Purpose
The capacity planner compares storage, processing and cost for a test-data estate under a ladder of scenarios: full copies per environment, a right-sized subset, compressed snapshots, a reusable certified dataset shared by all environments, a retention policy and incremental refresh. Each scenario keeps the previous one's choices and adds one lever.
It is available in the console (Capacity & Cost) and, as an educational calculator, in the public learning center. Every result is an estimate from the assumptions shown, not a measurement, benchmark or price quote. Prices are illustrative placeholders to replace with your own rates.
Where the numbers come from
The authoritative arithmetic is the control plane's POST /v1/capacity/estimate. The planner calls it once per scenario and converts bytes to cost with your prices. The console and learning center use an exact TypeScript port so the comparison can update as you type; shared golden vectors prove the port and the control plane return identical numbers. In the console, Estimate with the control plane re-checks your plan against the server and shows whether the results match.
Only assumptions are sent to the control plane: sizes, ratios, counts and prices. No data is involved.
Model
- Subset size = estate size × subset ratio; compressed size = subset × compression ratio.
- Storage per copy = compressed size × (1 + (versions kept − 1) × fraction rewritten per refresh). The first version is stored in full; later versions store what changed.
- Copies = 1 when a certified dataset is shared, otherwise one per environment.
- Processing per month = refreshes per month × subset size × fraction rewritten × copies.
- Cost = bytes ÷ 10^9 × price per GB, rounded half-to-even to cents.
The full-copy baseline uses a subset ratio and compression ratio of 1, no sharing, the baseline number of versions and full rebuilds.
Assumptions
| Assumption | Default | Notes |
|---|---|---|
| Production estate size | 100 TB | Decimal: 1 TB = 10^12 bytes (about 0.91 TiB). Hypothetical. |
| Rows in the estate | 400,000,000,000 | Only used for the subset row estimate. |
| Non-production environments | 6 | DEV, QA, SIT, UAT, PERF and similar. |
| Subset ratio | 0.05 | Share of the estate a referentially intact subset keeps. |
| Compression ratio | 0.3 | Depends heavily on your data; validate it. |
| Versions kept (full-copy baseline) | 3 | Copies kept today when cloning production. |
| Versions kept (retention policy) | 2 | Certified versions kept under a retention policy. |
| Refreshes per month | 4 | |
| Fraction rewritten per incremental refresh | 0.2 | 1 means a full rebuild. |
| Storage price | 0.023 per GB-month | Illustrative only. |
| Processing price | 0.01 per GB processed | Illustrative only. |
Limits
Byte and row counts are limited to 9 PB (9 × 10^15) so every implementation can represent them exactly. The model ignores network transfer, licensing, staff time and the compute used for masking or synthesis; add those separately if they matter to your decision.