Capacity Model

The capacity engine lives in src/engine/capacity.ts.

Formula flow

Raw Pool TB
  └─ apply per-drive efficiency   (driveSizeTB × 0.92 by default)
       └─ subtract reserve drives (min(nodeCount, 4) × usablePerDriveTB)
            └─ subtract infra CSV footprint
                 └─ apply resiliency factor
                      └─ effectiveUsableTB

Resiliency factors

Mode	Factor	Min nodes
Two-way mirror	0.5	1 (the only option on a single node)
Three-way mirror	0.333...	3
Dual parity	0.5 to 0.8 all-flash, 0.5 to 0.727 with HDD capacity, by node count	4
Nested two-way mirror	0.25	2
Four-way mirror	0.25	4, rack aware clusters only (2+2 to 4+4)

A rack aware 1+1 cluster uses two-way mirror. See single node and rack aware clusters for the rules and worked numbers, and the in-app References page for the Learn source of each.

Key outputs

Field	Description
rawPoolTB	Total physical capacity before any deduction
usablePerDriveTB	Per-drive usable capacity after the efficiency factor
reserveTB	Capacity held back by reserve drives
infraVolumeTB	Pool footprint of the cluster infrastructure volume
availableForVolumesTB	Pool space available for user volumes before resiliency
netPoolTB	After reserve deduction
resiliencyFactor	Fraction of available pool that becomes usable storage
effectiveUsableTB	Final usable TB; feed into volume planner

There is no blended efficiency factor: the old 0.92 constant was removed in 2.4.1. Raw drive bytes go through the pipeline above and the TB to TiB conversion happens once, at display.

Calculator TB vs WAC GB

effectiveUsableTB is in Calculator TB (1 TB = 1,000,000,000,000 bytes in the Excel model). When you create a volume in Windows Admin Center, you must enter WAC GB — the floor-to-nearest-GB conversion.

$$\text{WAC GB} = \lfloor \text{Calculator TB} \times 1000 \rfloor$$

This distinction is surfaced on every volume row and on the final export.