Abstract

This disclosure describes a method of building and operating data capacity as a modular data park whose permanent core is a governed data-custody backbone rather than a single large building. The backbone holds custody of the data, namely its copies, locations, offline tier and restore paths, and is independent of the compute and storage hardware that serve it. Capacity is added in standard modules, each connected to the park's power, cooling, network and custody backbone through declared standard interfaces. A module is added only when its use is contracted, so that capacity is not built ahead of demand. Before any module is built, its effects on power, water, heat and noise are forecast on the park's certified site twin and recorded, and the forecast is added to the park's community compact as a commitment for that module, checked afterwards against meters. Because the data's custody lives in the backbone, a module of an older hardware generation can be replaced by a newer one without moving custody: data is restored onto the new module from the backbone, the restore is drilled and reconciled against the twin's forecast, and only then is the old module retired. Retirement is recorded in the supply-chain and custody records, so that no data is stranded and no equipment leaves unrecorded. The park therefore grows, refreshes and shrinks module by module, with each step forecast, consented, contracted and proven.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Download

Share

COinS