It seems like there's a lot of overlap between InductionDB and what p2panda is doing, should investigate what parts may be useful