Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199368
Maozhen Li, P. V. Santen, D. Walker, O. Rana, M. Baker
Grid is computer-based infrastructure that provides dependable, consistent, pervasive access to distributed resources. Built on top of a Grid, a Semantic Grid is a service-oriented infrastructure that provides a range of computation, information and knowledge services. A purpose of a Grid portal is to provide easy and seamless access to Grid heterogeneous resources and services through a Web-based user interface. This paper presents PortalLab, a Web Services oriented toolkit for designing, integrating and building Semantic Grid portals. Portals built from PortalLab are composed from a collection of reusable Web Services oriented portlets that are themselves semantic Grid services. Each portlet has a WSDL interface and a semantic registry defined in a domain ontology repository. The use of software agents assists end users in formulating domain problems, searching possible solutions (solvers) and submitting user tasks to the Grid. Multiple agents work in a peer-to-peer environment to allow users to access federated Grid services across different domains to improve fault tolerance and quality of service in user job submission and execution on the Grid. Since portlets are context independent, a PortalLab portal provides the ability to interoperate with different Grid systems at a portal level.
{"title":"PortalLab: a web services toolkit for building semantic grid portals","authors":"Maozhen Li, P. V. Santen, D. Walker, O. Rana, M. Baker","doi":"10.1109/CCGRID.2003.1199368","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199368","url":null,"abstract":"Grid is computer-based infrastructure that provides dependable, consistent, pervasive access to distributed resources. Built on top of a Grid, a Semantic Grid is a service-oriented infrastructure that provides a range of computation, information and knowledge services. A purpose of a Grid portal is to provide easy and seamless access to Grid heterogeneous resources and services through a Web-based user interface. This paper presents PortalLab, a Web Services oriented toolkit for designing, integrating and building Semantic Grid portals. Portals built from PortalLab are composed from a collection of reusable Web Services oriented portlets that are themselves semantic Grid services. Each portlet has a WSDL interface and a semantic registry defined in a domain ontology repository. The use of software agents assists end users in formulating domain problems, searching possible solutions (solvers) and submitting user tasks to the Grid. Multiple agents work in a peer-to-peer environment to allow users to access federated Grid services across different domains to improve fault tolerance and quality of service in user job submission and execution on the Grid. Since portlets are context independent, a PortalLab portal provides the ability to interoperate with different Grid systems at a portal level.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"PP 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126354267","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199433
Marc-Laurent Aird, Walter Barbera-Medina, J. Padget
The MONET project is a two-year investigation into mathematical web services funded by the European Commission, as part of the Information Society Technologies (IST) Programme of the Fifth Framework. The project started on 1st April 2002. The long-term aim (beyond this project) is to weave together the emerging world of web services-the semantic web-and grid computing, in that we aim to deliver sophisticated mathematical problem analysis and the code to compute the answers, for which grid services will be required, in a common open agent-based framework for the description and provision of web-based mathematical services. The short-term aim of this project is proof of concept. That is to demonstrate a framework consisting of a set of example mathematical web services together with a brokering platform used to discover and invoke these services given a query by a user to solve a particular problem.
{"title":"MONET: service discovery and composition for mathematical problems","authors":"Marc-Laurent Aird, Walter Barbera-Medina, J. Padget","doi":"10.1109/CCGRID.2003.1199433","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199433","url":null,"abstract":"The MONET project is a two-year investigation into mathematical web services funded by the European Commission, as part of the Information Society Technologies (IST) Programme of the Fifth Framework. The project started on 1st April 2002. The long-term aim (beyond this project) is to weave together the emerging world of web services-the semantic web-and grid computing, in that we aim to deliver sophisticated mathematical problem analysis and the code to compute the answers, for which grid services will be required, in a common open agent-based framework for the description and provision of web-based mathematical services. The short-term aim of this project is proof of concept. That is to demonstrate a framework consisting of a set of example mathematical web services together with a brokering platform used to discover and invoke these services given a query by a user to solve a particular problem.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128224698","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199379
T. Okumura, D. Mossé, M. Minami, O. Nakamura
This paper tries to tackle the problem of providing retrofitting network QoS in clustered configurations. For this purpose, we designed a QoS manager which runs on each of the internal cluster nodes and controls network I/O of local interface cooperating with peer managers on other nodes towards a certain QoS policy. First, we show the design of control framework, contending that an end-host manager-based mechanism is a desirable approach, which utilizes an end-host oriented network control primitive, Netnice. Second, for flexibility of configuration, we propose object-oriented modeling of the QoS manager with event-handler based configuration mechanism, and show the design of an object-oriented configuration language that allow simple and flexible definition of QoS policies. Lastly, results from two simple experiments with a Web server cluster are analyzed.
