Object storage bills like the cloud: capacity, requests and traffic. Here is how to build, on Dell ObjectScale or ECS, a per-tenant and per-bucket cost model that survives scrutiny from a financial controller.
The hierarchy that structures everything
ObjectScale organizes data on three levels: the VDC (virtual data centre), the namespace, and the bucket. For billing, the simple rule that works: the namespace belongs to a tenant, and all its buckets follow. Reports then drill down to the bucket for detail — that is where teams recognize their applications.
The four dimensions to measure
- Capacity, in GiB-days. Measured daily then aggregated: a bucket that spikes to 50 TiB for three days then empties pays for three days. The monthly rate is divided by the number of days in the month and each daily charge is rounded to the cent — monthly totals therefore differ slightly from a naive capacity × rate calculation, and that is expected.
- Write operations — object creations and deletions, priced per 1,000 operations. This is the dimension that separates a dormant archive from a chatty application hammering the API.
- Ingress — data written to the platform, in $/GiB over the month.
- Egress — data read out, often priced higher: the standard cloud-provider model, applied to your private storage.
Building the rate profiles
An ObjectScale rate profile combines these four dimensions, with optional tiers on capacity and traffic. Three field-tested tips:
- Anchor your rates to a public comparable. If your $/GiB is higher than the equivalent commercial S3 offering, prepare the justification (sovereignty, latency, compliance) — the question will come.
- Do not price everything in month one. Start with capacity; add requests and traffic once consumers have accepted the principle.
- A bucket can override its namespace profile — useful for a grant-funded project or a contractual rate — as long as it keeps the same currency.
From monthly report to governance
The typical report shows, per tenant: each bucket’s average and end-of-month capacity, operation volumes, traffic in both directions, and the amount per dimension. Delivered automatically every month, it quickly becomes the main governance tool: it is what reveals the test bucket forgotten for eight months, the application performing ten million deletes a week, or the growth that justifies — numbers in hand — the next expansion.