{{ summaryTitle }}
{{ summaryValue }}
{{ summaryLine }}
{{ badge.label }}{{ badge.value }}
RAID array capacity and sizing assumptions
One capacity per installed drive, including spares; use commas, spaces or line breaks. Largest drives become hot spares first.
RAID capacity planning accepts equal drives or a mixed-size inventory.
Includes RAID 0, 1, 5, 6, 10, 50, 60 and approximate RAID-Z1, Z2 and Z3 vdev capacity.
RAID sizing can start with installed drives or explicit repeated groups / RAID-Z vdevs.
Editing this value selects explicit groups × width plus spares. RAID 10 widths must be even; fixed RAID 50/60 geometry stays fixed.
Editing selects explicit geometry. RAID 50/60 require at least two groups; spares are outside the active groups.
Includes spares; the target search is bounded to 2,520 installed drives.
drives
TB and GB are decimal; TiB and GiB are binary. Changing units preserves the physical capacity.
Use zero when every installed drive participates in the RAID set.
spares
Each RAID 50 group needs at least three active drives; each RAID 60 group needs four.
groups
Use zero to review the current array without a minimum-capacity target.
Maximum fill after metadata and filesystem overhead. 100% adds no fill headroom; reserve headroom below is a separate deduction.
%
Ten percent is the planning sample, not a vendor requirement.
%
{{ workflowFeedback }}
Additional filesystem allowance after metadata; zero leaves capacity unchanged.
%
Applied after target fill, before reserve headroom; zero leaves capacity unchanged.
%
Zero keeps the generic RAID-capacity result unchanged.
%
Zero disables the enclosure-fit check.
slots
Zero disables procurement estimates and leaves capacity unchanged.
$
{{ capacityExportStatus }}
Plan metricValuePlanning readCopy
{{ row.metric }}{{ row.value }}{{ row.note }}
{{ comparisonExportStatus }}
LayoutPlanned usableEfficiencyFault toleranceTarget fitCopy
{{ row.layout }}{{ row.usable }}{{ row.efficiency }}{{ row.faults }}{{ row.target }}
{{ pathExportStatus }}
Sizing routeRequirementResultDecisionCopy
{{ row.route }}{{ row.requirement }}{{ row.result }}{{ row.decision }}
{{ chartExportStatus }}

Similar tools

A storage target of 60 TB cannot be compared directly with eight drive labels that add up to 96 TB. Some installed drives may be dedicated spares, while the active set spends capacity on mirrors or parity. Filesystem overhead and unused operating reserve reduce the amount that should be promised to applications again.

Redundant array of independent disks (RAID) combines equal-sized members through striping, mirroring, parity, or nested parity groups. The chosen layout determines how many active drives contribute data capacity and how many provide protection. RAID 0 uses every active member for data but provides no redundancy. RAID 1 and RAID 10 trade roughly half or more of the active capacity for mirrors. RAID 5 and RAID 6 use one or two drive equivalents for parity; RAID 50 and RAID 60 repeat that cost across groups.

Capacity deductions in a RAID plan
Capacity stageWhat it includesWhy it differs from the drive labels
Installed rawEvery installed drive, including dedicated hot spares.It is procurement capacity, not active array capacity.
Usable before overheadData-drive equivalents after spares, mirrors, parity, and incomplete groups.Protection and idle members do not hold unique data capacity.
Planned usableUsable capacity after metadata/system overhead and reserve headroom.It represents the amount available to plan against the target.

Decimal and binary units are another frequent source of apparent loss. One TB is 1012 bytes, while one TiB is 240 bytes. Converting the same byte quantity between them changes the number displayed, not the physical capacity. Keep the input and target units explicit when comparing a vendor quote with an operating-system report.

Grouped layouts require complete groups. When active drives do not divide evenly by the chosen RAID 50 or RAID 60 group count, the remaining drives are shown as idle rather than silently added to usable capacity. RAID 10 similarly uses complete pairs. A drive-count recommendation can therefore rise in steps instead of gaining capacity with every added bay.

Capacity is only one storage decision. This sizing model does not estimate controller limits, random I/O, rebuild time, degraded performance, or failure probability. RAID also does not replace backups: it may keep an array available through certain member failures, but it does not recover deleted, corrupted, encrypted, or site-lost data.

How to Use This Tool:

Describe the installed shelf first, then apply the capacity target and operating deductions.

  1. Choose the RAID layout and enter installed drives, equal effective drive size, and capacity unit.
  2. Subtract dedicated hot spares from active membership. For RAID 50 or RAID 60, set the intended parity-group count and correct any group-width warning before using the result.
  3. Enter the target usable capacity, metadata or system overhead, and reserve headroom. Overhead is deducted first; reserve is then taken from the remaining usable amount.
  4. Add a chassis slot limit or cost per drive only when those checks affect the purchasing decision. A zero value disables either optional check.
  5. Read Planned usable capacity, target gap, and idle capacity before comparing alternative layouts or the minimum-drive path.

