Summary
This guide provides an overview of reflections in Dremio, focusing on their definition, purpose, and types. Reflections are optimized materializations of source data or queries that enhance query performance. The document outlines three primary types of reflections: Raw, Aggregate, and External, each serving distinct purposes in optimizing query execution. It details the role of reflection system tables, which manage metadata related to reflections, including their dependencies and materializations. The guide emphasizes the importance of evaluating reflection usage regularly, suggesting that users and administrators assess the impact of reflections and remove those that are not utilized. It also discusses key metrics for evaluating reflection effectiveness, such as Considered Count, Matched Count, and Accelerated Count. Furthermore, the document provides strategies for adding reflections based on query performance analysis and highlights the benefits of using raw reflections for large datasets and complex queries. Best practices for implementing reflections are also included.
