"network disruption as a service" and proof of local storage

The Bitcoin community is discussing ways to stop a single entity from running multiple nodes in order to harvest transaction origin IPs. One idea proposed is to challenge nodes to send a specific piece of data, such as SHA256(SHA256(concatenation of N pseudo-randomly selected bytes from the blockchain)). The response time would be too slow for anyone trying to cheat, but it would require each node to keep its own copy of the blockchain. Some participants have suggested that a company could host hundreds of lightweight servers all backed by a single copy of the blockchain, rather than forcing every machine to have its own copy. The discussion also includes improvements to the protocol that would allow verifiers to detect non-full nodes with less computation. There is also some debate about whether spinning disks or SSDs are better for reading random blocks of data from the blockchain. Ultimately, only nodes that wish to prove full-node-ness, such as those seeking bitnodes subsidy, would need to implement the protocol.