Apache Druid

Overview

Apache Druid is an analytics database for teams that need to query large, changing datasets quickly. It is a strong fit for user-facing analytics and exploratory work where fresh data and high query concurrency matter; it is a less natural choice for full-text search or general-purpose database duties.

Druid’s ingestion, query, and orchestration components are loosely coupled, with deep storage supporting scale-up and scale-out deployments. That architecture can serve workloads ranging from hundreds to 100,000 queries per second, and the project describes millisecond OLAP queries across billions to trillions of rows. These are ambitious target capabilities, not a substitute for sizing a deployment to its workload. Druid 37.0.0, released May 8, 2026, is the latest stable release, under the Apache License, Version 2.0.

Key features

Streaming and batch analytics

Native Apache Kafka and Amazon Kinesis integrations support low-latency ingestion and query-on-arrival, including ingestion at millions of events per second with guaranteed consistency. Druid also handles batch data, so teams can analyze streams alongside data from object stores, databases, and files. This breadth makes it useful when an application needs recent events and historical context in the same analytical system.

Indexed storage and query interfaces

Druid automatically turns ingested data into compressed columnar storage, applying time indexes, dictionary encoding, and bitmap indexes. This is designed to make selective analytical queries fast, especially on high-cardinality data. Users can query through Druid SQL or native JSON-over-HTTP queries; SQL joins are supported during ingestion and at query time.

Operations, integrations, and security

The web console can load data, manage datasources and tasks, show server and segment status, and run both query types. Core extensions connect Druid to systems and formats including S3, HDFS, Google Cloud Storage, Azure, Kafka, Kinesis, Avro, ORC, Parquet, MySQL, and PostgreSQL. Continuous backup, automated recovery, and multi-node replication support availability and durability, though these capabilities still require a deployment designed and operated for production.

Security features are disabled by default. Production operators must configure TLS, authentication, and authorization; documented authenticator extensions include HTTP Basic, LDAP, and Kerberos. That flexibility is useful in managed environments, but it makes security setup an essential part of deployment rather than an optional finishing touch.

Pricing

Apache Druid: free. This open-source analytics database is downloadable for self-hosting, with real-time ingestion and SQL support. There is no paid Druid plan in the stated offering; teams supply and operate their own infrastructure. The Apache License, Version 2.0 applies to the software and documentation.

For a local quickstart, the machine needs at least 6 GiB of RAM and Java 17. The quickstart supports Linux, Mac OS X, and other Unix-like systems, but not Windows. The project points users to Slack and GitHub for help and names Cloudera, Datumo, Deep.BI, Imply, and Rill Data as commercial support providers.

Platforms

Druid runs on Linux, macOS, Unix-like systems, and self-hosted environments, and is designed for AWS, GCP, Azure, and other cloud environments. The quickstart does not support Windows. API and web-console access provide additional ways to work with the system, but a local deployment still needs Java 17 and at least 6 GiB of RAM.

Who it's for

Consider Druid for streaming analytics, low-latency high-concurrency queries, applications that need immediate data visibility, and ad hoc exploration. It is particularly compelling when teams need to ingest and query structured or semi-structured data at scale. For full-text search over text logs, it is not a common choice; its ability to ingest and analyze semi-structured formats such as JSON does not make it a search engine.

Pros and cons

Fast analytics at substantial scale: Druid is designed for millisecond OLAP queries over very large datasets and high query volumes, making it suited to interactive and user-facing workloads.

Fresh data with broad ingestion: Kafka and Kinesis integrations enable low-latency streaming, while batch sources let teams combine event streams with stored data.

Operational flexibility: Loosely coupled components, deep storage, backup, recovery, and replication support distributed deployments, but operators must manage the system and configure production security themselves.

Local setup has real prerequisites: The quickstart requires Java 17 and at least 6 GiB of RAM, and Windows is unsupported.

Not aimed at full-text search: Teams searching text logs should choose a tool built for that workload rather than treating Druid’s semi-structured data support as a substitute.

Alternatives

Streaming Analytics Software, Columnar Databases, OLAP Databases, OLAP Software, and Database Software are useful category starting points.

Timeplus is worth considering for a lightweight streaming SQL engine: its open-source Proton plan is a single binary under 500MB, and the service also offers a free trial.

Confluent Cloud suits teams seeking managed, serverless Kafka resources, autoscaling, and more than 80 managed connectors, with a free Basic plan and free trial.

Materialize offers a self-managed Community License capped at 24 GiB of memory and 48 GiB of disk, or a Cloud Capacity plan priced at 1.50 USD; consider it when those deployment and capacity terms fit.

Apache Storm is a free, open-source option for teams looking for a distributed stream-processing system rather than Druid’s analytics-database focus.

Feldera offers an open-source edition with full SQL support and all connectors in a single-node, single-container setup; it is a candidate when that model suits the workload.

Apache Beam provides an open-source programming model; execution costs depend on the runner and infrastructure selected.

Ververica Platform has a free Community Edition that may require a license and may restrict use; check its terms if those conditions matter.

RisingWave offers self-managed Apache 2.0 open source without feature gates, license fees, or usage restrictions, for teams prioritizing those terms.

Verdict

Choose Apache Druid when your team needs fast, highly concurrent analytics across streaming and batch data and can operate a self-hosted distributed system. Its combination of query-on-arrival, indexed columnar storage, and open-source licensing is compelling for that job. Look elsewhere if you need full-text log search, Windows quickstart support, or a deployment that avoids security and operations work.