Pub Date : 2009-11-17DOI: 10.1109/ChinaGrid.2009.29
Yong-mei Lei, Shao-jun Chen
In this paper, we investigate the parallel reliable computational model and propose a parallel interval evolutionary algorithm that integrates interval arithmetic and Mind Evolutionary Computation method. The major aim is to explorer the new parallel interval decomposition scheme can solve computation intensive problem and can determine the all optimal solution reliably. The proposed algorithm is experimentally testified on the ZiQiang 3000 cluster of Shanghai High Education Grid-e-Grid Computational Application Platform with a test suit containing 6 complex multi-modal function optimization benchmarks.
{"title":"A Reliable Parallel Interval Global Optimization Algorithm Based on Mind Evolutionary Computation","authors":"Yong-mei Lei, Shao-jun Chen","doi":"10.1109/ChinaGrid.2009.29","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.29","url":null,"abstract":"In this paper, we investigate the parallel reliable computational model and propose a parallel interval evolutionary algorithm that integrates interval arithmetic and Mind Evolutionary Computation method. The major aim is to explorer the new parallel interval decomposition scheme can solve computation intensive problem and can determine the all optimal solution reliably. The proposed algorithm is experimentally testified on the ZiQiang 3000 cluster of Shanghai High Education Grid-e-Grid Computational Application Platform with a test suit containing 6 complex multi-modal function optimization benchmarks.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"57 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116577233","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 : 2009-08-21DOI: 10.1109/ChinaGrid.2009.9
Xiaohui Wei, Hongliang Li, Dexiong Li
Grid is attracting more and more attentions by its massive computational capacity. Tools like Globus Toolkit and MPICH-G2 have been developed to help scientists to facilitate their researches. As a Grid-enabled implementation of MPI, MPICH-G2 helps developers to port parallel applications to cross-domain environment. Since the current computationally-intensive parallel applications, especially long-running tasks, require high availability as well as high performance computing platform, dynamic job migration in Grid environment has became an essential issue. In this study, we present a dynamic job migration enabled MPICH-G2 version, MPICH-G-DM. We use Virtual Job Model (VJM) to reserve resources for the migrating jobs in advance to improve the efficiency of the system. An Asynchronous Migration Protocol (AMP) is proposed to enable the migrating sub jobs to checkpoint/restart and update their new addresses concurrently without a global synchronization. In order to reduce the communicating overhead of job migration, MPICH-G-DM minimized the number of control messages among domains to O(N). Experiment results show that MPICH-G-DM is effective and reliable.
{"title":"MPICH-G-DM: An Enhanced MPICH-G with Supporting Dynamic Job Migration","authors":"Xiaohui Wei, Hongliang Li, Dexiong Li","doi":"10.1109/ChinaGrid.2009.9","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.9","url":null,"abstract":"Grid is attracting more and more attentions by its massive computational capacity. Tools like Globus Toolkit and MPICH-G2 have been developed to help scientists to facilitate their researches. As a Grid-enabled implementation of MPI, MPICH-G2 helps developers to port parallel applications to cross-domain environment. Since the current computationally-intensive parallel applications, especially long-running tasks, require high availability as well as high performance computing platform, dynamic job migration in Grid environment has became an essential issue. In this study, we present a dynamic job migration enabled MPICH-G2 version, MPICH-G-DM. We use Virtual Job Model (VJM) to reserve resources for the migrating jobs in advance to improve the efficiency of the system. An Asynchronous Migration Protocol (AMP) is proposed to enable the migrating sub jobs to checkpoint/restart and update their new addresses concurrently without a global synchronization. In order to reduce the communicating overhead of job migration, MPICH-G-DM minimized the number of control messages among domains to O(N). Experiment results show that MPICH-G-DM is effective and reliable.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133301250","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 : 2009-08-21DOI: 10.1109/ChinaGrid.2009.26
Qingchao Cai, Xuejie Zhang
The data revelation policy in traditional mesh-based peer-to-peer streaming protocols always make peers announce all available data to their neighbors.However, we find that, in a low bandwidth network environment, this policy suffers from serious data loss. In addition, only data deadline and bandwidth are considered in the data scheduling algorithms in the existed protocols,which leads to the slow dissemination of newly generated content data.This paper first presents a strategic segment revelation policy that the source under-report the available data to its neighbors, while other peers announce all available data.According to simulation results, this policy can accelerate the dissemination of new segments, especially in low bandwidth environment. Then we propose a simple model of data scheduling which reveals the relation of download rate and other key factors in data scheduling. Based on this model, we further propose a greedy data scheduling algorithm, each data segment is given a priority according to its deadline and distribution among neighbors, and the segments with higher priority would be earlier requested such that these segments can be downloaded in the shortest time. The simulation results also indicate that the greedy data scheduling algorithm performs near-optimally in terms of bandwidth utilization.
