Trust

Method and provenance

Bayplan is a configuration engine with a shop window, not a review site that happens to have a form.

The three layers in this version

1. Storage target — current data, yearly photo and video growth, horizon, headroom. Output is a planning target in TB, before RAID and filesystem overhead.
2. Configuration — mutually exclusive rules pick bay count, redundancy class and performance class from that target plus a coarse media label.
3. Enclosure match — models whose bay count and class match. If the transcode class is empty, we show zero enclosures and no purchase step. We do not nearest-neighbour into another class.

V, R and D

• V — verified external fact. Bay count and the existence of a public product page for a named model.
• R — our deterministic or editorial rule. Headroom factors, the 8 TB line, performance class assignments, drive-class guidance.
• D — observed from our own data. Unused in this version. We will not relabel R as D.

Fail closed

Missing, negative or non-finite inputs, invalid shared-URL enums, or overlapping configuration rules return UNKNOWN or an error. A media label we cannot honour with a verified enclosure returns no match, not a nearby box with a buy button.

What this version will not invent

• Live prices, “from £x today”, or a shopping cart
• Drive part numbers or a live compatibility list (HCL)
• A transcode lab (CPU × codec × client)
• Cameras as a NAS category
• A filesystem overhead multiplier dressed as manufacturer data
• More than a handful of enclosure models

How merchant links work

After a named match, Amazon UK and Scan search links. Same pattern on enclosure pages, drive class searches, and the UPS / off-site complements. Manufacturer spec pages are not commercial. Ranking is not paid. Amazon Associates tracking is not live — URLs do not include a tag. Scan Awin tracking is not live.