TiDB is an open-source, cloud-native, distributed SQL database written in Go and licensed under Apache-2.0, aimed at teams that need ACID transactions, real-time analytics, and vector search at scale without data silos.

What it is

TiDB (/’taɪdiːbi:/, where "Ti" stands for Titanium) is a distributed SQL database built for high availability, horizontal and vertical scalability, strong consistency, and high performance. It lives in the cloud-native Go ecosystem and deploys in public clouds, on-premises, or natively in Kubernetes. The architecture separates computing from storage: the TiDB Server coordinates query execution, TiKV provides row-based storage, and TiFlash provides columnar storage. Repository topics include cloud-native, distributed-database, distributed-transactions, htap, agent, agent-context, agent-memory, agentic, ai, and hacktoberfest.

The concrete problem is scale and separation. TiDB is designed for deployments that outgrow a stand-alone MySQL 8.0 instance, and it is positioned for agentic workloads that grow unpredictably. Hybrid Transactional/Analytical Processing (HTAP) collapses transactional and analytical processing into one system rather than separate engines, so data does not sit in silos and workloads do not become noisy neighbors. MySQL 8.0 compatibility means applications migrate without code changes or with minimal modifications, using familiar protocols, frameworks, and tools, with a suite of data migration tools to move application data in.

Key capabilities

• Distributed transactions use a two-phase commit protocol for ACID compliance, spanning multiple nodes and preserving correctness under network partitions or node failures.
• Horizontal and vertical scalability add nodes or increase resources of existing nodes without downtime, with computing and storage adjusted independently.
• High availability uses the Raft consensus protocol for reliability and automated failover, storing data in multiple replicas and committing transactions only after writing to the majority of replicas, with geographic replica placement for disaster tolerance.
• HTAP pairs the TiKV row-based storage engine with the TiFlash columnar storage engine; TiFlash replicates data from TiKV in real time using the Multi-Raft Learner protocol, and the TiDB Server coordinates query execution across both.
• MySQL 8.0 compatibility lets existing applications use familiar protocols, frameworks, and tools, and migrate without code changes or with minimal modifications.
• Cloud-native deployment runs in public clouds, on-premises, or natively in Kubernetes, where TiDB Operator automates cluster operations, and TiDB Cloud provides a fully managed service.
• Native support covers transactions, analytics, and vector search for agentic workloads, with topics for agent, agent-context, and agent-memory.

Who uses it and how

• Teams building agentic workloads with unpredictable growth use TiDB for transactions, analytics, and vector search in one system, per the project description.
• Platform teams on Kubernetes run self-managed clusters through TiDB Operator, which automates cluster operations.
• Groups that prefer managed operations use TiDB Cloud, the fully managed service that provisions clusters in a few clicks.
• Organizations with mixed transactional and analytical queries run HTAP on one system, with TiKV serving row-based work and TiFlash serving columnar work.
• Deployments needing multi-AZ high availability and geographic replica placement use Raft consensus across public cloud or on-premises infrastructure.
• Teams migrating from MySQL 8.0 use the compatibility layer and data migration tools to move application data with no code changes or minimal modifications.

Getting started

Deploy TiDB with TiDB Operator on Kubernetes for a self-managed cluster, or use TiDB Cloud for a fully managed service that sets up clusters in a few clicks. The README also covers deployment in public clouds and on-premises.

How it compares

The facts name MySQL 8.0 as a compatibility target rather than a competing product, and name no other similar database. On the evidence provided, TiDB stands alone in this registry.

When to use it — and when not to

Self-hosting TiDB means operating the TiDB Server, the TiKV row-based storage engine, the TiFlash columnar storage engine, and Raft consensus, typically on Kubernetes through TiDB Operator, which is a distributed system footprint rather than a single binary. Teams that want a simple single-node MySQL drop-in, or that lack operational capacity for a distributed database, should not pick it. The facts show 6,976 open issues and a last push on 2026-09-17, with no release tag or version identified in the provided excerpt, so the issue backlog is substantial and the README excerpt is truncated.