We've done a lot of prototyping work in Ouija that leads into our "open" / "big" data initiatives for 2015, including analyzing test failure rates, identifying test machines with disproportionate numbers of failures and identifying extraneous test automation. I don't think maintaining these prototypes long term makes sense and standing them up in our "open" / "big" data system would provide some good initial use cases. I've already filed Bug 1071152 to get the machine failures working in Treeherder. I'll follow up with other blockers here.