PaScal-- A New Parallel and Scalable Server IO Networking Infrastructure for Supporting Global Storage/File Systems in Large-size Linux Clusters

Grider, Gary; Chen, H.; Nunez, James; Poole, Steve; Wacha, R.; Fields, Parks; Martı́nez, R.; Martínez, Pilar Martínez; Khalsa, S.; Matthews, Adrian; Gibson, Garth A.

doi:10.1109/.2006.1629424

Cited by 10 publications

(5 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most networks are provisioned such that the client's bandwidth to the switch should be the throughput bottleneck of any parallel data transfer [11,12]. Unfortunately, when performing synchronized reads for data blocks across an increasing number of servers, a client may observe a TCP throughput drop of one or two orders of magnitude below its link capacity.…”

Section: Throughput Collapse Of Synchronized Readsmentioning

confidence: 99%

See 1 more Smart Citation

On application-level approaches to avoiding TCP throughput collapse in cluster-based storage systems

Krevat

Vasudevan

Phanishayee

et al. 2007

Proceedings of the 2nd International Workshop on Petascale Data Storage: Held in Conjunction With Supercomputing '07

View full text Add to dashboard Cite

show abstract

Section: Throughput Collapse Of Synchronized Readsmentioning

confidence: 99%

“…To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Supercomputing'07 Nov. [10][11][12][13][14][15][16]2007, Reno, NV Copyright 2007 ACM 978-1-59593-899-2/07/11 ...$5.00.…”

Section: Introductionmentioning

confidence: 99%

On application-level approaches to avoiding TCP throughput collapse in cluster-based storage systems

Krevat

Vasudevan

Phanishayee

et al. 2007

Proceedings of the 2nd International Workshop on Petascale Data Storage: Held in Conjunction With Supercomputing '07

View full text Add to dashboard Cite

show abstract

“…The first workflow is the QCD suite from USQCD. 11 For this workflow, the simulation component QCD uses POSIX as its default means of file I/O. However, QCD can allow for MPI-IO as its file I/O layer, as it is not dependent on POSIX for file I/O support.…”

Section: Usfd: Supporting Different I/o Semanticsmentioning

confidence: 99%

“…Data transferred between compute nodes and the storage system are routed through these IO nodes. This architecture allows the storage infrastructure to be shared by multiple supercomputers [11,18]. The disadvantage of this approach is that the additional cost of the network makes it much more expensive than a comparable amount of storage attached directly to the compute nodes as is the case in DIFS.…”

Section: Introductionmentioning

confidence: 99%

USFD: a unified storage framework for SOAR HPC scientific workflows

Mackey

Sehrish

Mitchell

et al. 2012

International Journal of Parallel, Emergent and Distributed Sys

View full text Add to dashboard Cite

Emerging scientific workflows in high performance computing (HPC) focus more on analysis rather than on simulation. Simulation output is so dense with information that copious amounts of analysis must be performed on a single output to understand the results of that simulation. We identify this repetitive analysis as a new application type, simulate once analyse repeatedly (SOAR) computing. Current scientific HPC, when extended to SOAR computing, results in excessive data migration between compute and storage resources. For a workflow bound by file I/O, a large data migration overhead is unacceptable. We propose a framework that uses a data-intensive storage cluster coupled with an interoperability layer, called unified storage framework designed (USFD).USFD is a better support SOAR HPC scientific workloads through enhanced file I/O support and co-located storage and analysis. In this work, we analyse the performance of USFD and other traditional HPC approaches for SOAR scientific workloads. Our results show that SOAR workflows using USFD complete 7.5 times faster over other approaches with quantum chromodynamic workflows and 4 times faster with FLASH workflows.

show abstract

“…The Quadrics network [20] uses a network operating system and specialized hardware to support highperformance data transfers. PaScal [16] is an I/O networking infrastructure which employs several novel technologies, including multi-level switch-fabric interconnection network, bandwidth on demand, and so forth. It has been implemented and evaluated on a 1024 node Linux cluster in LANL installed with the Panasas parallel filesystem.…”

Section: Introductionmentioning

confidence: 99%

Enhancing I/O throughput via efficient routing and placement for large-scale parallel file systems

Dillow

Shipman

Oral

et al. 2011

30th IEEE International Performance Computing and Communications Conference

View full text Add to dashboard Cite

As storage systems get larger to meet the demands of petascale systems, careful planning must be applied to avoid congestion points and extract the maximum performance. In addition, the large data sets generated by such systems makes it desirable for all compute resources to have common access to this data without needing to copy it to each machine. This paper describes a method of placing I/O close to the storage nodes to minimize contention on Cray's SeaStar2+ network, and extends it to a routed Lustre configuration to gain the same benefits when running against a center-wide file system. Our experiments using half of the resources of Spider -the center-wide file system at the Oak Ridge Leadership Computing Facility -show that I/O write bandwidth can be improved by up to 45% (from 71.9 to 104 GB/s) for a direct-attached configuration and by 137% (47.6 GB/s to 115 GB/s) for a routed configuration. We demonstrated up to 20.7% reduction in run-time for production scientific applications. With the full Spider system, we demonstrated over 240 GB/s of aggregate bandwidth using our techniques.

show abstract

PaScal-- A New Parallel and Scalable Server IO Networking Infrastructure for Supporting Global Storage/File Systems in Large-size Linux Clusters

Cited by 10 publications

References 6 publications

On application-level approaches to avoiding TCP throughput collapse in cluster-based storage systems

On application-level approaches to avoiding TCP throughput collapse in cluster-based storage systems

USFD: a unified storage framework for SOAR HPC scientific workflows

Enhancing I/O throughput via efficient routing and placement for large-scale parallel file systems

Contact Info

Product

Resources

About