Public Repository

Last pushed: 8 hours ago
Short Description
Time-series database for the modern hardware.
Full Description

Github repo -
Only stable builds.

Getting started.

Pull latest stable image:

> sudo docker pull akumuli/akumuli:0.7.45-skylake

You should have a directory for the database files e.g. /tmp/akumuli. The database files will be placed there.
Also, you should provide some configuration details for the database. First parameter is a nvolumes (default value is 4). It should be used to set the number of volumes that database should contain. The second parameter is a volume_size. It can be used to change the size of each volume (default value is 4GB). This parameter can't be larger than 4GB.

To run the container use this command:

> sudo docker run -d -p 8181:8181 -p 8282:8282 -p 8383:8383 -p 4242:4242 \
       -v /tmp/akumuli:/Akumuli -e nvolumes='16' \
       -e volume_size='1GB' akumuli/akumuli:0.7.45-skylake

The database volumes will be created on the first start. Now you can check that the database is running:

> curl -XGET "http://localhost:8181/api/stats"

You should see the JSON output with the list of available volumes.

Alternatively, you can use this Helm chart for Kubernetes.


The repository contains two types of tags:

  • X.X.XX-skylake is an optimized version of the image. It's built using the -march=corei7-avx option that allows compiler to use various SIMD instructions and POPCNT instruction which is required for the compression algorithm. Because of this, it may not work on old hardware.
  • X.X.XX-generic is also optimized but for older architectures. It's built using the -mtune=generic option. This image should work on any 64bit x86 compatible CPU but it couldn't get performance advantage on modern hardware.


Akumuli follows semver policy. Image tag starts with the version number (e.g 0.7.44) followed by the build type (skylake or generic).

Docker Pull Command