Node Architecture

In the vast landscape of database management, CrateDB distinguishes itself with its unique node architecture, offering unparalleled functionality and configurability. Rather than a primary-secondary setup, CrateDB utilizes a system where every node in the cluster can perform every operation. Each node is accessible via three ports: SQL queries are accepted, and results are returned via HTTP or the PostgreSQL Wire Protocol. Additionally, an extra transport port is used for inter-cluster communication.

CrateDB's node architecture is built on four major components:

• SQL Handler: The SQL Handler serves as the first point of contact for incoming client requests. It parses and analyzes SQL statements, then creates an execution plan.
• Job execution service: This service manages the execution of plans, also known as "jobs." These jobs, which consist of multiple operations, are distributed to nodes involved in the process via the Transport Protocol.
• Cluster state service: This service is at the heart of cluster setup, managing the cluster state. Its responsibilities include, but are not limited to, master node election and node discovery.
• Data storage component: This component is responsible for storing and retrieving data from disk based on the execution plan. CrateDB stores data in sharded tables, which are divided and stored across multiple nodes. Each shard is a distinct Lucene index stored physically on the file system.