StarRocks: Features, Licence and Alternatives (2026)

Vendor

StarRocks project (Linux Foundation) / PhoenixAI, formerly CelerData

Price

Free

Platforms

Self-hosted (Linux, Docker, Kubernetes), PhoenixAI Cloud (managed BYOC), PhoenixAI Anywhere (self-managed commercial)

Licence

Apache License 2.0

Works with

Apache IcebergApache HudiDelta LakeApache PaimonApache HiveS3Google Cloud StorageAzure Blob StorageHDFSMySQL clientsElasticsearch

At a glance

SQL support	Standard SQL over the MySQL protocol
Storage and compute architecture	FE plus BE or CN nodes; shared-nothing (local disks) or shared-data (object storage with local cache)
Workload types	Sub-second and high-concurrency analytics, lakehouse queries, real-time updates
Integrations	Hive, Iceberg, Hudi, Delta Lake, Paimon, JDBC and Elasticsearch catalogs
Deployment	Self-hosted open source, or PhoenixAI Cloud and PhoenixAI Anywhere
Licensing	Apache License 2.0; commercial offerings priced by quote

Short answer

StarRocks suits teams that want fast, high-concurrency SQL analytics either on data loaded into StarRocks or on open table formats in place, with MySQL-protocol connectivity and an open source licence. Expect to run and tune the cluster yourself unless you buy the commercial service from PhoenixAI, whose prices are quote-only.

How we know: This profile is research-based: licence, governance, architecture, catalogs, materialized views and releases were checked against starrocks.io, the StarRocks documentation and release notes, and the StarRocks GitHub repository; the commercial vendor's products and pricing were checked on phoenixdata.ai, on 7 November 2026. We have not deployed StarRocks or measured its performance.

Overview

StarRocks is an open source analytical database whose repository describes it as an open query engine for sub-second analytics on and off the data lakehouse. It is licensed under the Apache License 2.0 and run as a Linux Foundation project (LF Projects). A cluster has two kinds of components: frontends (FE), which manage metadata, connections and query planning, and backends (BE) or stateless compute nodes (CN), which execute SQL. It is compatible with the MySQL protocol, so MySQL clients can connect.

StarRocks runs in a shared-nothing mode, with data on local BE disks for lowest latency, or a shared-data mode, with data in object storage such as S3, GCS, Azure Blob Storage or MinIO and hot data cached on compute nodes. External catalogs let it query Hive, Iceberg, Hudi, Delta Lake, Paimon, JDBC sources and Elasticsearch without loading. Commercial services come from PhoenixAI, the company formerly called CelerData, whose CTO chairs the StarRocks Technical Steering Committee.

Key features

Shared-nothing or shared-data architecture, with data in S3, GCS, Azure Blob Storage, MinIO or HDFS in shared-data mode
External catalogs for Hive, Iceberg, Hudi, Delta Lake, Paimon, JDBC and Elasticsearch, plus a unified catalog
Asynchronous materialized views with automatic, transparent query rewrite
MySQL protocol compatibility for clients and BI tools
Primary Key tables that support real-time updates and deletes
Stateless compute nodes that scale without rebalancing data in shared-data clusters

Pros and cons

Pros

Apache 2.0 licence under Linux Foundation project governance
Queries lakehouse formats in place as well as its own native tables
Materialized views with automatic query rewrite
MySQL protocol compatibility eases connection from existing tools

Cons

Commercial offerings have no published prices
Frequent patch releases, with documented upgrade cautions such as not downgrading below 4.0.6 after moving to 4.1
Running and tuning FE and BE or CN nodes yourself takes operational effort

Pricing

StarRocks is free and open source under the Apache License 2.0. The commercial offerings come from PhoenixAI (formerly CelerData): PhoenixAI Cloud, a managed bring-your-own-cloud service, and PhoenixAI Anywhere for self-managed VPC, on-premises or air-gapped Kubernetes. Pricing is quote-based and depends on workload, data scale, region and support tier; no list prices are published. Both offer a 30-day free trial with no credit card.

Pricing checked on the vendor's official page on 7 November 2026. Prices change; confirm before buying.

Who it is best for

Data teams that want fast, concurrent SQL analytics over both native tables and open lakehouse formats, with an open source licence and MySQL-compatible access.

Who should look elsewhere

Teams that want a fully serverless warehouse with published per-query prices: see BigQuery alternatives.
Single-user analysis of local files, where an embedded engine is simpler: see DuckDB.

Verdict

Our take

StarRocks is a credible open source option for teams that want low-latency analytics and lakehouse querying in one engine, backed by Linux Foundation governance. The trade-offs are the operational work of running the cluster and quote-only pricing for the commercial service, which recently changed its company name from CelerData to PhoenixAI.

Frequently asked questions

Is StarRocks open source?

Yes. StarRocks is licensed under the Apache License 2.0 and run as a Linux Foundation project. Commercial managed and self-managed offerings are sold separately by PhoenixAI.

What happened to CelerData?

CelerData, the main commercial company behind StarRocks, rebranded as PhoenixAI in 2026, according to its own about page. Its commercial products are now called PhoenixAI Cloud and PhoenixAI Anywhere.

Can StarRocks query Iceberg tables without loading them?

Yes. StarRocks documents external catalogs for Iceberg, Hudi, Delta Lake, Paimon and Hive, plus a unified catalog that treats several of them as one data source.

Sources

Checked November 2026.

How we research reviews and what star ratings mean: our editorial method.