Make the storage plan inspectable.

Units: drive labels and entered data use decimal TB and GB. Results also show TiB, where 1 TiB = 2⁴⁰ bytes, so the unit conversion is visible instead of being mistaken for filesystem overhead.

Forecast data: irreplaceable and replaceable data are forecast independently as current decimal TB + monthly decimal GB × planning months ÷ 1,000.

Decision path: a first-purchase plan carries forecast data, horizon, operating reserve, and failure tolerance into the bay lifecycle comparison before one selected path enters an exact RAID check. An existing-system plan carries the installed equal-drive baseline and the reserve-inclusive capacity target into the detailed replacement sequence.

Capacity target: total forecast data + the selected operating reserve. The reserve remains a separate value rather than being hidden in a fixed multiplier.

Bay lifecycle comparison: every 2-, 4-, 6-, and 8-bay path receives the same reserve-inclusive target. The per-drive minimum is target capacity ÷ data-drive equivalents, rounded up to the maintained drive-label classes. Modelled runway is the point where forecast data consumes layout capacity after retaining the selected operating reserve.

Equal-drive layouts: raw drive labels − mirror/parity capacity − operating reserve. The remaining planning capacity is compared with forecast data, so the reserve is not counted twice. Filesystem, metadata, snapshots, and vendor-specific overhead remain platform checks.

Inventory upgrade path: standard RAID is constrained by the smallest member across the array; SHR-1 is modelled in protected capacity layers and enforces Synology’s mixed-drive replacement-size boundary. Each checkpoint is compared with the usable-capacity target carried from the unified workspace.

Rebuild-window range: the smallest entered healthy-member read, replacement write, and controller/pipeline rate defines the raw bottleneck. Reconstruction scope ÷ raw bottleneck is the transfer floor. Reserving the selected foreground-workload share and applying the editable planning efficiency produces the degraded-state planning window compared with the user-owned time target.

Operating envelope: aggregate network demand = concurrent clients or jobs × entered MB/s per client + background MB/s. Effective link capacity = entered Gb/s × 1,000 ÷ 8 × editable network efficiency. Memory compares a workload-owned requirement with candidate capacity. Average drive watts interpolate between entered idle and active values by active-time share; average system watts add base power and every drive, while peak watts place every drive at the entered active value. Annual energy applies operating hours per day across 365.25 days. Candidate dBA is compared directly with the user-owned placement maximum without combining unmatched acoustic measurements.

Constraint order: unfinished candidate specifications are resolved before a pass is emitted. Completed gaps are ordered by the declared workload focus: application/VM paths inspect memory first, while file, media, backup, and mixed paths inspect aggregate network demand first; peak power and active placement noise remain explicit checks.

Lifecycle budget: primary array = enclosure or host quote + drive count × per-drive quote + optional spare. Acquisition subtotal keeps the primary array, independent backup, UPS, and networking values separate. Optional lifecycle energy = average whole-day system watts × 24 × 365.25 × years ÷ 1,000 × entered energy price. A zero-price purchase remains unresolved until the requirement is explicitly marked already available.

Independent backup baseline: forecast irreplaceable data + selected reserve. Version history, retention, and additional copies can increase the required destination.

Protection layers: array redundancy, snapshots, independent local copies, and offsite copies remain separate because they cover different failure boundaries. Only a copy eligible for the selected array, enclosure, or site-loss boundary is used for the independent recovery objective.

Recovery-point objective: the eligible independent RPO is the shortest enabled copy interval that survives the selected failure boundary. Snapshot cadence remains visible as the fastest rollback point but is not promoted to an independent recovery point.

Restore-time range: priority restore data ÷ the smallest of network throughput, source read, and destination write gives a transfer floor. Applying the editable protocol/workload efficiency produces the planning edge compared with the RTO.

Measured restore verification: restored decimal TB × 1,000,000 MB/TB ÷ elapsed seconds gives observed MB/s. Priority restore data ÷ that measured rate produces the projected full recovery window; verification also requires the recovered point to meet RPO and every recorded usability-evidence check to be complete.