{"title":"Towards Data Revelation and Scheduling Policy in Mesh-Based Peer-to-Peer Streaming","authors":"Qingchao Cai, Xuejie Zhang","doi":"10.1109/ChinaGrid.2009.26","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.26","url":null,"abstract":"The data revelation policy in traditional mesh-based peer-to-peer streaming protocols always make peers announce all available data to their neighbors.However, we find that, in a low bandwidth network environment, this policy suffers from serious data loss. In addition, only data deadline and bandwidth are considered in the data scheduling algorithms in the existed protocols,which leads to the slow dissemination of newly generated content data.This paper first presents a strategic segment revelation policy that the source under-report the available data to its neighbors, while other peers announce all available data.According to simulation results, this policy can accelerate the dissemination of new segments, especially in low bandwidth environment. Then we propose a simple model of data scheduling which reveals the relation of download rate and other key factors in data scheduling. Based on this model, we further propose a greedy data scheduling algorithm, each data segment is given a priority according to its deadline and distribution among neighbors, and the segments with higher priority would be earlier requested such that these segments can be downloaded in the shortest time. The simulation results also indicate that the greedy data scheduling algorithm performs near-optimally in terms of bandwidth utilization.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"80 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127190490","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}
Classification is a significant technique in data mining research and applications. C4.5 is a widely used classification method, and ensemble learning adopts a parallel and distributed computing model for classification. Based on analyses of the MapReduce computing paradigm and the process of ensemble learning, we find that the parallel and distributed computing model in MapReduce is appropriate for implementing ensemble learning. This paper takes the advantages of C4.5, ensemble learning and the MapReduce computing model, and proposes a new method MReC4.5 for parallel and distributed ensemble classification. Our experimental results show that increasing the number of nodes would benefit the effectiveness of classification modeling, and serialization operations at the model level make the MReC4.5 classifier “construct once, use anywhere”.
{"title":"MReC4.5: C4.5 Ensemble Classification with MapReduce","authors":"Gongqing Wu, Hai-Guang Li, Xuegang Hu, Yuan-Jun Bi, J. Zhang, Xindong Wu","doi":"10.1109/ChinaGrid.2009.39","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.39","url":null,"abstract":"Classification is a significant technique in data mining research and applications. C4.5 is a widely used classification method, and ensemble learning adopts a parallel and distributed computing model for classification. Based on analyses of the MapReduce computing paradigm and the process of ensemble learning, we find that the parallel and distributed computing model in MapReduce is appropriate for implementing ensemble learning. This paper takes the advantages of C4.5, ensemble learning and the MapReduce computing model, and proposes a new method MReC4.5 for parallel and distributed ensemble classification. Our experimental results show that increasing the number of nodes would benefit the effectiveness of classification modeling, and serialization operations at the model level make the MReC4.5 classifier “construct once, use anywhere”.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132843409","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 : 2009-08-21DOI: 10.1109/ChinaGrid.2009.23
Hongwei Liu, Jiong Yu, Guozhong Tian, Hongcui Gong
According to the dependence and deadline of the grid workflow tasks, the effective degrees and MIPS of the grid resources, a new algorithm called the priority tasks scheduling algorithm based on the grid resource prediction is presented. The algorithm uses DAG to find the critical path, obtain the deadline of every task and compute their PRI (priority). The algorithm takes the below problems into consideration: the request of user, the type of resources and re-scheduling of failed tasks. The result shows that the algorithm is effective.
