@mwlucas Hopefully I don't do that, but if I needed to I would use Himpy [1] which would integrate in our Rieman [2] → Samplerr [3] → ElasticSearch → Grafana metric processing pipeline.
References:
1. https://github.com/pyr/himpy
2. http://riemann.io/
3. https://github.com/ccin2p3/samplerr