A correlation-aware data placement strategy for key-value stores
Technical Report DI-CCTC-10-08, Centro de Ciências e Tecnologias de Computacão, Universidade do Minho
A correlation-aware data placement strategy for key-value stores
Technical Report DI-CCTC-10-08, Centro de Ciências e Tecnologias de Computacão, Universidade do Minho
AuthorsRicardo Vilaça, Rui Oliveira and José PereiraAbstractKey-value stores hold the bulk of the data produced by the unprecedented activity of social networking applications. Their scalability and availability requirements often outweigh sacrificing richer data and processing models, and even elementary data consistency. In this report we exploit arbitrary data relations easily expressed by the application to foster data locality and improve the performance of complex queries common in social network read-intensive workloads. To this end, we present the prototype of an elastic key-value data store embodying a novel data placement strategy based on multidimensional locality-preserving mappings. The system is built on the basis of an efficient peer-to-peer overlay, provides atomic access to tuples and flexible data replication. We evaluate different data placement strategies under the workload of a typical social network application and show that the proposed correlation-aware data placement offers a major improvement on the system's overall response time and network requirements. The elasticity of the system is also put under test by measuring the impact of a significant change on the number of nodes of the system.Full TextPDF |