{"title":"Network QoS management framework for server clusters an end-host retrofitting event-handler approach using netnice","authors":"T. Okumura, D. Mossé, M. Minami, O. Nakamura","doi":"10.1109/CCGRID.2003.1199379","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199379","url":null,"abstract":"This paper tries to tackle the problem of providing retrofitting network QoS in clustered configurations. For this purpose, we designed a QoS manager which runs on each of the internal cluster nodes and controls network I/O of local interface cooperating with peer managers on other nodes towards a certain QoS policy. First, we show the design of control framework, contending that an end-host manager-based mechanism is a desirable approach, which utilizes an end-host oriented network control primitive, Netnice. Second, for flexibility of configuration, we propose object-oriented modeling of the QoS manager with event-handler based configuration mechanism, and show the design of an object-oriented configuration language that allow simple and flexible definition of QoS policies. Lastly, results from two simple experiments with a Web server cluster are analyzed.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130426612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199406
Liang Peng, W. Wong, C. Yuen
Distributed Shared Memory (DSM) is a highly desirable programming model for cluster based computing. Even though a number of software DSMs have been developed with their performance evaluated, few of them have a theoretical performance model. In this paper, we propose and analyze the performance model of the software DSM of SilkRoad,, a multithreaded runtime system for cluster computing. SilkRoad is built on the Cilk system with an extended memory consistency model which we call RC/sub dag/ consistency: Extending Cilk's theoretical performance model, we show that with the RC/sub dag/ consistent DSM, the expected execution.time T/sub P/ of a partially strict multithreaded computation on P processors is O((T/sub 1/(Z, L) + N)/P + /spl mu/HT/sub /spl infin//), where T/sub 1/(Z, L) is the total work of computation, T/sub /spl infin// is the critical path, N is the number of lock acquisitions, /spl mu/ is the service time of a cache miss, and H is the height of memory: Finally, we present experimental evidence that verify the performance model.
{"title":"The performance model of SilkRoad - a multithreaded DSM system for clusters","authors":"Liang Peng, W. Wong, C. Yuen","doi":"10.1109/CCGRID.2003.1199406","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199406","url":null,"abstract":"Distributed Shared Memory (DSM) is a highly desirable programming model for cluster based computing. Even though a number of software DSMs have been developed with their performance evaluated, few of them have a theoretical performance model. In this paper, we propose and analyze the performance model of the software DSM of SilkRoad,, a multithreaded runtime system for cluster computing. SilkRoad is built on the Cilk system with an extended memory consistency model which we call RC/sub dag/ consistency: Extending Cilk's theoretical performance model, we show that with the RC/sub dag/ consistent DSM, the expected execution.time T/sub P/ of a partially strict multithreaded computation on P processors is O((T/sub 1/(Z, L) + N)/P + /spl mu/HT/sub /spl infin//), where T/sub 1/(Z, L) is the total work of computation, T/sub /spl infin// is the critical path, N is the number of lock acquisitions, /spl mu/ is the service time of a cache miss, and H is the height of memory: Finally, we present experimental evidence that verify the performance model.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130480112","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199356
F. Dehne, Todd Eavis, A. Rau-Chaplin
This paper addresses the query performance issue for Relational OLAP (ROLAP) datacubes. We present a distributed multi-dimensional ROLAP indexing scheme which is practical to implement, requires only a small communication volume, and is fully adapted to distributed disks. Our solution is efficient for spatial searches in high dimensions and scalable in terms of data sizes, dimensions, and number of processors. Our method is also incrementally maintainable. Using "surrogate" group-bys, it allows for the efficient processing of arbitrary OLAP queries on partial cubes, where not all of the group-bys have been materialized. Our experiments show that the ROLAP advantage of better scalability, in comparison to MOLAP can be maintained while providing, at the same time, a fast and flexible index for OLAP queries.
