Home-based Cooperative Cache for parallel I/O applications

Hwang, Inseok; Maeng, Seungryoul; Cho, Jungwan

doi:10.1016/j.future.2005.09.004

Cited by 3 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Because a grid network has a much lower bandwidth and much higher latency than a typical cluster network, the design aims to remove these potential performance bottlenecks and improve system scalability by eliminating redundant cached data, and simultaneously transfer blocks from multiple nodes to the requestor. The technique of cooperative caching has also been introduced to cluster computing by using a socalled home-based strategy wherein a strong consistency requirement necessitates tracking the locations of each cached copy of a block [27]. To avoid a bottleneck at the server, the home-based strategy uses a peer-to-peer distributed index, similar to the distributed hash table used to organize a structured peer-to-peer system.…”

Section: Other Related Workmentioning

confidence: 99%

Improving Networked File System Performance Using a Locality-Aware Cooperative Cache Protocol

Jiang

Zhang

Liang

et al. 2010

IEEE Trans. Comput.

View full text Add to dashboard Cite

Abstract-In a distributed environment the utilization of file buffer caches in different clients may greatly vary. Cooperative caching has been proposed to increase cache utilization by coordinating the shared usage of distributed caches. It allows clients that would more greatly benefit from larger caches to forward data objects to peer clients with relatively underutilized caches. To support such coordination, global cache utilization must be dynamically evaluated. This in turn requires an effective analysis of application data access patterns. Existing coordination protocols are demonstrably suboptimal in this respect, exhibiting inefficient memory utilization and undue interference among clients. We propose a locality-aware cooperative caching protocol, called LAC, that is based on analysis and manipulation of data-block reuse distance to effectively predict cache utilization and the probability of data reuse at each client. Using a dynamically adaptive synchronization technique, we keep local information up to date and consistently comparable across clients. The system is highly scalable in the sense that global coordination is achieved without centralized control. We have conducted thorough trace-driven simulation experiments to assess the performance differences between LAC and various existing protocols representative of the general class. Using a realistic and representative cost model, we show that the LAC protocol significantly and consistently outperforms existing cooperative caching protocols, demonstrating high and balanced utilization of caches across all clients. In our experiments LAC reduces block access time by up to 36%, with an average of 31%, over the system without peer cache coordination, and reduces block access time by up to 22%, with an average of 13%, over the best performer of the existing protocols.

show abstract

Section: Other Related Workmentioning

confidence: 99%

Improving Networked File System Performance Using a Locality-Aware Cooperative Cache Protocol

Jiang

Zhang

Liang

et al. 2010

IEEE Trans. Comput.

View full text Add to dashboard Cite

show abstract

“…Client caches also reduce disk accesses when servers have disk caches because it reduces server load. In the bibliography can be found several data cache proposed for distributed file systems ( [7], [8], [9], [10], [11], [12], [13]) and in a higher level, for example, for MPI-IO ( [14], [15]). In particular, for the first PVFS version has been implemented some prototype of data cache ( [16,17], [13]).…”

Section: Related Workmentioning

confidence: 99%

“…In the bibliography can be found several data cache proposed for distributed file systems ( [7], [8], [9], [10], [11], [12], [13]) and in a higher level, for example, for MPI-IO ( [14], [15]). In particular, for the first PVFS version has been implemented some prototype of data cache ( [16,17], [13]). Some of these works proposed cooperative caches ( [13] implements cooperative cache in the first PVFS version, [7], [8] implement cooperative cache in NFSv4 clients, [12] proposes it for Clusterfile).…”

Section: Related Workmentioning

confidence: 99%

“…In particular, for the first PVFS version has been implemented some prototype of data cache ( [16,17], [13]). Some of these works proposed cooperative caches ( [13] implements cooperative cache in the first PVFS version, [7], [8] implement cooperative cache in NFSv4 clients, [12] proposes it for Clusterfile). The performance improves in this kind of cache because it reduces servers' load and because the access time of another client's memory has faster response than that of a server's disk.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Client cache for PVFS2

Camacho

Nieto

Anguita

et al. 2010

2010 First International Conference on Parallel, Distributed and Grid Computing (PDGC 2010)

View full text Add to dashboard Cite

Engineering and scientific applications usually require to manage large quantities of data with different programs. The I/O demands of these applications get higher as they get larger because processor and memory speed increase. In order to improve performance and make programming much easier these applications can use parallel file systems. The relation between performance and easiness of programming can be additionally improved by using caches in clients. PVFS is a free parallel file system developed by a multi-institution team of parallel I/O, networking and storage experts. This work proposes a cache in the second version of PVFS in order to improve performance of data-intensive applications with several programs and to improve performance in a parallel program without modifying it to take advantage of the several cores in a processor. We analyze the performance of the cache implementation.

show abstract