Akumuli is a time-series database for modern hardware. It can be used to capture, store and process time-series data in real-time. The word "akumuli" can be translated from Esperanto as "accumulate".
- True column-oriented format (not PAX).
- Based on novel LSM and B+tree hybrid datastructure with multiversion concurrency control (no concurrency bugs, parallel writes, optimized for SSD and NVMe).
- Crash safety and recovery.
- Fast aggregation without pre-configured rollups or materialized views.
- Queries can be executed without decompressing the data.
- Fast compression algorithm (dictionary + entropy) with small memory overhead (about 2.5 bytes per element on appropriate data).
- Compressed in-memory storage for recent data.
- Can be used as a server application or an embedded library.
- Simple query language based on JSON and HTTP.
- Fast range scans and joins, read speed doesn't depend on database cardinality.
- Fast data ingestion over the network:
- 4.5M data points per second on 8-core Intel Xeon E5-2670 v2 (m3.2xlarge EC2 instance).
- 16.1M data points per second on 32-core Intel Xeon E5-2680 v2 (c3.8xlarge EC2 instance).
- Query results are streamed to client using the chunked transfer encoding of the HTTP protocol.
- Decompression algorithm and input parsers were fuzz-tested.
- Grafana datasource plugin.
- Fast and compact inverted index for series lookup.
|Storage engine features||Current version||Future versions|
|Inserts||In order||Out of order|
|Query language features||Current version||Future versions|
|Merge & aggregate||+||+|
|Group-aggregate & merge||-||+|
|Join & merge||-||+|
|Join & group-aggregate||-||+|
|Join & group-aggregate & merge||-||+|
|Filter by value||+||+|
|Filter & group-aggregate||+||+|
|Filter & join||+||+|
- You can find documentation here
- Installation & build instructions
- Getting started guide
- Writing data
Pre-built Debian/RPM packages for the following platforms are available via packagecloud:
- Ubuntu 14.04
- Ubuntu 16.04
- Debian Jessie
- Debian Stretch
- CentOS 7
Docker image is availabe through Docker Hub.
Tools for monitoring
Akumuli supports OpenTSDB telnet-style API for writing. This means that many collectors works with it
without any trouble, for instance
datasource plugin is availabe as well.