{"title":"Parallel multi-dimensional ROLAP indexing","authors":"F. Dehne, Todd Eavis, A. Rau-Chaplin","doi":"10.1109/CCGRID.2003.1199356","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199356","url":null,"abstract":"This paper addresses the query performance issue for Relational OLAP (ROLAP) datacubes. We present a distributed multi-dimensional ROLAP indexing scheme which is practical to implement, requires only a small communication volume, and is fully adapted to distributed disks. Our solution is efficient for spatial searches in high dimensions and scalable in terms of data sizes, dimensions, and number of processors. Our method is also incrementally maintainable. Using \"surrogate\" group-bys, it allows for the efficient processing of arbitrary OLAP queries on partial cubes, where not all of the group-bys have been materialized. Our experiments show that the ROLAP advantage of better scalability, in comparison to MOLAP can be maintained while providing, at the same time, a fast and flexible index for OLAP queries.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"87 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131969894","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199422
Tieyan Li, Zhi-Gang Zhao, Siting Yo
Peer-to-Peer (p2p), as an emerging technology, is exerting huge influence on various application scenarios. Meanwhile, agents, assisting applications in traditional paradigm with intelligent ways, have to fit in this new trend. Observing the potential benefits of combining agent and p2p technologies, we propose a novel platform- "A-peer" in our approach. A-peer is an agent based secure p2p platform which empowers agents to be deployed in p2p environment. In this paper, we define the agent properties and mainly describe its protocol design, essentially agent transmission protocol, agent discovery protocol and agent messaging protocol. We also discuss the security aspect of the platform. Our implementation is based on Aglets system [2] and JXTA platform [1]. A-peer could be efficient for agent based p2p programming for distributed computing, decentralized collaboration, information searching and retrieval, and E-business.
{"title":"A-peer: an agent platform integrating peer-to-peer network","authors":"Tieyan Li, Zhi-Gang Zhao, Siting Yo","doi":"10.1109/CCGRID.2003.1199422","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199422","url":null,"abstract":"Peer-to-Peer (p2p), as an emerging technology, is exerting huge influence on various application scenarios. Meanwhile, agents, assisting applications in traditional paradigm with intelligent ways, have to fit in this new trend. Observing the potential benefits of combining agent and p2p technologies, we propose a novel platform- \"A-peer\" in our approach. A-peer is an agent based secure p2p platform which empowers agents to be deployed in p2p environment. In this paper, we define the agent properties and mainly describe its protocol design, essentially agent transmission protocol, agent discovery protocol and agent messaging protocol. We also discuss the security aspect of the platform. Our implementation is based on Aglets system [2] and JXTA platform [1]. A-peer could be efficient for agent based p2p programming for distributed computing, decentralized collaboration, information searching and retrieval, and E-business.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133866903","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199398
Nodoka Mimura, K. Nakauchi, H. Morikawa, T. Aoyama
Application-level multicast (ALM) is being increasingly recognized as a solution to support multipoint applications without the need for a network layer multicast protocol. Though several ALM systems have been proposed, all the multicast functions are independently developed and integrated into individual applications. However, we can find out the common functions of ALM among the existing systems. Our goal is to abstract the common functions and incorporate them in the middleware called RelayCast. RelayCast meets the fundamental requirements of various applications by combining several components of ALM functions. In this paper, we firstly present the architecture and implementation of RelayCast. Then, we present the multipath routing mechanism as a component. The multi-path routing achieves fast tree rebuilding at the time of tree partition caused by end-host's leave or failure.
