Open Source
Time Series Database


For Modern Hardware

Learn More

Akumuli Time Series Database


Akumuli is an open source Time Series Database available under Apache 2.0 License. The key goals of the project are efficiency and ability to work in stand-alone mode with little or no maintenance.

The database can scale up by utilizing multiple cores or CPUs. It's also can scale down and work on small and cheap VMs or in a "sidecar" container.

Write


Up to ~14 million writes per second on 32-core server.

Data ingestion is parallelized and optimized for SSD/NVMe.

State of the art compression, ~1GB/s encoding throughput per core.





Read


Column-oriented format.

Fast downsampling without pre-configured rollups.

Suitable for interactive applications.





Storage

Specialized
Based on B+tree
Crash safe
Space constrained

Performance

Written in C++
Small memory footprint
Multi-threaded
SSD/NVMe optimized

Queries

Downsample
Filter
Aggregate
Join

Integrations

Grafana plugin
Collectors via the OpenTSDB protocol

Download


Ubuntu PPA

Grafana Plugin

Docker images



License


Akumuli is available under Apache 2.0