API Latest Releases: Java Core, C++ Core, Python, Memory, Pig, Hive,

Apache DataSketches GitHub Components

Our library is made up of components that are partitioned into GitHub repositories by language and dependencies. The dependencies of the core components are kept to a bare minimum to enable flexible integration into many different environments. Meanwhile, the Hive and Pig components, for example, have major dependencies on those envionments.

If you have a specific issue or bug report that impacts only one of these components please open an issue on the respective component. If you are a developer and wish to submit a PR, please choose the appropriate repository.

List of Component Repositories (Explained below)

Repository URL
Java Core https://github.com/apache/datasketches-java
C++ Core https://github.com/apache/datasketches-cpp
Hive Adaptor https://github.com/apache/datasketches-hive
Pig Adaptor https://github.com/apache/datasketches-pig
PostgreSQL Adaptor https://github.com/apache/datasketches-postgresql
Memory https://github.com/apache/datasketches-memory
Characterization https://github.com/apache/datasketches-characterization
Website https://github.com/apache/datasketches-website
Vector (Experimental) https://github.com/apache/datasketches-vector
Server (Under Development) https://github.com/apache/datasketches-server
Reserved for future use https://github.com/apache/datasketches

Core Algorithms

If you like what you see give us a Star on one of these two sites!

  • Java (Versioned, Apache Released) This is the original and the most comprehensive collection of sketch algorithms. It has a dependence on the Memory component and the Java Adaptors have a dependence on this component.

  • C++/Python (Versioned, Apache Released) This is newer and provides most of the major algorithms available in Java. Our C++ adaptors have a dependence on this component. The Pybind adaptors for Python are included for all the C++ sketches.


Adapters integrate the core components into the aggregation APIs of specific data processing systems. Some of these adapters are available as part of the library, other adapters are directly integrated into the target data processing application.

Java Adaptors

C++ Adaptors

  • PostgreSQL (Versioned, Apache Released) This site provides the postgres-specific adaptors that wrap the C++ implementations making them available to the PostgreSQL database users. PostgreSQL users should download the PostgreSQL extension from pgxn.org. For examples refer to the README on the component site.

Other Components

  • Memory: (Versioned, Apache Released) This is a low-level library that enables fast access to off-heap memory for Java.
  • Characterization: This is a collection of Java and C++ code that we use for long-running studies of accuracy and speed performance over many different parameters. Feel free to run these tests to reproduce many of the graphs and charts you see on our website.
  • Vector (Experimental): This component implements the Frequent Directions Algorithm [GLP16]. It is still experimental in that the theoretical work has not yet supplied a suitable measure of error for production work. It can be used as is, but it will not go through a formal Apache Release until we can find a way to provide better error properties. It has a dependence on the Memory component.
  • Website: This repository is the home of our website and is constantly being updated with new material.
  • Server (Under Development)

Deprecated Components

The code in these components are no longer maintained and will eventually be removed.


This is a new repository dedicated to sketches designed to be run in a mobile client, such as a cell phone. It is still in development and should be considered experimental.


This repository is an experimental staging area for code that will eventually end up in another repository. This code is not versioned.


Demos, command-line access, characterization testing and other code not related to production deployment.

This code is offered “as is” and primarily as a reference so that users can understand how some of the performance characterization plots were obtained. This code has few unit tests, if any, and was never intended for production use. Nonetheless, some folks have found it useful. If you find it useful, go for it. This code is not versioned.

Sketches-misc Packages Package Description
org.apache.datasketches Utility functions used by the sketches-misc packages
org.apache.datasketches.cmd Support for Command Line functions Being Redesigned
org.apache.datasketches.demo Simple demo for brute-force vs Theta and HLL sketches Will be superceded by Command Line functions
org.apache.datasketches.quantiles Utility for computing & printing space table for Quantiles Sketches (only in the test branch)
org.apache.datasketches.sampling Benchmarks and Entropy testing for sampling sketches


This is the parallel characterization repository with a parallel objective to the Java characterization repository.


This repository is an experimental staging area for C++ code that will eventually end up in another repository.

Command-Line Tool

These repositories provide a command-line tool that provides access to the following sketches:

  • Frequent Items
  • HLL
  • Quantiles
  • Reservoir Sampling
  • Theta Sketches
  • VarOpt Sampling

This tool can be installed from Homebrew.