Cloud computing has emerged as a new technology that provides large amounts of computing and data storage capacity to its users with a promise of increased scalability, high availability, and reduced administration and maintenance costs. As the use of cloud computing environments increases, it becomes crucial to understand the performance of these environments. So, it is of great importance to assess the performance of computing clouds in terms of various metrics, such as the overhead of acquiring and releasing the virtual computing resources, and other virtualization and network communications overheads. To address these issues, we have designed and implemented C-Meter, which is a portable, extensible, and easy-to-use framework for generating and submitting test workloads to computing clouds. In this paper, first we state the requirements for frameworks to assess the performance of computing clouds. Then, we present the architecture of the C-Meter framework and discuss several cloud resource management alternatives. Finally, we present ourearly experiences with C-Meter in Amazon EC2. We show how C-Meter can be used for assessing the overhead of acquiring and releasing the virtual computing resources, for comparing different configurations, and for evaluating different scheduling algorithms.
{"title":"C-Meter: A Framework for Performance Analysis of Computing Clouds","authors":"N. Yigitbasi, A. Iosup, D. Epema, S. Ostermann","doi":"10.1109/CCGRID.2009.40","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.40","url":null,"abstract":"Cloud computing has emerged as a new technology that provides large amounts of computing and data storage capacity to its users with a promise of increased scalability, high availability, and reduced administration and maintenance costs. As the use of cloud computing environments increases, it becomes crucial to understand the performance of these environments. So, it is of great importance to assess the performance of computing clouds in terms of various metrics, such as the overhead of acquiring and releasing the virtual computing resources, and other virtualization and network communications overheads. To address these issues, we have designed and implemented C-Meter, which is a portable, extensible, and easy-to-use framework for generating and submitting test workloads to computing clouds. In this paper, first we state the requirements for frameworks to assess the performance of computing clouds. Then, we present the architecture of the C-Meter framework and discuss several cloud resource management alternatives. Finally, we present ourearly experiences with C-Meter in Amazon EC2. We show how C-Meter can be used for assessing the overhead of acquiring and releasing the virtual computing resources, for comparing different configurations, and for evaluating different scheduling algorithms.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"141 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124472809","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}
Whenever a resource allocation fails although enough free capacity being available, fragmentation is easily spotted as cause. But how the fragmentation in a system requiring continuous allocations like time schedules or memory can be quantified is hardly analyzed. A Grid environment using advance reservation even combines two-dimensions: time and resource dimension. In this paper a new way to measure the fragmentation of a system in one dimension is proposed. This measure is then extended to incorporate also the second dimension. Extensive simulations showed that the proposed fragmentation measure is a good indicator of the state of the system.
{"title":"Measuring Fragmentation of Two-Dimensional Resources Applied to Advance Reservation Grid Scheduling","authors":"J. Gehr, Jörg Schneider","doi":"10.1109/CCGRID.2009.81","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.81","url":null,"abstract":"Whenever a resource allocation fails although enough free capacity being available, fragmentation is easily spotted as cause. But how the fragmentation in a system requiring continuous allocations like time schedules or memory can be quantified is hardly analyzed. A Grid environment using advance reservation even combines two-dimensions: time and resource dimension. In this paper a new way to measure the fragmentation of a system in one dimension is proposed. This measure is then extended to incorporate also the second dimension. Extensive simulations showed that the proposed fragmentation measure is a good indicator of the state of the system.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123440541","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}
Workflow management system is widely accepted and used in the wide area network environment, especially in the e-Science application scenarios, to coordinate the operation of different functional components and to provide more powerful functions. The error-prone nature of the wide area network environment makes the fault-tolerance requirements of workflow management become more and more urgent. In this paper, we propose Cesar-FD, a stateful fault detection mechanism, which builds up states related to the runtime and external environments of workflow management system by aggregating multiple messages and provides more accurate notifications asynchronously. We demonstrate the use of this mechanism in the Drug Discovery Grid environment by two use cases. We also show that it can be used to detect faulty situations more accurately.