{"title":"The Priority Tasks Scheduling Algorithm Based on Grid Resource Prediction","authors":"Hongwei Liu, Jiong Yu, Guozhong Tian, Hongcui Gong","doi":"10.1109/ChinaGrid.2009.23","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.23","url":null,"abstract":"According to the dependence and deadline of the grid workflow tasks, the effective degrees and MIPS of the grid resources, a new algorithm called the priority tasks scheduling algorithm based on the grid resource prediction is presented. The algorithm uses DAG to find the critical path, obtain the deadline of every task and compute their PRI (priority). The algorithm takes the below problems into consideration: the request of user, the type of resources and re-scheduling of failed tasks. The result shows that the algorithm is effective.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133683421","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 : 2009-08-21DOI: 10.1109/ChinaGrid.2009.28
Jun Gu, Junzhou Luo
Grid monitoring, as an important part of any grid systems, is needed to query the state of grid resources and match user requirements with available grid resources. In order to ensure the availability of grid monitoring, the reliability imposed by software or hardware failure happened with unpredictable probability must be assessed. This paper contributes to study the reliability analysis approach of grid monitoring in the context of grid monitoring architecture (GMA) that has been de facto standards for many areas of grid computing. Failure types and contributing factors in GMA are analyzed, which are likely to take place in comprised components, channels or process behaviors. Then, the respective evaluation equations are suggested via Markov procedure, queue model, and probability theory. Furthermore, the reliability issue of hierarchical GMA is discussed based on four basic architectural relations. Numerical examples are given to illustrate the proposed computing equations. The results show that our approach is feasible.
{"title":"Reliability Analysis Approach of Grid Monitoring Architecture","authors":"Jun Gu, Junzhou Luo","doi":"10.1109/ChinaGrid.2009.28","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.28","url":null,"abstract":"Grid monitoring, as an important part of any grid systems, is needed to query the state of grid resources and match user requirements with available grid resources. In order to ensure the availability of grid monitoring, the reliability imposed by software or hardware failure happened with unpredictable probability must be assessed. This paper contributes to study the reliability analysis approach of grid monitoring in the context of grid monitoring architecture (GMA) that has been de facto standards for many areas of grid computing. Failure types and contributing factors in GMA are analyzed, which are likely to take place in comprised components, channels or process behaviors. Then, the respective evaluation equations are suggested via Markov procedure, queue model, and probability theory. Furthermore, the reliability issue of hierarchical GMA is discussed based on four basic architectural relations. Numerical examples are given to illustrate the proposed computing equations. The results show that our approach is feasible.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131881835","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 : 2009-08-21DOI: 10.1109/ChinaGrid.2009.18
ZhongAn Lao, Nong Xiao, ChangSheng Fu
Memory intensive applications and I/O intensive applications often suffer from the poor performance of disk swapping when memory is inadequate.RAM Grid, which combines Network Memory, Service-oriented Computing and Grid Computing technology, focuses on solving these problems. Being excellent at providing shared memory to improve system performance, RAM Grid can’t guarantee busy node load balance or optimal file remote cache efficiency. After studying data placement policy of large-scale network storage systems, we propose a file classification caching service (FCCS) based on RAM Grid, which promises fairness and high availability, to give a method that tries to solve those shortages. As experiment results show: FCCS improve system performance greatly.
{"title":"FCCS: File Classification Caching Service Based on RAM Grid","authors":"ZhongAn Lao, Nong Xiao, ChangSheng Fu","doi":"10.1109/ChinaGrid.2009.18","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.18","url":null,"abstract":"Memory intensive applications and I/O intensive applications often suffer from the poor performance of disk swapping when memory is inadequate.RAM Grid, which combines Network Memory, Service-oriented Computing and Grid Computing technology, focuses on solving these problems. Being excellent at providing shared memory to improve system performance, RAM Grid can’t guarantee busy node load balance or optimal file remote cache efficiency. After studying data placement policy of large-scale network storage systems, we propose a file classification caching service (FCCS) based on RAM Grid, which promises fairness and high availability, to give a method that tries to solve those shortages. As experiment results show: FCCS improve system performance greatly.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"86 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122494901","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 : 2009-08-21DOI: 10.1109/ChinaGrid.2009.43
Bin Cheng, Hai Jin, Xiaofei Liao, Bo Li
Scheduling is essential in Peer-to-Peer (P2P) Video- on-Demand (VoD) applications to obtain desirable user experience and high sharing efficiency. Existing systems and proposals largely utilize scheduling mechanisms that are periodical in nature. We refer these as time-driven scheduling. In this paper we argue that the time-driven scheduling algorithm is not efficient for P2P VoD due to its inadequate resource utilization and inability in quickly adapting to the potential dynamics in such systems. We propose an event-driven scheduling algorithm for P2P VoD systems, in which the scheduler is triggered by events such as peer churn or/and random seeks. We evaluate our design through extensive simulations, and the experimental results show that event-driven scheduling outperforms time-driven scheduling in terms of both user experience and system scalability.
