Back to catalog
TiDB
Apache-2.0TiDB is a distributed MySQL-compatible database that replaces sharded clusters with native horizontal scaling and real-time columnar HTAP analytics.
Open-source alternative to — compare it against:
Overview
TiDB separates SQL parsing from storage by running stateless Go compute nodes over TiKV, a Rust-based transactional key-value engine using Multi-Raft consensus. Built by PingCAP since 2015, it scales reads and writes horizontally while speaking the MySQL wire protocol. Applications connect to a single logical database without manual sharding or middleware proxies.
Why TiDB instead of PlanetScale
Both TiDB and PlanetScale scale MySQL horizontally, but their integration models differ. PlanetScale wraps Vitess around sharded MySQL instances, relying on custom connection poolers to serve serverless apps. TiDB exposes a native MySQL 8.0 server endpoint over standard TCP, allowing standard MySQL client libraries, connection poolers like ProxySQL, and ORMs to connect directly. Because TiDB coordinates distributed transactions natively through two-phase commits, developers avoid configuring Vitess VSchema keys or handling cross-shard join constraints.
The integration difference also applies to analytics. PlanetScale focuses on transactional workloads and offloads analytical queries to external warehouses through ETL pipelines. TiDB integrates TiFlash, a columnar engine that replicates row data in real time via Raft learners. Complex aggregations and reporting queries run against columnar data at the same endpoint without slowing transactional writes. While PlanetScale removes maintenance through managed cloud hosting, TiDB offers deployment control across bare metal or Kubernetes via the TiDB Operator.
What it does well
- Native MySQL 8.0 wire compatibility lets developers connect existing applications with standard drivers and ORMs without code changes.
- TiFlash columnar engine replicates data asynchronously via Raft learners, enabling real-time analytics without degrading OLTP throughput.
- TiDB Operator automates Kubernetes deployments, handling rolling upgrades, pod scaling, and multi-zone failover for DevOps teams.
- Online schema changes execute across the cluster without locking tables or interrupting active read and write queries.
- Built-in placement rules allow DevOps engineers to pin tables, partitions, or replicas to designated hardware or cloud regions.
- Native Change Data Capture through TiCDC streams row-level changes to Apache Kafka or downstream databases with sub-second latency.