{"title":"Cesar-FD: An Effective Stateful Fault Detection Mechanism in Drug Discovery Grid","authors":"Yongjian Wang, Yinan Ren, Ting-Wen Chen, Yuanqiang Huang, Zhongzhi Luan, Zhongxin Wu, D. Qian","doi":"10.1109/CCGRID.2009.28","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.28","url":null,"abstract":"Workflow management system is widely accepted and used in the wide area network environment, especially in the e-Science application scenarios, to coordinate the operation of different functional components and to provide more powerful functions. The error-prone nature of the wide area network environment makes the fault-tolerance requirements of workflow management become more and more urgent. In this paper, we propose Cesar-FD, a stateful fault detection mechanism, which builds up states related to the runtime and external environments of workflow management system by aggregating multiple messages and provides more accurate notifications asynchronously. We demonstrate the use of this mechanism in the Drug Discovery Grid environment by two use cases. We also show that it can be used to detect faulty situations more accurately.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121682333","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}
Replication in grid file systems can significantly improve I/O performance of data-intensive applications. However, most of existing replication techniques apply to individual files, which may introduce inefficient replication overheads for a large number of files. We propose a file clustering based replication algorithm for grid file systems. Our algorithm groups files according to a relationship of simultaneous accesses between files and stores replicas of the clustered files into storage nodes, to satisfy expected most of future read access times to the clustered files and replication times for individual files being minimized under the given storage capacity limitation. Our experiments on a given grid environment, 20 nodes of 5 sites, suggest that the proposed algorithm achieves accurate file clustering and efficient replica management; our clustering policy with the file cluster size limit of 5120 MB and the storage capacity limit for replicas of 10240 MB exhibits 1.58 times efficiency than the policy that never groups related files. The results also indicate that the overheads required for introducing our algorithm significantly affect I/O performance of running applications.
{"title":"File Clustering Based Replication Algorithm in a Grid Environment","authors":"Hitoshi Sato, S. Matsuoka, Toshio Endo","doi":"10.1109/CCGRID.2009.73","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.73","url":null,"abstract":"Replication in grid file systems can significantly improve I/O performance of data-intensive applications. However, most of existing replication techniques apply to individual files, which may introduce inefficient replication overheads for a large number of files. We propose a file clustering based replication algorithm for grid file systems. Our algorithm groups files according to a relationship of simultaneous accesses between files and stores replicas of the clustered files into storage nodes, to satisfy expected most of future read access times to the clustered files and replication times for individual files being minimized under the given storage capacity limitation. Our experiments on a given grid environment, 20 nodes of 5 sites, suggest that the proposed algorithm achieves accurate file clustering and efficient replica management; our clustering policy with the file cluster size limit of 5120 MB and the storage capacity limit for replicas of 10240 MB exhibits 1.58 times efficiency than the policy that never groups related files. The results also indicate that the overheads required for introducing our algorithm significantly affect I/O performance of running applications.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126730950","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}
Grid infrastructures are in operation around the world, federating an impressive collection of computational resources and a wide variety of application software. In this context, it is important to establish advanced software discovery services that could help end-users locate software components suitable to their needs. In this paper, we present the design, architecture and implementation of an open-source keyword-based paradigm for the search of software resources in Grid infrastructures, called Minersoft. A key goal of Minersoft is to annotate automatically all the software resources with keyword-rich metadata. Using advanced Information Retrieval techniques, we locate software resources with respect to users queries. Experiments were conducted in EGEE, one of the largest Grid production services currently in operation. Results showed that Minersoft successfully crawled 12.3 million valid files (620 GB size) and sustained, in most sites, high crawling rates.
