Conventional wisdom dictates that cloud data should be stored at a trusted data center with standards for privacy & security—and in a clearly identified jurisdiction. But convention is outdated. Distributed, peer-to-peer storage is more robust, less susceptible to hacking and—with private key user credentials—impossible to subpoena. With proper architecture (what I call “torrent reacquisition”), it can be fast and transparent.

A few cloud storage providers have baked-in my suggested architecture, but the fulll vision has yet to be adopted. I envision a combination of three novel twists on the existing model: a) A hubless, fully-distributed, P2P storage architecture (no data center); b) Reverse the model: Live cloud with Local/Physical backup); c) Torrent reacquisition to restore data.

The market and methods for specifying a network storage architecture is highly fragmented. Decisions made regarding network layout, and capital equipment will have repercussions into the future. Which architecture is best for a given set of criteria. Which technologies are durable for the long haul: RAID, fiber-bus, load balancing, real-time encryption, hashing, peer matrices, etc. Should the whole project be hosted by a virtual service? I am working toward a standards-based approach for optimizing tradeoffs when specifying a storage architecture: Access speed, fault tolerance, initial cost, ongoing maintenance cost, changing demands, etc.