> Scales to terabytes of data and hundreds of billions of keys per store > Efficiently handles thousands of concurrent operations > Dramatically improves thro

> Scales to terabytes of data and hundreds of billions of keys per store > Efficiently handles thousands of concurrent operations > Dramatically improves throughput, latency, write-amplification, and read-amplification versus common alternatives for many workloads > Optionally combines multiple classes of solid-state storage to optimize performance and endurance It would be nice to see some evidence/benchmarks supporting these statements. Perhaps I missed something but I haven't been able to find any on their page or github.