Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

Apache Druid

druid.apache.org

  • Free plan
  • macOS
  • Linux
  • Web
  • API
  • Self-hosted
8.0 Club score Visit druid.apache.org

What Apache Druid is

Apache Druid is an open-source real-time analytics database for querying streaming and batch data at scale. It is designed for OLAP queries in milliseconds on high-cardinality datasets with billions to trillions of rows, and the project describes applications ranging from hundreds to 100,000 queries per second. Native Apache Kafka and Amazon Kinesis integrations support low-latency streaming ingestion and query-on-arrival. Druid organizes ingested data into compressed, indexed columns, with time indexing and dictionary and bitmap indexes. Queries can use Druid SQL or JSON-over-HTTP native queries, and joins are supported during ingestion or at query time. Its web console can load data, manage datasources and tasks, show server status and segments, and run queries. Extensions connect it to storage, databases, and formats including S3, HDFS, Azure, PostgreSQL, Avro, ORC, and Parquet. Druid is free under Apache License 2.0 and downloadable for self-hosting on Linux, macOS, and other Unix-like systems; Windows is not supported. The local quickstart requires at least 6 GiB of RAM and Java 17. Security controls are disabled by default, so production deployments need TLS, authentication, and authorization configured.

Who it suits

Druid suits teams building low-latency, high-concurrency analytics applications, especially where streaming data or instant visibility matters. It may also fit teams exploring data ad hoc or querying large datasets.

Good to know, and what to watch

  • Open-source and downloadable for self-hosting
  • Supports streaming and batch data
  • Provides Druid SQL and JSON-over-HTTP queries
  • Web console can manage data and run queries
  • Supports Kafka and Amazon Kinesis integrations
  • Windows is not supported
  • Local quickstart needs at least 6 GiB RAM and Java 17
  • Security controls are disabled by default

Apache Druid

Free

Open source

Getting started

  1. Visit the Apache Druid website
  2. Prepare a Linux, macOS, or other Unix-like system
  3. Provide at least 6 GiB of RAM and Java 17 for the local quickstart
  4. Download and run Druid for self-hosting
  5. Connect streaming, object-store, database, or file data sources
  6. Configure TLS, authentication, and authorization for production

Limits to check

Apache Druid is free at no cost and is downloadable for self-hosting. Windows is unsupported, the local quickstart requires at least 6 GiB of RAM and Java 17, and production security features require manual configuration.

The Club's review

Apache Druid is a serious choice for teams that need fresh analytical results from very large event streams and batch datasets. It suits user-facing dashboards, exploratory analysis, and high-concurrency applications more than ordinary transactional workloads. The trade-off is operational: you get an open-source distributed database, but you must run it and secure it yourself.

Overview

Druid is built for low-latency online analytical processing rather than general-purpose application storage. Its architecture separates ingestion, query, and orchestration services, allowing each layer to scale independently while deep storage supports larger deployments. The project describes millisecond OLAP queries on high-cardinality datasets containing billions to trillions of rows, with workloads ranging from hundreds to as many as 100,000 queries per second. Those targets make Druid compelling when response time and concurrent access matter more than a simple single-node setup.

The current stable release is Apache Druid 37.0.0, released May 8, 2026. The software and documentation use the Apache License, Version 2.0, and the project can run on commodity hardware or cloud environments including AWS, GCP, and Azure.

Key features

Streaming and batch ingestion

Native Apache Kafka and Amazon Kinesis integrations provide low-latency ingestion and query-on-arrival. Druid is designed for millions of events per second with guaranteed consistency, while also accepting batch data from object stores, databases, files, and other sources. That combination fits products that must expose new events quickly without maintaining separate systems for historical analysis.

Columnar indexing for fast OLAP

During ingestion, Druid columnarizes, time-indexes, dictionary-encodes, bitmap-indexes, and compresses data. These choices are aimed at scanning only the relevant columns and time ranges, which is why Druid is a stronger fit for aggregations and time-oriented analytics than for arbitrary record-by-record application queries.

