That seems to be focused on throughput. I'm focused on how much data I can store into the system and if it's redundant so long as N nodes don't die (and if they do I'd like it to attempt to reorganize the network so they system keeps working with N node failure).
I don' t think so. My experience with it is non-existent, but I read up on it before and now again, and self-healing and replication was pretty high on the feature list. Anyway, what you describes is a distributed/clustered (fault-tolerant) file system. Tahoe-LAFS as mentioned by jabl is one of those, https://en.wikipedia.org/wiki/List_of_file_systems#Distribut... has a list.
