Authors: Bing Zhang; Brandon Ross; Tevfik Kosar
Addresses: Department of Computer Science and Engineering, University at Buffalo (SUNY), Buffalo, New York, 14260, USA ' Department of Computer Science and Engineering, University at Buffalo (SUNY), Buffalo, New York, 14260, USA ' Department of Computer Science and Engineering, University at Buffalo (SUNY), Buffalo, New York, 14260, USA
Abstract: Due to the diverse approaches and characteristics of scientific research, valuable resources within most important scientific disciplines exist in heterogeneous storage systems in distributed environments as they are often created and maintained by different information providers. Considering more than half of all data access operations in traditional storage systems are metadata access operations, it is important to design a framework that can effectively query and access valuable metadata information in distributed environments. In this paper, we present a highly efficient caching and prefetching mechanism tailored to reduce metadata access latency and improve responsiveness in wide-area data transfers. We designed dynamically provisioned parallel TCP streams and non-blocking concurrent in-memory cache for the system performance. We have implemented these mechanisms in a cloud-hosted metadata retrieval, caching, and prefetching system called directory listing service (DLS) and have evaluated its performance on both local area and wide area settings.
Keywords: metadata retrieval; wide-area transfers; cloud-hosted data caching; prefetching; software as a service; SaaS; distributed metadata access; cloud computing; access latency; access responsiveness.
International Journal of Big Data Intelligence, 2015 Vol.2 No.3, pp.183 - 200
Available online: 11 Jul 2015 *Full-text access for editors Access for subscribers Purchase this article Comment on this article