SQL and native query APIs

Druid SQL gives analysts and applications a familiar interface, while JSON-over-HTTP native queries expose Druid-specific controls. Joins are supported both during ingestion and at query time. The web console can load data, manage datasources and tasks, show server status and segments, and run either query style, making it useful for administration and investigation as well as application development.

Integrations and resilience

Core extensions cover S3, HDFS, Google Cloud Storage, Azure, Kafka, Kinesis, Avro, ORC, Parquet, MySQL, and PostgreSQL, among other storage, format, and metadata options. Continuous backup, automated recovery, and multi-node replication are intended to provide durability and high availability. They also add planning and operational work that smaller teams may not want.

Security requirements

Security controls are disabled by default, so a production deployment must configure TLS, authentication, and authorization. Documented authentication extensions include HTTP Basic, LDAP, and Kerberos. This is manageable for an experienced platform team, but it is an important deployment task rather than a safe default.

Pricing

Apache Druid — free. The open-source analytics database is downloadable for self-hosting, with no software license charge. You supply the infrastructure and operations needed to run it. There are no paid tiers, seat quotas, trial periods, or renewal terms in this plan; commercial support is available from providers including Cloudera, Datumo, Deep.BI, Imply, and Rill Data.

Platforms

Druid is offered through an API, web console, Linux and macOS environments, and self-hosted deployment. The local quickstart supports Linux, Mac OS X, and other Unix-like systems, but not Windows. Running that quickstart requires at least 6 GiB of RAM and Java 17, so a small laptop may need preparation before development can begin.

Who it's for

Choose Druid for user-facing analytics, low-latency high-concurrency queries, instant visibility into streaming data, ad hoc exploration, or systems that combine live events with large historical stores. It is particularly attractive when Kafka or Kinesis is already central to the data pipeline and when independent scaling of ingestion and query capacity is valuable.

It is a poor fit for full-text search over text logs; the project says Druid is not commonly used for that job, even though it can ingest and analyze semi-structured JSON. Teams seeking a conventional transactional database or a managed service with minimal operations should also look elsewhere.

Pros and cons

  • Pro: Millisecond-scale OLAP is designed for datasets from billions to trillions of rows.
  • Pro: Kafka and Kinesis integrations support query-on-arrival and high-rate streaming ingestion.
  • Pro: Independent components, deep storage, replication, backup, and recovery support large, durable deployments.
  • Con: Self-hosting means your team handles infrastructure, scaling, upgrades, and configuration.
  • Con: TLS, authentication, and authorization require explicit production setup because security is off by default.
  • Con: Windows is not supported by the local quickstart, and the 6 GiB RAM plus Java 17 requirement raises the entry barrier.

Alternatives

Timeplus is worth choosing when you want an open-source, single-binary streaming SQL engine rather than Druid's distributed analytics database. Confluent Cloud is the better direction for fully managed Kafka clusters with autoscaling and managed connectors. Materialize suits teams wanting self-managed or cloud stream processing with a stated Community License memory and disk ceiling. Ververica Platform offers a community edition and a cloud option for teams focused on stream processing. Feldera is an alternative open-source streaming SQL system supporting a single-node deployment and broad connectors. Apache Beam is a programming model whose execution depends on the chosen runner and infrastructure. Apache Storm is another free, Apache-licensed streaming framework. kdb+ is the alternative to consider when its paid platform and personal-edition terms fit your use case.

Verdict

Apache Druid is a strong recommendation for engineering teams that need fast, concurrent analytics on continuously arriving and historical data, especially around Kafka or Kinesis. Its main reason to choose it is the combination of real-time ingestion, indexed columnar storage, and scale-out OLAP performance without a license fee. Look elsewhere if you need Windows support, built-in production security defaults, full-text log search, or a managed experience that removes distributed-system operations.

