Follow the white rabbit to find *very* detailed in-source docs on the internals: github.com/blambov/cassandra…
cassandra/MemtableTrie.md at CASSANDRA-17240 · blambov/cassandra
Mirror of Apache Cassandra. Contribute to blambov/cassandra development by creating an account on GitHub.github.com
Yes, we have performance results and should share. Tries have been in @DataStax Enterprise 6 line for years, and the difference is tens of percents, and more for some specific workloads. (Part of the benefit has more to do with Java heap and GC tricks than the index structure.)
According to dbdb.io/browse?indexes=patri… not many real world databases use Tries yet, and even then not all of those are open source. So I expect the Cassandra implementation to be of interest to the wider #database community too.
Database of Databases
Encyclopedia of databases systems from Carnegie Mellon University. Curated by @andy_pavlodbdb.io