We rely heavily on Apache Cassandra as our primary data storage at Librato, using it to store both real-time metrics and historical rollup time-series. In this post, we explore a lesser known storage option that can have a large impact on Cassandra storage efficiency.