{"title":"Harvesting Large-Scale Grids for Software Resources","authors":"Asterios Katsifodimos, G. Pallis, M. Dikaiakos","doi":"10.1109/CCGRID.2009.51","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.51","url":null,"abstract":"Grid infrastructures are in operation around the world, federating an impressive collection of computational resources and a wide variety of application software. In this context, it is important to establish advanced software discovery services that could help end-users locate software components suitable to their needs. In this paper, we present the design, architecture and implementation of an open-source keyword-based paradigm for the search of software resources in Grid infrastructures, called Minersoft. A key goal of Minersoft is to annotate automatically all the software resources with keyword-rich metadata. Using advanced Information Retrieval techniques, we locate software resources with respect to users queries. Experiments were conducted in EGEE, one of the largest Grid production services currently in operation. Results showed that Minersoft successfully crawled 12.3 million valid files (620 GB size) and sustained, in most sites, high crawling rates.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"59 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130534934","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}
This paper presents the transient performancestudy on the adaptability of an application layer multicastprotocol, namely Adaptive Overlay Multicast (AOM). The studyfocuses on how the application layer multicast tree structure adapts to network dynamics and faults by using efficient tree management and adaptation algorithms, thus reducing the adverse effects of network faults on the application performance. We present extensive simulation studies as well as studies using real Internet data The results show that AOM is highly adaptive to network dynamics while incurring low overhead.
{"title":"Transient Analysis of an Overlay Multicast Protocol","authors":"Xiaobing Hou, S. Wu","doi":"10.1109/CCGRID.2009.27","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.27","url":null,"abstract":"This paper presents the transient performancestudy on the adaptability of an application layer multicastprotocol, namely Adaptive Overlay Multicast (AOM). The studyfocuses on how the application layer multicast tree structure adapts to network dynamics and faults by using efficient tree management and adaptation algorithms, thus reducing the adverse effects of network faults on the application performance. We present extensive simulation studies as well as studies using real Internet data The results show that AOM is highly adaptive to network dynamics while incurring low overhead.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115864684","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}
The Science & Technology Facilities Council is home to international Facilities such as the ISIS Neutron Spallation Source ≪http://www.isis.rl.ac.uk≫, Central Laser Facility ≪http://www.clf.rl.ac.uk/≫ and Diamond Light Source ≪http://www.diamond.ac.uk/default.htm≫, the National Grid Service ≪http://www.grid-support.ac.uk/≫ including national super computers ≪http://www.cse.scitech.ac.uk≫, Tier1 data service for CERN particle physics experiment, the British Atmospheric data Centre and the British Oceanographic Data Centre at the Space Science and Technology department. Together these Facilities generate several Terabytes of data per month which needs to be handled, catalogued and provided access to. In addition, the scientists within STFC departments also develop complex simulations and undertake data analysis for their own experiments. Facilities also have strong ongoing collaborations with UK academic and commercial users through their involvement with Collaborative Computational Programme, generating very large simulation datasets. There is thus the need to support high resolution data analysis using distributed compute, data and visualization resources. At the same time, these requirements offer the computational and visualization scientists within STFC unique opportunities to advocate the take up of advanced visualization techniques and toolkits in distributed high performance, high resolution hardware environment. It gives an opportunity to understand the requirements and usefulness of distributed visualization. Given this seemingly advantageous position, the STFC vizNET ≪http://www.viznet.ac.uk/≫ partners have been actively pursuing visualization awareness activities and services aimed at application holders of various scientific disciplines. These activities include holding workshops, hands-on tutorials, show case demonstrations and the setting up of hardware based visualization services with technical support. This report provides details of these activities, the outcomes, the status and some suggestions as to the way forward.
