This document discusses Presto connectors and how Treasure Data optimizes the Presto connector for cloud storage. It provides details on:
1) How Treasure Data uses Presto as a distributed SQL query engine and developed its own Presto connector to interface with its cloud-based data storage system called PlazmaDB.
2) Key aspects of PlazmaDB including using PostgreSQL for metadata and S3 for storage, with transactions managed across these systems.
3) How data is partitioned in PlazmaDB to optimize query performance, including time index partitioning based on ingestion time and user-defined partitioning.