Session

Back

ZomboDB: Making Postgres & Elasticsearch Work Together Like it's 2018

Date: 2018-09-07
Time: 11:30 - 12:20
Room: Mission
Level: Intermediate

ZomboDB lets you "CREATE INDEX" on your existing Postgres tables, and it automatically creates, synchronizes, and manages Elasticsearch indices for you. It also has an expressive query language allowing you to "SELECT" records that match complex full-text queries. Additionally, ZomboDB stores key Postgres MVCC metadata in its Elasticsearch indices to enable Elasticsearch to mimic Postgres' visibility rules and answer certain queries wholly within Elasticsearch.

ZomboDB's ultimate feature is that it eliminates the problems of ghost-reads, dead-reads, and otherwise out-of-sync text indices that typically occur when Postgres and Elasticsearch are used together via application code.

We developed ZomboDB so that you don't have to. We would be honored to present at PostgresOpen SV 2018 with a "live-demo" demonstrating ZomboDB's features using a familiar data set. We'll describe some of ZomboDB's implementation details and demonstrate how to solve common text- search record visibility problems. We'll also demonstrate advanced features: Query language, aggregation support, document scoring, hit-highlighting, and Elasticsearch cluster management. All of these features are possible thanks to the Postgres extension framework, including the Index Access Method API.

Speaker

Emmett Plant
Eric B. Ridge