RAID Capacity & Fault-Tolerance Calculator

Inspectable records

Ledger state

The current ledger marks this instrument verified clean within the source-bound verification corpus.

Area

Computing & Information

Tranche

tranche-4

Evidence ID

t4-040

Ledger rows

1

No ledger note is attached.

Review findings

The capacity formulas are broadly standard under equal-drive assumptions, but the specification needs explicit validity constraints, an overhead formula, and a defined RAID10 tolerance policy. The honest-label warning is appropriate, but it should explicitly state that RAID10 tolerance varies by mirror-pair placement and that displayed capacity excludes vendor and filesystem overhead unless those are modeled.

Formula and scope findings

• RAID0 capacity N×S is correct for equal-sized drives, but RAID0 tolerates zero drive failures and requires a valid drive-count constraint.
• RAID1 usable capacity S is correct for a mirrored set of equal-sized drives, but the formula omits the required minimum drive count and does not define tolerated failures for N-way mirrors.
• RAID5 capacity (N-1)×S and one-drive tolerance are correct only for N≥3 equal-sized drives.
• RAID6 capacity (N-2)×S and two-drive tolerance are correct only for N≥4 equal-sized drives.
• RAID10 capacity (N/2)×S is correct only when N is even and drives are arranged as equal-sized mirror pairs; tolerated failures are not a single fixed number. Four drives can tolerate one failure in the worst case and two failures if they occur in different mirror pairs.
• The specification claims an overhead-percentage output but provides no formula. Assuming raw capacity is N×S, overhead should be (raw capacity−usable capacity)/raw capacity×100.

Unit assumptions

• All drives have equal usable capacity S.
• Drive capacity and usable capacity use the same unit; the test vectors use decimal TB, where 1 TB=1,000,000,000,000 bytes.
• Overhead percentage is interpreted relative to raw capacity N×S, excluding filesystem, metadata, hot-spare, vendor-reserved, and thin-provisioning overhead.
• RAID10 drive count is even and represents complete mirror pairs.
• The RAID level is implemented as conventional capacity-only RAID, without vendor-specific reserved space or formatting losses.

Boundary hardening

BREAK_CLAIMED · SELF_FLAGGED

The zero-drive input causes a negative usable-capacity result and an undefined redundancy-overhead division by zero, so the instrument must reject it. The unit ambiguity can also produce materially different byte capacities unless TB versus TiB is made explicit.

Errata and corrections

No applied erratum is attached to this unique-name record.

Inspectable example

4 review vectors are recorded. One source-bound sample is shown below.

{
  "inputs": {
    "raid_level": "RAID0",
    "number_of_drives": 4,
    "drive_capacity": 2,
    "capacity_unit": "decimal TB"
  },
  "expected_outputs": {
    "usable_capacity": 8,
    "capacity_unit": "TB",
    "redundancy_overhead_percent": 0,
    "tolerated_simultaneous_failures": 0,
    "rebuild_risk_note": "No redundancy; any single drive failure loses the array. Vendor- and workload-specific rebuild/URE analysis is not performed."
  },
  "working": "Raw capacity = N×S = 4×2 = 8 TB. Usable capacity = 4×2 = 8 TB. Redundancy overhead = (8−8)/8×100 = 0/8×100 = 0%. RAID0 tolerated failures = 0."
}

Related workflows

No source-defined multi-instrument workflow membership is recorded.

Sources and evidence

These pointers bind the page to evidence. They do not grant authority or collapse verification into deployment.

• VERIFIED_LEDGER.json · VERIFIED_CLEAN
• evidence-tranche4/tranche4-results.json
• BOUNDARY_HAZARD_CATALOG.json