The verdict

Apache Druid is a self-hosted option for real-time analytics across streaming and batch data, with SQL and native query interfaces. Production use requires configuring security controls, and the local quickstart has stated system requirements.

Questions people ask

How much does Apache Druid cost?

The Apache Druid plan is free at no cost. It is open-source and licensed under Apache License, Version 2.0.

Which operating systems support Druid?

The quickstart supports Linux, macOS, and other Unix-like operating systems. Windows is not supported.

What query languages are available?

Druid provides Druid SQL and JSON-over-HTTP native queries. Joins can run during ingestion and at query time.

Which streaming systems integrate natively?

Apache Kafka and Amazon Kinesis provide native streaming ingestion, query-on-arrival, and low-latency data access.

What does the local quickstart require?

The local quickstart requires a machine with at least 6 GiB of RAM and Java 17.

Are security features enabled automatically?

No. Security controls are disabled by default, so production deployments must configure TLS, authentication, and authorization.

What the maker publishes

Architecture
Loosely coupled ingestion, query, and orchestration components with deep storage support scale-up and scale-out. source
Authentication options
Documented authenticator extensions include HTTP Basic authentication, LDAP, and Kerberos. source
Concurrency
Druid supports applications ranging from hundreds to 100,000 queries per second at consistent performance. source
Concurrency
The project describes Druid as supporting workloads from hundreds to tens of thousands of queries per second at consistent performance. source
Deployment
Druid can run on commodity hardware in *NIX environments and is designed to run in AWS, GCP, Azure, and other cloud environments. source
Extensions
Core extensions add support for storage, metadata stores, formats, authentication, and other capabilities; examples include S3, HDFS, Azure, Kafka, and PostgreSQL. source
Ingestion
Druid integrates natively with Apache Kafka and Amazon Kinesis for low-latency streaming ingestion and query-on-arrival. source
Integrations
Core extensions support systems and formats including S3, HDFS, Google Cloud Storage, Azure, Kafka, Kinesis, Avro, ORC, Parquet, MySQL, and PostgreSQL. source
Intended workloads
The FAQ recommends considering Druid for user-facing applications, low-latency high-concurrency queries, instant data visibility, ad hoc exploration, and streaming data. source
Latest release
The latest stable release is Apache Druid 37.0.0, released May 8, 2026. source
License
Apache Druid and its documentation are licensed under the Apache License, Version 2.0. source
Notable limitation
The FAQ says Druid is not commonly used for full-text search over text logs, though it is often used to ingest and analyze semi-structured data such as JSON. source
OLAP scale
Druid executes OLAP queries in milliseconds on high-cardinality datasets containing billions to trillions of rows. source
Operating systems
The quickstart supports Linux, Mac OS X, and other Unix-like operating systems; Windows is not supported. source
Purpose
Apache Druid is a high-performance real-time analytics database for sub-second queries on streaming and batch data at scale. source
Query languages
Druid supports both Druid SQL and JSON-over-HTTP native queries. source
Query performance
The project says Druid can execute OLAP queries in milliseconds over datasets with billions to trillions of rows. source
Reliability
Druid provides continuous backup, automated recovery, and multi-node replication for high availability and durability. source
Security
Druid security features are disabled by default and production deployments must configure TLS, authentication, and authorization. source
Security
Security features include TLS, authentication, and authorization, but they are disabled by default and the documentation says to configure them for production. source
SQL and joins
Druid provides a SQL API and supports joins during ingestion and at query time. source
Storage and indexing
Ingested data is columnarized, time-indexed, dictionary-encoded, bitmap-indexed, and compressed. source
Storage format
Druid automatically columnarizes, time-indexes, dictionary-encodes, bitmap-indexes, and compresses ingested data. source
Streaming
Native Apache Kafka and Amazon Kinesis integrations provide query-on-arrival, ingestion at millions of events per second, low latency, and guaranteed consistency. source

Checked against the maker's pages on 2 Oct 2026.