Interpreting Results:

Planned usable capacity is the main sizing result. It has already removed dedicated spares and protection drive equivalents, then applied metadata/system overhead, filesystem overhead, target maximum fill, additional safety reserve, and reserve headroom in that order. Installed efficiency compares that result with all installed raw bytes, including spares and any idle members.

  • Minimum drive count searches up to 2,520 installed drives with the same layout, hot spares, and percentage settings. Added drives use the current effective member size. Explicit-group sizing adds whole groups at the chosen width, up to 64 groups; installed-drive sizing keeps the selected group count fixed.
  • Required drive size keeps the current geometry and estimates the equal member size needed to meet the target after all selected capacity deductions.
  • A negative target gap is a shortfall; a non-negative gap means the target is met. A zero target disables the fit comparison.
  • Idle uneven capacity means active drives could not complete a RAID 10 pair or nested parity group under the selected geometry.

Higher installed efficiency is not automatically a better design. RAID 0 maximizes capacity while offering no redundancy, and parity or mirror capacity may be an intentional availability cost. Confirm layout support and recovery behavior with the actual controller or storage platform.

Technical Details:

The calculation uses equal-member drive equivalents. Active drives equal installed drives minus dedicated hot spares. The layout then resolves active members into data, protection, and possibly idle counts before any percentage deductions are applied.

Formula Core:

All capacity arithmetic is performed in bytes. Data capacity is reduced by metadata overhead, then filesystem overhead, target maximum fill, additional safety reserve, and reserve headroom. Each percentage applies to the capacity remaining after the preceding stage; the retained fraction is k.

Nactive=Ninstalled-Nspare Cpre=NdataSdrive k=(1-o100)(1-f100)t100(1-s100)(1-r100) Cplan=Cprek Srequired=CtargetNdatak

Ndata is the data-drive equivalent count and Sdrive is the smallest engaged member's byte capacity. The percentages are o for metadata/system overhead, f for filesystem overhead, t for target maximum fill, s for additional safety reserve, and r for reserve headroom. The required-size equation assumes equal replacement members and is used only when the target is positive and the resolved layout has data members. Display values are rounded after byte-level calculation.

For eight 12 TB RAID 6 members with 80% target fill, 10% reserve headroom, and zero metadata, filesystem, and additional safety deductions, planned usable capacity is 51.84 TB. A 60 TB target requires 13.89 TB per member when rounded to two decimal places.

Rule Core:

RAID data-drive and buildability rules
LayoutMinimum active drivesData-drive ruleModeled guaranteed failures
RAID 01All active drives0
RAID 12One data copy; remaining active members are mirrorsActive drives minus 1
RAID 53Active drives minus 11
RAID 64Active drives minus 22
RAID 104Half of the largest even active countAt least 1; failure placement matters beyond that
RAID 506Complete groups minus one parity member per group1 per group in the geometry model
RAID 608Complete groups minus two parity members per group2 per group in the geometry model

This capacity model accepts one active RAID 0 member: one 12 TB drive contributes 12 TB before deductions, with no redundancy. Confirm that the storage platform supports the intended member count.

Nested parity groups use the floor of active drives divided by group count. RAID 50 requires at least three active drives per group; RAID 60 requires at least four. Engaged drives equal whole group count times group width, and any remainder becomes idle active capacity. RAID 10 also drops one active member from the engaged set when the count is odd.

The optional drive-count ladder evaluates nearby whole counts with the same spares and group setting. A candidate that cannot form the selected layout is marked unbuildable rather than assigned zero-capacity meaning. Chassis warnings appear when installed drives or the first target-meeting count exceed a positive slot limit.

Accuracy Notes:

Real usable capacity can differ after controller metadata, filesystem allocation, volume-manager rules, snapshots, thin provisioning, and vendor-specific reserve are applied.

  • Use the smallest effective member size for mixed-drive arrays unless the platform documents different behavior.
  • Check whether the chosen controller supports the RAID level, drive count, group width, and hot-spare policy.
  • Do not infer performance or rebuild safety from capacity efficiency.
  • RAID redundancy does not replace independent backups and tested restores.
  • The values are calculated in the browser and are not sent to a dedicated storage-sizing service.

Worked Examples:

Eight-drive dual-parity shelf

Eight 12 TB drives in RAID 6 with no hot spare provide six data-drive equivalents, or 72 TB before deductions. With no metadata overhead and 10% reserve, planned usable capacity is 64.8 TB. A 60 TB target is met with 4.8 TB remaining.

One drive cannot complete the nested groups

Twenty-five active drives split across four RAID 60 groups form four groups of six. Eight members are protection equivalents, 16 are data equivalents, and one active drive is idle because it cannot be assigned evenly to the four groups.

References: