Sustainability and TCO
Scope
This domain tests total cost of ownership and environmental accounting for tape: what to include, how to compute per-TB-year figures, and why offline media win on power.
Core rules
- Include everything: media, drives, library, floor space, power and cooling at the PUE-adjusted rate, software licences, staff time, migration labour and media refresh. Excluding migration labour is the classic way a disk comparison gets flattered.
- Media at rest draws no power and needs no cooling. A tape library powers robotics and drives only while mounting and streaming; an idle disk array draws power continuously and typically needs refreshing every five years.
- LTO media is rated for up to 30 years' archival life, so a tier can hold data for a decade with verification rather than continual copying. Industry studies, such as Brad Johns Consulting's 2021 comparison, put tape's carbon advantage at up to 87% versus disk for long-term archive - cite it as an industry study, not a standard.
- Report two metrics: $/TB-year and kWh/TB-year (or kgCO2e/TB-year). Anything else invites apples-to-oranges comparison.
- Security carries TCO value: an offline copy shortens ransomware recovery, which is a cost line, not only a risk line.
Worked example
5 PB, two copies, five years, LTO-9 at 18 TB native. Per copy: 5,000 ÷ 18 = 278 cartridges; two copies plus 10% margin = 612 tapes. Illustrative prices: tapes at $150 = $91.8k; eight drives $120k; library $150k; capex ≈ $362k, about $72k/year, roughly $14/TB-year before staff. The same 5 PB on disk pays continuously for power and cooling and refreshes every five years - that recurring line, not the purchase price, is where tape wins.
Traps
- Comparing compressed tape capacity against raw disk capacity.
- Leaving power, cooling and migration out of the model.
- Presenting a vendor carbon figure as an audited standard.