DataLayer Conference: Rethinking Indexing in Data Stores with Replex

Next up at DataLayer, Amy Tai from Princeton University took the stage to discuss Replex and rethinking how to index in the modern, distributed datastore. Replex is a datastore that enables efficient querying on multiple keys.

Partitioning tends to use a single fixed index key which is not ideal when you need to do searches which involve a secondary index. Replex rethinks where data is placed during replication using knowledge of what keys we would be using to query data using multiple keys. It combines replication and indexing, hence Repl-ex.

Amy demonstrates the why and how of Replex and introduces hybrid Replexes which enable a new design space where architects can trade off steady-state performance for faster recovery.

If you're interested in taking a closer look at Amy's slide deck, download it here.


If you have any feedback about this or any other Compose article, drop the Compose Articles team a line at articles@compose.com. We're happy to hear from you.