Paper Info
Title
Elastic online analytical processing on RAMCloud
Abstract
A shared-nothing architecture is state-of-the-art for deploying a distributed analytical in-memory database management system: it preserves the in-memory performance advantage by processing data locally on each node but is difficult to scale out. Modern switched fabric communication links such as InfiniBand narrow the performance gap between local and remote DRAM data access to a single order of magnitude. Based on these premises, we introduce a distributed in-memory database architecture that separates the query execution engine and data access: this enables a) the usage of a large-scale DRAM-based storage system such as Stanford's RAMCloud and b) the push-down of bandwidth-intensive database operators into the storage system. We address the resulting challenges such as finding the optimal operator execution strategy and partitioning scheme. We demonstrate that such an architecture delivers both: the elasticity of a shared-storage approach and the performance characteristics of operating on local DRAM.
Year
DOI
Venue
2013
10.1145/2452376.2452429
EDBT
Keywords
Field
DocType
shared-nothing architecture,in-memory database architecture,analytical in-memory database management,elastic online analytical processing,remote dram data access,bandwidth-intensive database operator,performance characteristic,large-scale dram-based storage system,performance gap,data access,in-memory performance advantage,elasticity,performance,analytics
Dram,Data architecture,InfiniBand,Computer science,Computer data storage,Switched fabric,Online analytical processing,Data access,Database,Scalability
Conference
Citations 
PageRank 
References 
7
0.45
22
Authors
7
Name
Order
Citations
PageRank
Christian Tinnefeld1445.77
Donald Kossmann26220603.55
Martin Grund327016.74
Joos-Hendrik Boese4101.46
Frank Renkes5130.94
Vishal Sikka621215.63
Hasso Plattner777063.43