#capacity-planning

RAID storage-array maths as an API, computed locally and deterministically. The capacity endpoint computes the usable and raw capacity, the storage efficiency and the fault tolerance of a RAID level — RAID 0 stripes for n×disk with no redundancy, RAID 1 mirrors to one disk and tolerates n−1 failures, RAID 5 gives (n−1)×disk with one-disk tolerance, RAID 6 gives (n−2)×disk with two-disk tolerance, and RAID 10 gives (n/2)×disk — and reports the minimum disks each level needs. The compare endpoint lays the levels side by side for the same disks and disk size so you can weigh capacity against redundancy. The rebuild endpoint estimates how long it takes to rebuild a single disk at a given rebuild speed, the window during which a second failure would lose data in RAID 5/6. Everything is computed locally and deterministically, so it is instant and private. Ideal for storage, NAS, server and IT-admin app developers, capacity-planning and procurement tools, and homelab calculators. Pure local computation — no key, no third-party service, instant. Live, nothing stored. 3 endpoints. This is RAID array sizing; for data-transfer time use a transfer API.

api.oanor.com/raid-api

Queueing-theory maths as an API, computed locally and deterministically. The littles-law endpoint applies Little's law, L = λ·W — the average number in a system equals the arrival rate times the average time in the system — and solves for whichever of the three you leave out; it holds for any stable system, from a checkout line to a request pipeline. The mm1 endpoint gives the full steady-state metrics of a single-server M/M/1 queue from the arrival rate λ and the service rate μ: the utilization ρ = λ/μ, the average number in the system and in the queue, the average time in the system and waiting, and the probability the system is empty — and it flags an unstable queue when ρ ≥ 1. The mmc endpoint extends this to a multi-server M/M/c queue with the Erlang-C waiting probability, returning the offered load in erlangs, the per-server utilization, the chance an arrival has to wait, and the same length and time metrics. Rates must share a time unit, and the times come out in that unit. Everything is computed locally and deterministically, so it is instant and private. Ideal for capacity-planning and operations tools, call-centre and staffing apps, server and throughput sizing, and operations-research education. Pure local computation — no key, no third-party service, instant. Live, nothing stored. 3 endpoints. This is queueing theory; for descriptive statistics on a list of numbers use a statistics API.

api.oanor.com/queue-api