{"title":"Supporting Distributed Visualization Services for High Performance Science and Engineering Applications A Service Provider Perspective","authors":"L. Sastry, R. Fowler, S. Nagella, J. Churchill","doi":"10.1109/CCGRID.2009.94","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.94","url":null,"abstract":"The Science & Technology Facilities Council is home to international Facilities such as the ISIS Neutron Spallation Source ≪http://www.isis.rl.ac.uk≫, Central Laser Facility ≪http://www.clf.rl.ac.uk/≫ and Diamond Light Source ≪http://www.diamond.ac.uk/default.htm≫, the National Grid Service ≪http://www.grid-support.ac.uk/≫ including national super computers ≪http://www.cse.scitech.ac.uk≫, Tier1 data service for CERN particle physics experiment, the British Atmospheric data Centre and the British Oceanographic Data Centre at the Space Science and Technology department. Together these Facilities generate several Terabytes of data per month which needs to be handled, catalogued and provided access to. In addition, the scientists within STFC departments also develop complex simulations and undertake data analysis for their own experiments. Facilities also have strong ongoing collaborations with UK academic and commercial users through their involvement with Collaborative Computational Programme, generating very large simulation datasets. There is thus the need to support high resolution data analysis using distributed compute, data and visualization resources. At the same time, these requirements offer the computational and visualization scientists within STFC unique opportunities to advocate the take up of advanced visualization techniques and toolkits in distributed high performance, high resolution hardware environment. It gives an opportunity to understand the requirements and usefulness of distributed visualization. Given this seemingly advantageous position, the STFC vizNET ≪http://www.viznet.ac.uk/≫ partners have been actively pursuing visualization awareness activities and services aimed at application holders of various scientific disciplines. These activities include holding workshops, hands-on tutorials, show case demonstrations and the setting up of hardware based visualization services with technical support. This report provides details of these activities, the outcomes, the status and some suggestions as to the way forward.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124973855","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}
Today many Peer-to-peer (P2P) applications are widely used on the Internet. Especially file sharing is a popular P2P application that has, at least partially, replaced the centralized file sharing infrastructure. However, there are still a number of legacy applications that utilize a centralized infrastructure as opposed to a decentralized approach. In this paper we present a generic framework for decentralizing legacy applications. Even though we focus especially on Voice over IP (VoIP), email, and web applications, we believe that our framework could also be utilized with other legacy applications. A notable feature of our framework is that it does not require any changes to legacy applications.
{"title":"Framework for Decentralizing Legacy Applications","authors":"J. Hautakorpi, G. Camarillo, David López","doi":"10.1109/CCGRID.2009.75","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.75","url":null,"abstract":"Today many Peer-to-peer (P2P) applications are widely used on the Internet. Especially file sharing is a popular P2P application that has, at least partially, replaced the centralized file sharing infrastructure. However, there are still a number of legacy applications that utilize a centralized infrastructure as opposed to a decentralized approach. In this paper we present a generic framework for decentralizing legacy applications. Even though we focus especially on Voice over IP (VoIP), email, and web applications, we believe that our framework could also be utilized with other legacy applications. A notable feature of our framework is that it does not require any changes to legacy applications.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"108 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122756914","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}
Range query in Peer-to-Peer networks based on Distributed Hash Table (DHT) is still an open problem. The traditional way uses order-preserving hashing functions to create value indexes that are placed and stored on the corresponding peers to support range query. The way, however, suffers from high index maintenance costs. To avoid the issue, a scalable blind search method over DHTs - recursive partition search (RPS) can be used. But, RPS still easily incurs high network overhead as network size grows. Thus, in this paper, a learning-aware RPS (LARPS) is proposed to overcome the disadvantages of two approaches above mentioned. Extensive experiments show LARPS is a scalable and robust approach for range query, especially in the following cases: a) query range is wide, b) the requested resources follow Zipf distribution, and c) the number of required resources is small.
