Different resources are bound in fundamentally different ways, and that changes how you plan them. **Peak-driven (elastic)** — web servers, databases, caches consume a range of resources over a day and have a breaking point near resource saturation. The question is *how high does the peak go?* You find the periodic peaks, drill into what happens during them, and size for the peak. Load can be shed or shifted. **Consumption-driven (monotonic)** — storage only grows. The question is not "what's the peak" but **"when will I run out?"** You track the consumption rate (data volume per unit time) and project the run-out date. You can't shed what's already stored. Knowing which kind a resource is tells you which question to ask and which forecast to build — a peak envelope versus a depletion timeline. --- *Source: [[The Art of Capacity Planning]] (John Allspaw, O'Reilly 2008) — Ch 3 — Measurement: Units of Capacity*