TiDB

TiDB is an open-source distributed SQL database compatible with MySQL, run as the TiDB Cloud managed service or self-managed, for apps that outgrow one server.

TiDB spreads one SQL database across many machines, so an application keeps a single relational schema while its data and traffic grow. It is compatible with MySQL, which eases moving existing MySQL workloads onto it. PingCAP builds TiDB and keeps the engine's source public under the Apache License. The same engine runs as TiDB Cloud, the managed service, and as TiDB Self-Managed. TiDB keeps its SQL layer apart from storage, so each side scales on its own. TiKV holds rows as a distributed key-value store replicated with the Raft protocol, and a write commits once a majority of replicas confirm it. When a minority of replicas fail, automatic failover keeps the cluster serving, and the Placement Driver hands out transaction timestamps and decides where data lives. TiFlash keeps columnar copies of chosen tables, so analytical queries read live transactional data without waiting on ETL. TiCDC streams row changes from selected tables onward, log backup restores a database to a chosen moment, and native vector search sits beside SQL. The database suits engineering teams whose MySQL server has reached its ceiling, and products that report on the same data they write. PingCAP builds its solutions for fintech, e-commerce, SaaS, logistics and web3, and for AI teams that want agent memory and vector retrieval in one store. Buyers choose a deployment rather than a separate product. TiDB Cloud runs managed on AWS, Google Cloud or Azure, a bring-your-own-cloud option places it in the customer's account, and Self-Managed runs on their own hardware. A common first project is moving a MySQL application onto a cluster that grows by adding nodes, then pointing Metabase at the same tables for live dashboards. Another is publishing order or payment changes through TiCDC into Kafka as Canal-JSON, Avro or Debezium messages, so downstream services react to each write. AI teams keep embeddings next to relational rows and use vector search to fetch context for agents. Platform teams that run their own infrastructure deploy and upgrade clusters with TiUP, or with TiDB Operator on Kubernetes. TiDB fits a team that wants familiar MySQL syntax while keeping transactions, analytics and AI retrieval in one database.