{"title":"RelayCast: a middleware for application-level multicast services","authors":"Nodoka Mimura, K. Nakauchi, H. Morikawa, T. Aoyama","doi":"10.1109/CCGRID.2003.1199398","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199398","url":null,"abstract":"Application-level multicast (ALM) is being increasingly recognized as a solution to support multipoint applications without the need for a network layer multicast protocol. Though several ALM systems have been proposed, all the multicast functions are independently developed and integrated into individual applications. However, we can find out the common functions of ALM among the existing systems. Our goal is to abstract the common functions and incorporate them in the middleware called RelayCast. RelayCast meets the fundamental requirements of various applications by combining several components of ALM functions. In this paper, we firstly present the architecture and implementation of RelayCast. Then, we present the multipath routing mechanism as a component. The multi-path routing achieves fast tree rebuilding at the time of tree partition caused by end-host's leave or failure.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132288393","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199405
Cristian Tapus
Interprocess communication (IPC) is ubiquitous in today's computing world. One of the simplest mechanisms for IPC is shared memory. We present a system that enhances the System V IPC API to support distributed shared memory (DSM) by using speculations. Speculations provide performance improvements by enabling rollback of overly optimistic speculative executions. This paper describes a speculative total order communication protocol, a speculative sequential consistency model, and a speculative distributed locking mechanism. All these are supported by a mathematical model showing the advantages of speculative execution over traditional execution. Our DSM system is part of the Mojave system, which consists of a compiler and the extensions of the operating system designed to support speculations and process migration. The goal of our system is to provide a simpler programming paradigm for designers of distributed systems.
进程间通信(IPC)在当今的计算世界中无处不在。最简单的IPC机制之一是共享内存。我们提出了一个系统，通过使用推测来增强system V IPC API以支持分布式共享内存。通过启用过度乐观的投机执行的回滚，投机提供了性能改进。本文描述了一种推测全序通信协议、推测序列一致性模型和推测分布式锁定机制。所有这些都得到了一个数学模型的支持，该模型显示了投机执行相对于传统执行的优势。我们的DSM系统是Mojave系统的一部分，该系统由编译器和操作系统的扩展组成，旨在支持推测和进程迁移。我们系统的目标是为分布式系统的设计者提供一个更简单的编程范例。
{"title":"Kernel level speculative DSM","authors":"Cristian Tapus","doi":"10.1109/CCGRID.2003.1199405","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199405","url":null,"abstract":"Interprocess communication (IPC) is ubiquitous in today's computing world. One of the simplest mechanisms for IPC is shared memory. We present a system that enhances the System V IPC API to support distributed shared memory (DSM) by using speculations. Speculations provide performance improvements by enabling rollback of overly optimistic speculative executions. This paper describes a speculative total order communication protocol, a speculative sequential consistency model, and a speculative distributed locking mechanism. All these are supported by a mathematical model showing the advantages of speculative execution over traditional execution. Our DSM system is part of the Mojave system, which consists of a compiler and the extensions of the operating system designed to support speculations and process migration. The goal of our system is to provide a simpler programming paradigm for designers of distributed systems.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128686289","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199354
James Cheetham, F. Dehne, A. Rau-Chaplin, U. Stege, Peter J. Taillon
Fixed-parameter tractability (FPT) techniques have recently been successful in solving NP-complete problem instances of practical importance which were too large to be solved with previous methods. In this paper we show how to enhance this approach through the addition of parallelism, thereby allowing even larger problem instances to be solved in practice. More precisely, we demonstrate the potential of parallelism when applied to the bounded-tree search phase of FPT algorithms. We apply our methodology to the k-VERTEX COVER problem which has important applications, e.g., in multiple sequence alignments for computational biochemistry. We have implemented our parallel FPT method for the k-VERTEX COVER problem using C and the MPI communication library, and tested it on a PC cluster. This is the first experimental examination of parallel FPT techniques. We have tested our parallel k-VERTEX COVER method on protein sequences obtained from the National Center for Biotechnology Information. As part of our experiments, we solved larger instances of k-VERTEX COVER than in any previously reported implementations. For example, our code can solve problem instances with k /spl ges/ 400 in less than 1.5 hours. Since our parallel FPT algorithm requires only very little communication between processors, we expect our method to also perform well on Grids.
