Introduction
VeloDB Cloud is a fully managed real-time analytics and search database built on Apache Doris. A single engine handles a broad range of real-time analytics, from SQL analytics to full-text and vector search.
What you can build
Each of these usually requires a separate system for OLAP, search, or lakehouse federation. VeloDB Cloud runs them on one engine:
| If you need to | VeloDB Cloud |
|---|---|
| Power real-time dashboards and apps for many concurrent users and AI agents | Ingests streaming data and serves sub-second SQL queries at high concurrency. |
| Keep analytics in sync with your operational databases | Continuously ingests inserts, updates, and deletes from MySQL and PostgreSQL through CDC, using primary-key upserts. |
| Store and search logs, metrics, and traces | Keeps observability data at object-storage cost and searches it with full-text inverted indexes. |
| Retrieve context for LLMs and agents | Runs vector search, full-text search, and SQL filters in one query for RAG. |
| Query across your data lakes | Reads Iceberg, Hudi, Paimon, and Hive tables, plus external databases, in place without copying data. |
VeloDB Cloud runs on AWS, with storage and compute separated so you can scale each independently. Sign in at velodb.cloud to deploy a warehouse, or follow the Quick Start to load data and run your first query in about 10 minutes.
Core capabilities
- Real-time ingestion and updates. Stream data in with low latency and query it immediately. Primary-key upserts and deletes keep tables current as source data changes. Compute scales up and down on demand.
- Full-text and vector search. Run full-text search, vector search, and structured SQL filters in a single query. The
search()function uses an Elasticsearch-compatible syntax for phrase, wildcard, and regex queries. - AI agents. Connect AI agents to your data through the MCP Server, so an agent can query and reason over live data.
- Lakehouse and federation. Read and write open table formats and query external databases in place.
- Apache Doris compatibility. VeloDB Cloud speaks the MySQL protocol and is wire-compatible with Apache Doris SQL, so existing Doris clients, BI tools, and connectors work without changes.
How it works
VeloDB Cloud separates storage from compute. Object storage holds the full dataset at low cost, and one or more compute clusters query it. Clusters share the same storage but isolate their workloads, and each scales independently on demand.
Key concepts
| Concept | Definition |
|---|---|
| Organization | Your company or team's top-level account. Billing, resources, and data are isolated between organizations. |
| Warehouse | A logical unit that holds compute and storage. An organization can run many, isolated from each other, to separate business domains. |
| Cluster | A set of compute nodes inside a warehouse. Clusters share the warehouse's data but isolate workloads, and scale independently. |
| Storage | The shared object-storage layer that holds the full dataset for all clusters in a warehouse. |
Relationship to Apache Doris
VeloDB is the commercial company behind Apache Doris, founded in 2023 by the project's original team. VeloDB Cloud 26.0 is built on Apache Doris 4.0. For how VeloDB Cloud releases map to Apache Doris versions, see Version Mapping.
VeloDB Cloud is wire-compatible with Apache Doris SQL, so you can migrate data and queries between the two without rewriting them. Compared with self-managed Apache Doris, VeloDB Cloud adds:
- A managed control plane. No software to install, deploy, patch, or upgrade.
- A built-in console. SQL editor, monitoring, alerting, and cluster management in the browser.
- Elastic, usage-based compute. Clusters scale on demand, and you pay only for what you use.
- Enterprise security and compliance. SOC 2, encryption at rest and in transit, SSO, and a BYOC option that runs in your own cloud account.
- Managed reliability. Automated backups, high availability, and expert support.
Next step
Create a warehouse and run your first query with a 14-day free trial.