Exploring I/O virtualization data paths for MPI applications in a cluster of VMs: A networking perspective

Nanos, A; Goumas, G; Koziris, N

dc.contributor.author	Nanos, A	en
dc.contributor.author	Goumas, G	en
dc.contributor.author	Koziris, N	en
dc.date.accessioned	2014-03-01T02:53:16Z
dc.date.available	2014-03-01T02:53:16Z
dc.date.issued	2011	en
dc.identifier.issn	03029743	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/36206
dc.subject.other	Data paths	en
dc.subject.other	Evaluating method	en
dc.subject.other	High-performance computing	en
dc.subject.other	MPI applications	en
dc.subject.other	Network behaviors	en
dc.subject.other	Network configuration	en
dc.subject.other	Network devices	en
dc.subject.other	Network load	en
dc.subject.other	Network virtualization	en
dc.subject.other	Optimized solutions	en
dc.subject.other	Performance bottlenecks	en
dc.subject.other	Scientific applications	en
dc.subject.other	Virtualizations	en
dc.subject.other	Virtualized environment	en
dc.subject.other	Behavioral research	en
dc.subject.other	Cluster computing	en
dc.subject.other	Computer software selection and evaluation	en
dc.subject.other	Message passing	en
dc.subject.other	Network performance	en
dc.subject.other	Optimization	en
dc.subject.other	Parallel architectures	en
dc.subject.other	Virtual reality	en
dc.subject.other	Benchmarking	en
dc.title	Exploring I/O virtualization data paths for MPI applications in a cluster of VMs: A networking perspective	en
heal.type	conferenceItem	en
heal.identifier.primary	10.1007/978-3-642-21878-1_82	en
heal.identifier.secondary	http://dx.doi.org/10.1007/978-3-642-21878-1_82	en
heal.publicationDate	2011	en
heal.abstract	Nowadays, seeking optimized data paths that can increase I/O throughput in Virtualized environments is an intriguing task, especially in a high-performance computing context. This study endeavors to address this issue by evaluating methods for optimized network device access using scientific applications and micro-benchmarks. We examine the network performance bottlenecks that appear in a Cluster of Xen VMs using both generic and intelligent network adapters. We study the network behavior of MPI applications. Our goal is to: (a) explore the implications of alternative data paths between applications and network hardware and (b) specify optimized solutions for scientific applications that put pressure on network devices. To monitor the network load and the applications' total throughput we build a custom testbed using different network configurations. We use the Xen bridge mechanism and I/O Virtualization techniques and examine the trade-offs. Preliminary results show that a combination of these techniques is essential to overcome network virtualization overheads and achieve near-native performance. © 2011 Springer-Verlag Berlin Heidelberg.	en
heal.journalName	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)	en
dc.identifier.doi	10.1007/978-3-642-21878-1_82	en
dc.identifier.volume	6586 LNCS	en
dc.identifier.spage	665	en
dc.identifier.epage	671	en