KairosDB
| Developer(s) | KairosDB |
|---|---|
| Initial release | August 21, 2013 |
| Stable release | v1.1.3
/ January 3, 2017.[1] |
| Written in | Java |
| Engine | |
| Operating system | Cross-platform |
| Type | Time series database |
| License | Apache License 2 |
| Website | www |
Search KairosDB on Amazon.
KairosDB is a high-performance[2] open source time series database[3] designed for storing application, system[4], and IOT[5] metrics.
Design[edit]
KairosDB is a stand-alone service built in Java. Metrics[6] are the fundamental unit of this time series database. A metric is a named set of data points with an optional set of tags. Data points consist of a timestamp and a value. Metrics can be grouped and filtered by tags.
KairosDB is architected with five main components; an ingest/query interface, a data store, an event bus, groupers, and aggregators. It is built with a pluggable interface leveraging Guice Modules so each of the five pieces can be customized
Ingest/Query Interface[edit]
Metrics are ingested via a Telnet or REST API. KairosDB is designed for a high-performance ingestion rate.
Queries are done via a REST API. Data is retrieved from the data store, filtered and grouped, and then run through zero or more aggregators. Data is returned in JSON format.
Data Store[edit]
The data store is the storage for the metrics. KariosDB ships with two default data store modules H2 and Cassandra. ScyllaDB can also be used as a data store since it is API compatible with Cassandra[7][8]
Event Bus[edit]
A modified version of the EventBus provides the ability to add data filters and listeners.
Groupers[edit]
Metrics returned from a query can be grouped into buckets of data. For example, metrics can be separated by host, data center, day of the week, data point value, etc.
Aggregators[edit]
Queried metrics are run through a pipeline of aggregators. Each aggregator works on the output of the previous aggregator. Aggregators manipulate data and can do such oprations as calculating a max, min, average, standard deviation, sum, etc. A set of predefined aggregators ship with KariosDB.
History[edit]
KairosDB was built in 2013 by Brian Hawkins and Jeff Sabin while working at Proofpoint. They created KairosDB because commercial offerings were too expensive for the amount of data Proofpoint wanted to store, and open source alternatives were difficult to manage, scale and did not support the precision of time needed.
Philosophy[edit]
You don't know which metrics you need until you need them. So add all your metrics. This requires a time series database that can handle a high ingestion load. Metrics can have a TTL set or be rolled-up later to save store space.
Etymology[edit]
Kairos is a Greek word indicating an opportune time. KairosDB is a play on words since it stores time series data that may be needed for analysis at the right time.
References[edit]
- ↑ "Version 1.1.3 Release· kairosdb/kairosdb". GitHub. KairosDB. Retrieved 2017-07-10.
- ↑ Goldschmidt, Thomas; Jansen, Anton; Koziolek, Heiko; Doppelhamer, Jens; Pei Breivold, Hongyu. "Scalability and Robustness of Time-Series Databases for Cloud-Native Monitoring of Industrial Processes" (PDF). Heiko Koziolek. Industrial Software Systems. Retrieved 11 July 2017.
- ↑ Acreman, Steven (2016-08-28). ""Top 10 Time Series Databases Database". Outlyer.
- ↑ "vCloud Architecture Toolkit (vCAT) Blog". vmware.com Blogs. Retrieved 15 July 2017.
- ↑ "VMware vCloud Director Virtual Machine Metric Database - vCloud Architecture Toolkit (vCAT) Blog - VMware Blogs". blogs.vmware.com. Retrieved 2017-07-25.
- ↑ Valsorda, Filippo. "UNDERSTANDING METRICS IN THE AGE OF THE TSDB". FILIPPO.IO. Retrieved 15 July 2017.
- ↑ "Documentation". docs.scylladb.com. Retrieved 2017-07-25.
- ↑ "9 Steps for Building a Time-Series Solution with Scylla and KairosDB". ScyllaDB. 2017-05-24. Retrieved 2017-07-25.
This article "KairosDB" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:KairosDB. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
