What smaller interventions might be worth comparing?

How a strong PM thinks

Find the constrained step.

Measure request traces, query behavior, data size, and peak workload. If a database scan dominates, new service boundaries may add complexity without improving the query.

Compare indexing, narrower scope, precomputed summaries, background work, or additional capacity with engineering. Define freshness and latency requirements so each option can be assessed against the same user outcome.

Takeaway

Architecture should answer an observed constraint.

Measure the bottleneck and specify the workload before paying for more moving parts.