#5353rd of 204 in Backends, databases and hosting
starrocks
Distributed vectorized SQL engine for real-time and data lake analytics
- Stars
- 12.3k
- License
- Apache-2.0
- Last commit
- Nov 2026
- Last release
- Nov 2026
- Language
- Java
11,535 on 25 Apr 202612,264 on 14 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
StarRocks is an MySQL-protocol-compatible analytical database with a cost-based optimizer and a vectorized execution engine, written in Java (frontend) and C++ (backend). It supports primary-key upserts and deletes, materialized views, and direct queries over Hive, Iceberg, Delta Lake and Hudi tables without importing. Since 3.0 it also offers a shared-data architecture. It is a general OLAP database, not an AI-specific project.
Who it is for: Data engineers running real-time or lakehouse analytics on large datasets
Strengths
- MySQL protocol and ANSI SQL, so existing clients and BI tools connect directly
- Queries Hive, Iceberg, Delta Lake and Hudi tables in place, without importing
- Primary-key upsert/delete plus auto-refreshed materialized views chosen at query time
- Shared-data architecture available since 3.0; per-tenant resource isolation
Weaknesses
- Cluster of separate Frontend and Backend node types, heavier than a single-binary database
- Speed claims (3x, 5-10x) come from vendor benchmarks, not reproduced in the README
- No AI, LLM or vector features described; it is an analytics database
What it needs
- Compose
What changed
- 4.0.17 ↗ (opens in a new tab)Security fix
“Security vulnerabilities fixed by bumping dependencies: ZooKeeper to 3.8.7 and 3.9.6, the broker Tomcat to 9.0.121, BouncyCastle to 1.86, jackson to 2.21.7, zstd-jni to 1.5.7-15, parquet to 1.18.1, a…”
- 4.0.16 ↗ (opens in a new tab)Security fix
“Addressed security vulnerabilities (CVE): upgraded Apache Ranger to 2.9.0 and Bouncy Castle to 1.85, Apache HttpClient5 to 5.6.4 and HttpCore to 5.4.3, Netty to 4.1.137.Final, ZooKeeper to 3.8.7 and …”
- 4.1.7 ↗ (opens in a new tab)Security fix
“Bumped pprof to release/20261012. #80439”
From the release notes. No flag means the note does not say so, not that the upgrade is safe.
Known problems
Open now, reported in the last 90 days
- Shared-data Storage Volume on EKS falls back to EC2 instance role instead of IRSA (starlet v4.1.1 -> v4.1.4 regression) ↗ (opens in a new tab)4 reactions
- Column statistics are UNKNOWN for queries that read a JSON subfield (4.1.6) ↗ (opens in a new tab)3 reactions
- [Bug] Completed transactions retain tablet writing-size map entries ↗ (opens in a new tab)1 reactions
- [Bug] "io ranges are overlapped" reading the newest partitions of a shared-data Primary Key table when `io_coalesce_lake_read_enable=true` ↗ (opens in a new tab)1 reactions
- Query Queue V2 does not enforce resource group concurrency_limit: bug or intended behavior? ↗ (opens in a new tab)1 reactions
Reported, then fixed by a linked change
- Shared-data: Workload Identity for ADLS2 storage volume is silently converted to Managed Identity, breaking authentication on CN ↗ (opens in a new tab)4 reactions · closed 22 Sept 2026Fixed by [BugFix] Preserve ADLS2 workload identity in storage volumes ↗ (opens in a new tab)
- [Bug] Cluster snapshots fail with Azure Blob Storage (TYPE = AZBLOB/ADLS2) - UnsupportedFileSystemException ↗ (opens in a new tab)3 reactions · closed 15 Jan 2026Fixed by [Enhancement] Map azblob/adls2 paths to wasb/abfs with endpoint injection ↗ (opens in a new tab)
- [Bug] Iceberg dataFileCache can serve permanently stale partial data, causing silent wrong query results ↗ (opens in a new tab)2 reactions · closed 25 Mar 2026Fixed by [BugFix] Fix partial manifest cache writes in Iceberg ManifestReader ↗ (opens in a new tab)
- [Bug] BE Segfault in Parquet Page Index Filter (GCS/Iceberg) ↗ (opens in a new tab)2 reactions · closed 27 Jan 2026Fixed by [BugFix] Segfault in Parquet Page Index Filter when null_counts empty ↗ (opens in a new tab)
- Missing cleanup in handleCompactionFinished can cause permanent compaction exclusion ↗ (opens in a new tab)2 reactions · closed 24 Dec 2025Fixed by [BugFix] Clear startup lake compaction txn map on txn finalization ↗ (opens in a new tab)
Titles are the reporters' own words, taken from the project's GitHub issues and ranked by reactions. We have not tried to reproduce any of them.
Also in Backends, databases and hosting
See all 204| Rank | Project | Score |
|---|---|---|
| 1 | bunJavaScript and TypeScript runtime with bundler, test runner and package manager | 94 out of 100 |
| 2 | coolifySelf-hosted Heroku and Vercel alternative that deploys apps and databases over SSH | 93 out of 100 |
| 3 | strapiHeadless CMS that generates REST and GraphQL APIs from content models | 92 out of 100 |
| #4 | traefikHTTP reverse proxy and load balancer that auto-discovers routes from orchestrators | 92 out of 100 |
| #5 | appwriteBackend with auth, databases, storage, functions and an MCP server for agents | 92 out of 100 |