{"title":"Event-Driven Scheduling for P2P VoD","authors":"Bin Cheng, Hai Jin, Xiaofei Liao, Bo Li","doi":"10.1109/ChinaGrid.2009.43","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.43","url":null,"abstract":"Scheduling is essential in Peer-to-Peer (P2P) Video- on-Demand (VoD) applications to obtain desirable user experience and high sharing efficiency. Existing systems and proposals largely utilize scheduling mechanisms that are periodical in nature. We refer these as time-driven scheduling. In this paper we argue that the time-driven scheduling algorithm is not efficient for P2P VoD due to its inadequate resource utilization and inability in quickly adapting to the potential dynamics in such systems. We propose an event-driven scheduling algorithm for P2P VoD systems, in which the scheduler is triggered by events such as peer churn or/and random seeks. We evaluate our design through extensive simulations, and the experimental results show that event-driven scheduling outperforms time-driven scheduling in terms of both user experience and system scalability.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"65 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127028949","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 : 2009-08-21DOI: 10.1109/ChinaGrid.2009.42
Haikun Liu, Hai Jin, Xiaofei Liao
Virtual Machine (VM) level checkpoints bring several advantages which process-level checkpoint implementation can hardly provide: compatibility, transparence, flexibility, and simplicity. However, the size of VM-level checkpoint may be very large and even in the order of gigabytes. This disadvantage causes the VM checkpointing and restart time become very long. To reduce the size of VM checkpoint, this paper proposes a memory exclusion scheme using ballooning mechanism, which omits saving unnecessary free pages in the VM. We implement our prototype in Xen environment. Experimental measurements show our approach can significantly reduce the size of VM checkpoint with minimal runtime overhead, thereby greatly improve the checkpoint performance.
{"title":"Optimize Performance of Virtual Machine Checkpointing via Memory Exclusion","authors":"Haikun Liu, Hai Jin, Xiaofei Liao","doi":"10.1109/ChinaGrid.2009.42","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.42","url":null,"abstract":"Virtual Machine (VM) level checkpoints bring several advantages which process-level checkpoint implementation can hardly provide: compatibility, transparence, flexibility, and simplicity. However, the size of VM-level checkpoint may be very large and even in the order of gigabytes. This disadvantage causes the VM checkpointing and restart time become very long. To reduce the size of VM checkpoint, this paper proposes a memory exclusion scheme using ballooning mechanism, which omits saving unnecessary free pages in the VM. We implement our prototype in Xen environment. Experimental measurements show our approach can significantly reduce the size of VM checkpoint with minimal runtime overhead, thereby greatly improve the checkpoint performance.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130511326","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 : 2009-08-21DOI: 10.1109/ChinaGrid.2009.33
Hao Long, Hai Lu, Ruihua Di
Combining ontology with grid technology and applying them into the structure engineering is an inevitable trend and demand for engineers and researchers. Concerned with the semantic conflicts during resource integration and sharing, an ontology-based framework of the structure engineering grid is introduced, a method for designing the ontology model of the structure engineering grid is proposed, which consists of three types of ontology concept: Data types of Ontology, Core Ontology and Grid service Ontology, and the particular description of their functions and relations are given. The framework and the model are flexible and suitable for structural experimental resource integrating and searching.
{"title":"Research of Ontology Modeling in Structure Engineering Grid","authors":"Hao Long, Hai Lu, Ruihua Di","doi":"10.1109/ChinaGrid.2009.33","DOIUrl":"https://doi.org/10.1109/ChinaGrid.2009.33","url":null,"abstract":"Combining ontology with grid technology and applying them into the structure engineering is an inevitable trend and demand for engineers and researchers. Concerned with the semantic conflicts during resource integration and sharing, an ontology-based framework of the structure engineering grid is introduced, a method for designing the ontology model of the structure engineering grid is proposed, which consists of three types of ontology concept: Data types of Ontology, Core Ontology and Grid service Ontology, and the particular description of their functions and relations are given. The framework and the model are flexible and suitable for structural experimental resource integrating and searching.","PeriodicalId":212445,"journal":{"name":"2009 Fourth ChinaGrid Annual Conference","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116769496","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：