Lightweight Compression Methods for Storing Time Series in Databases and Their GPU Implementation

Nauka i Obrazovanie

V. V. Belous¹  N. V. Pivovarova¹ 

[1] Bauman Moscow State Technical University;

关键词: data compression;    database;    time series;   

DOI  :  10.7463/1014.0728338

来源: DOAJ

【 摘 要 】

Different lightweight methods for time series compression are discussed in the work. An approach to managing this compression is proposed, in the context of which a cascade of compression algorithms is considered; moreover these algorithms are chosen dynamically based on features of the data to be compressed. Implementation of these algorithms on graphics processing units (using OpenCL) allows a speed of operation on the order of 200 Gbit/s to be achieved. The methods proposed in the paper may find wide application in tasks of on-line storage and processing of telemetry data of different complex objects and systems.

【 授权许可】

Unknown