{"title":"Range Query Using Learning-Aware RPS in DHT-Based Peer-to-Peer Networks","authors":"Ze Deng, D. Feng, Ke Zhou, Zhan Shi, Chao Luo","doi":"10.1109/CCGRID.2009.25","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.25","url":null,"abstract":"Range query in Peer-to-Peer networks based on Distributed Hash Table (DHT) is still an open problem. The traditional way uses order-preserving hashing functions to create value indexes that are placed and stored on the corresponding peers to support range query. The way, however, suffers from high index maintenance costs. To avoid the issue, a scalable blind search method over DHTs - recursive partition search (RPS) can be used. But, RPS still easily incurs high network overhead as network size grows. Thus, in this paper, a learning-aware RPS (LARPS) is proposed to overcome the disadvantages of two approaches above mentioned. Extensive experiments show LARPS is a scalable and robust approach for range query, especially in the following cases: a) query range is wide, b) the requested resources follow Zipf distribution, and c) the number of required resources is small.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125156573","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}
Wim Depoorter, R. Y. V. Bossche, K. Vanmechelen, J. Broeckhove
The ef¿cient scheduling of jobs is an essential part of any grid resource management system. At its core, it involves ¿nding a solution to a problem which is NP-complete by reduction to the knapsack problem. Consequently, this problem is often tackled by using heuristics to derive a more pragmatic solution. Other than the use of heuristics, simpli¿cations and abstractions of the workload model may also be employed to increase the tractability of the scheduling problem. A possible abstraction in this context is the use of Divisible Load Theory (DLT), in which it is assumed that an application consists of an arbitrarily divisible load (ADL). Many applications however, are composed of a number of atomic tasks and are only modularly divisible. In this paper we evaluate the consequences of the ADL assumption on the performance of economic scheduling approaches for grids, in the context of CPU-bound modularly divisible applications with hard deadlines. Our goal is to evaluate to what extent DLT can still serve as a useful workload abstraction for obtaining tractable scheduling algorithms in this setting. The focus of our evaluation is on the recently proposed tsfGrid heuristic for economic scheduling of grid workloads which operates under the assumptions of ADL. We demonstrate the effect of the ADL assumption on the actual instantiation of schedules and on the user value realized by the RMS. In addition we describe how the usage of a DLT heuristic in a high-level admission controller for a mechanism which does take into account the atomicity of individual tasks, can signi¿cantly reduce communication and computational overhead.
{"title":"Evaluating the Divisible Load Assumption in the Context of Economic Grid Scheduling with Deadline-Based QoS guarantees","authors":"Wim Depoorter, R. Y. V. Bossche, K. Vanmechelen, J. Broeckhove","doi":"10.1109/CCGRID.2009.36","DOIUrl":"https://doi.org/10.1109/CCGRID.2009.36","url":null,"abstract":"The ef¿cient scheduling of jobs is an essential part of any grid resource management system. At its core, it involves ¿nding a solution to a problem which is NP-complete by reduction to the knapsack problem. Consequently, this problem is often tackled by using heuristics to derive a more pragmatic solution. Other than the use of heuristics, simpli¿cations and abstractions of the workload model may also be employed to increase the tractability of the scheduling problem. A possible abstraction in this context is the use of Divisible Load Theory (DLT), in which it is assumed that an application consists of an arbitrarily divisible load (ADL). Many applications however, are composed of a number of atomic tasks and are only modularly divisible. In this paper we evaluate the consequences of the ADL assumption on the performance of economic scheduling approaches for grids, in the context of CPU-bound modularly divisible applications with hard deadlines. Our goal is to evaluate to what extent DLT can still serve as a useful workload abstraction for obtaining tractable scheduling algorithms in this setting. The focus of our evaluation is on the recently proposed tsfGrid heuristic for economic scheduling of grid workloads which operates under the assumptions of ADL. We demonstrate the effect of the ADL assumption on the actual instantiation of schedules and on the user value realized by the RMS. In addition we describe how the usage of a DLT heuristic in a high-level admission controller for a mechanism which does take into account the atomicity of individual tasks, can signi¿cantly reduce communication and computational overhead.","PeriodicalId":118263,"journal":{"name":"2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130208296","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：