PROVIDER Β· Real-time data warehouse and OLAP engine
Apache Doris
Apache Doris is the Chinese open source MPP OLAP engine from Baidu's Palo, MySQL protocol compatible, where materialised views and Multi-Catalog unify real-time reporting with lake analytics in one engine.
Pros
Cons
Product Information
Detailed Notes
Overview
Apache Doris descends from Baidu's 2018 Palo project and became an Apache top-level project in 2022, using an MPP architecture with columnar storage and MySQL protocol compatibility, among the most used open source OLAP projects with a China-centred community.
For Backend & Database work the advantage is one engine covering reporting and real-time pipelines, since Kafka streams write in real time alongside batch loads from Hive or lake data, materialised views rewrite queries automatically, and MySQL protocol lets existing BI tools and clients connect with low learning and migration cost.
Assess scale and governance, because single-cluster capacity depends on node sizing and shard design with very large deployments needing planning, community and enterprise editions differ on security, multi-tenancy, and operations tooling, and overlap with StarRocks means comparing specific version features and community activity.
Core Strengths
- MPP columnar architecture reaches sub-second aggregation over hundreds of millions of rows
- MySQL protocol compatibility connects existing BI tools and clients directly
- Materialised views rewrite queries transparently to accelerate frequent reports
- Kafka real-time writes plus Hive and lake federation unify batch and streaming
Product Ecosystem
Coupled and separated compute
Since 3.x a separated form keeps compute stateless and elastic with data on object storage.
Data lake and federation
Multi-Catalog queries Hive, Iceberg, Hudi, and Paimon directly.
Query acceleration
Materialised views, runtime filters, vectorised execution, and a cost-based optimiser deliver high-concurrency low latency.
Drawbacks
- Very large clusters still need sharding and node planning expertise
- Enterprise security, multi-tenancy, and tooling are commercial with community limits
- Heavy overlap with StarRocks demands comparing concrete version features
- Overseas community and English material are smaller than international projects like ClickHouse
Use Cases
- Real-time reporting and BI dashboards (β β β β β ) - Sub-second queries with materialised views
- Unified batch and streaming warehousing (β β β β β ) - Kafka plus offline loads
- Unified lake analytics (β β β β ) - Multi-Catalog federation
- Multinational overseas deployment (β β β ) - Smaller English ecosystem
Pricing
| Plan | Price | Scope |
|---|---|---|
| Apache Doris open source | $0 | Apache 2.0, self-managed |
| SelectDB Enterprise | Per node and capability | Security, multi-tenancy, support |
| SelectDB Cloud | Compute and storage usage | Managed with elastic scaling |
The open source edition is free with cost in servers and staff, typically three BE nodes minimum for replicas, while enterprise and cloud plans bill per node or usage mainly for security compliance, multi-tenancy, and tooling aligned with real data volume and concurrency.
FAQ
-
Doris or StarRocks? They share lineage and are highly similar, so choose Doris for Apache governance and vendor neutrality and StarRocks for optimiser maturity and commercial support, deciding by version features and benchmarks.
-
Can it replace ClickHouse? Doris fits better when joins, updates, and unified analytics matter, while ClickHouse remains strong for extreme single-table aggregation and very large detail.
-
What concurrency does it handle? Thousands of QPS of dashboards work with sensible sharding, and higher concurrency needs more frontend and backend nodes with materialised views and caching.