Tool 01 · Backup & Storage

Backup retention & storage sizing calculator

Model a GFS retention scheme against your real environment, data growth, daily change rate, deduplication and secondary copies, and get the capacity you actually need over the next five years. Everything runs in your browser; nothing is uploaded.

Environment

Backup engine
Source data
TB
%/yr
%
: 1
GFS retention
days
weeks
months
years
Copies

Required capacity

at year 5

Year 1
Year 3
Year 5

Storage composition · year 5

Full backups () Incrementals ()

60-month projection

Retention pools fill up over the first months (yearlies take up to 5 years to accrue), then growth drives the curve.

Breakdown

HorizonSource dataFulls retainedBackend used+20% headroom
Link copied

How it's calculated

Source data grows monthly at your compounded annual rate: FETB(m) = FETB₀ · (1+g)^(m/12). Each retained full costs FETB(m) / reduction ratio; each retained daily incremental costs FETB(m) · change rate / reduction ratio.

Retained fulls follow the GFS accrual: weeklies cap at your weekly retention, monthlies fill one per month, yearlies one per year until each pool reaches its limit. That is why the curve climbs steeply early on. Overlap between pools (a monthly that is also the current weekly) is counted once. The secondary copy doubles backend usage with the same retention. The headroom column adds 20% for indexing, housekeeping and safety margin. Size your library against that number, not the raw figure.

The deduplication model changes the cost of retained fulls. Global dedup (Commvault DDB, Data Domain, appliance-style engines) stores shared blocks once across all copies: each retained full only adds the blocks changed since the previous one (change rate × interval, capped at 100%). That is why weekly fulls are nearly free but a 5-year-old yearly still costs close to a full copy. Per-copy applies the reduction ratio to each copy independently (per-job dedup or plain compression): every retained full costs its whole reduced size. None models raw storage, tape-style, ignoring the reduction ratio entirely.

This is a planning model, not a quote: real dedup ratios vary by workload, and synthetic fulls or forever-incremental engines change the math. Validate against your vendor's own sizer before purchase.