Skip to main content
Version: Nightly

GreptimeDB Table Engines

Overview​

GreptimeDB offers multiple specialized table engines, each designed to excel at specific workloads and use cases. This document provides a comprehensive introduction to these engines and guidance on when to use each one.

Mito Engine​

Mito is the default storage engine of GreptimeDB, responsible for efficiently storing and managing database data. Built on the LSMT (Log-structured Merge-tree) architecture, Mito has been extensively optimized for time-series data workloads.

The engine features a robust architecture including Write-Ahead Logging (WAL) for durability, a memory table system, and an efficient compaction strategy based on Time Window Compaction Strategy (TWCS). This design enables Mito to handle high-throughput write operations while maintaining excellent query performance.

Mito seamlessly integrates with various object storage solutions including S3, GCS, and Azure Blob Storage, providing native support without additional plugins. It implements a tiered cache system on top of object storage, optimizing both storage costs and access speeds for time-series data at any scale.

Metric Engine​

As the name suggests, the Metric Engine is designed to process metrics data efficiently. It specializes in handling scenarios with a large number of small tables typical in observability and monitoring workloads.

The key innovation of the Metric Engine is its use of synthetic wide physical tables to store data from numerous small tables. This approach enables efficient column and metadata reuse across tables, significantly reducing storage overhead while enhancing columnar compression efficiency. Under the Metric Engine, tables become lightweight logical constructs, making it ideal for cloud-native monitoring scenarios where thousands of small metric tables are common.

The Metric Engine is built on top of the Mito Engine, meaning that its data is actually stored in the Mito Engine. This architecture leverages Mito's robust storage capabilities while adding specialized optimizations for metrics data management.

File Engine​

The File Engine is a specialized storage engine in GreptimeDB designed for handling file-based data sources. It allows GreptimeDB to directly query and process data stored in external files without requiring data import or conversion.

This engine supports various file formats commonly used in data analytics workflows, enabling seamless integration with existing data pipelines. By treating external files as virtual tables, the File Engine provides SQL query capabilities over file-based data while maintaining the performance optimizations of GreptimeDB's query engine.

The File Engine is particularly useful for scenarios where data already exists in file formats, allowing users to analyze this data alongside time-series data stored in other GreptimeDB engines. This capability makes GreptimeDB more versatile as a unified analytics platform that can work with diverse data sources.

Engine Selection Guide​

When to Use Each Engine​

  • Mito Engine: As the default storage engine, Mito is suitable for most general time-series workloads. It's an excellent choice when you need a balance of write throughput, query performance, and storage efficiency. Use Mito for applications requiring durable storage with good all-around performance characteristics.

  • Metric Engine: Choose the Metric Engine when dealing with observability and monitoring scenarios that involve thousands of small tables with similar schemas. This engine excels at reducing storage overhead and improving query performance in cloud-native monitoring environments where metrics data is predominant.

  • File Engine: Opt for the File Engine when you need to query data that already exists in external files without importing it into the database. This is ideal for data exploration, one-time analysis tasks, or when integrating with existing file-based data pipelines.

Specifying Engine Type in SQL​

When creating a table in GreptimeDB, you can specify which engine to use through the ENGINE clause in your CREATE TABLE statement. For more detailed information on syntax and options, please refer to the CREATE TABLE documentation.