On the Performance of Three In-Memory Data Systems for On Line Analytical Processing

Spremljeno u:

Bibliografski detalji
Izdano u:	Informatica Economica vol. 21, no. 1 (2017), p. 5-15
Glavni autor:	Hrubaru, Ionut
Daljnji autori:	Fotache, Marin
Izdano:	INFOREC Association
Teme:	Servers Software upgrading Design Data processing Distributed processing Concurrency control Optimization techniques Queries Databases Big Data Workloads Product development Relational data bases Case studies
Online pristup:	Citation/Abstract Full Text Full Text - PDF
Oznake:	Dodaj oznaku Bez oznaka, Budi prvi tko označuje ovaj zapis!

Opis
Sažetak:	In-memory database systems are among the most recent and most promising Big Data technologies, being developed and released either as brand new distributed systems or as extensions of old monolith (centralized) database systems. As name suggests, in-memory systems cache all the data into special memory structures. Many are part of the NewSQL strand and target to bridge the gap between OLTP and OLAP into so-called Hybrid Transactional Analytical Systems (HTAP). This paper aims to test the performance of using such type of systems for TPCH analytical workloads. Performance is analyzed in terms of data loading, memory footprint and execution time of the TPCH query set for three in-memory data systems: Oracle, SQL Server and MemSQL. Tests are subsequently deployed on classical on-disk architectures and results compared to in-memory solutions. As in-memory is an enterprise edition feature, associated costs are also considered.
ISSN:	1453-1305 1842-8088
Digitalni identifikator objekta:	10.12948/issn14531305/21.1.2017.01
Izvor:	ABI/INFORM Global