{"title":"A parallel FPT application for clusters","authors":"James Cheetham, F. Dehne, A. Rau-Chaplin, U. Stege, Peter J. Taillon","doi":"10.1109/CCGRID.2003.1199354","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199354","url":null,"abstract":"Fixed-parameter tractability (FPT) techniques have recently been successful in solving NP-complete problem instances of practical importance which were too large to be solved with previous methods. In this paper we show how to enhance this approach through the addition of parallelism, thereby allowing even larger problem instances to be solved in practice. More precisely, we demonstrate the potential of parallelism when applied to the bounded-tree search phase of FPT algorithms. We apply our methodology to the k-VERTEX COVER problem which has important applications, e.g., in multiple sequence alignments for computational biochemistry. We have implemented our parallel FPT method for the k-VERTEX COVER problem using C and the MPI communication library, and tested it on a PC cluster. This is the first experimental examination of parallel FPT techniques. We have tested our parallel k-VERTEX COVER method on protein sequences obtained from the National Center for Biotechnology Information. As part of our experiments, we solved larger instances of k-VERTEX COVER than in any previously reported implementations. For example, our code can solve problem instances with k /spl ges/ 400 in less than 1.5 hours. Since our parallel FPT algorithm requires only very little communication between processors, we expect our method to also perform well on Grids.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126034900","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2003-05-12DOI: 10.1109/CCGRID.2003.1199435
Olivier Cozette, Cyril Randriamaro, G. Utard
Grand challenge applications have to process large amounts of data, and then require high performance IO systems. Cluster computing is a good alternative to proprietary system for building cost effective IO intensive platform: some cluster architectures won sorting benchmark (MinuteSort, Datamation)! Recent advances in IO component technologies (disk, controller and network) let us expect higher IO performance for data intensive applications on cluster. The counterpart of this evolution is that much stress is put on the different buses (memory, IO) of each node which cannot be scaled. In this paper we investigate a strategy we called READ/sup 2/ (Remote Efficient Access to Distant Device) to reduce this stress. With READ/sup 2/ any cluster node accesses directly to remote disk: the remote processor and the remote memory are removed from the control and data path: Inputs/Outputs don't interfere with the host processor and the host memory activity. With READ/sup 2/ strategy, a cluster can be considered as a shared disk architecture instead of a shared nothing one. This papers describes an implementation of READ/sup 2/ on Myrinet Networks. First experimental results show IO performance improvement.
{"title":"READ/sup 2/: Put disks at network level","authors":"Olivier Cozette, Cyril Randriamaro, G. Utard","doi":"10.1109/CCGRID.2003.1199435","DOIUrl":"https://doi.org/10.1109/CCGRID.2003.1199435","url":null,"abstract":"Grand challenge applications have to process large amounts of data, and then require high performance IO systems. Cluster computing is a good alternative to proprietary system for building cost effective IO intensive platform: some cluster architectures won sorting benchmark (MinuteSort, Datamation)! Recent advances in IO component technologies (disk, controller and network) let us expect higher IO performance for data intensive applications on cluster. The counterpart of this evolution is that much stress is put on the different buses (memory, IO) of each node which cannot be scaled. In this paper we investigate a strategy we called READ/sup 2/ (Remote Efficient Access to Distant Device) to reduce this stress. With READ/sup 2/ any cluster node accesses directly to remote disk: the remote processor and the remote memory are removed from the control and data path: Inputs/Outputs don't interfere with the host processor and the host memory activity. With READ/sup 2/ strategy, a cluster can be considered as a shared disk architecture instead of a shared nothing one. This papers describes an implementation of READ/sup 2/ on Myrinet Networks. First experimental results show IO performance improvement.","PeriodicalId":433323,"journal":{"name":"CCGrid 2003. 3rd IEEE/ACM International Symposium on Cluster Computing and the Grid, 2003. Proceedings.","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2003-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116843077","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：