Pub Date : 2002-08-07DOI: 10.1109/ISPAN.2002.1004276
P. Uthayopas, T. Angskun, Jullawadee Maneesilp
Since their beginning in a NASA project around 1994, Beowulf clusters have become one of the most important classes of high performance computing systems. Although this platform has been widely used for several years, it is still difficult to build a large Beowulf cluster. The reason for this is the lack of tools that allow people to effectively maintain the system and develop applications. This problem significantly reduces the benefit gained from a Beowulf system. In this paper, the design of the next generation integrated environment for Beowulf clusters, called SCE (Scalable Cluster Environment), is presented. The goal of the SCE project is to build a powerful, scalable, portable, and easy to use cluster environment. The paper discusses the important issues in designing a good cluster environment and proposes a solution to these issues. SCE design has been used to demonstrate the ideas. Finally, the experimental results are presented to illustrated that while it provides a flexible and portable environment, SCE can also provide a scalable, fast, and efficient cluster environment for users.
{"title":"On the building of the next generation integrated environment for Beowulf clusters","authors":"P. Uthayopas, T. Angskun, Jullawadee Maneesilp","doi":"10.1109/ISPAN.2002.1004276","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004276","url":null,"abstract":"Since their beginning in a NASA project around 1994, Beowulf clusters have become one of the most important classes of high performance computing systems. Although this platform has been widely used for several years, it is still difficult to build a large Beowulf cluster. The reason for this is the lack of tools that allow people to effectively maintain the system and develop applications. This problem significantly reduces the benefit gained from a Beowulf system. In this paper, the design of the next generation integrated environment for Beowulf clusters, called SCE (Scalable Cluster Environment), is presented. The goal of the SCE project is to build a powerful, scalable, portable, and easy to use cluster environment. The paper discusses the important issues in designing a good cluster environment and proposes a solution to these issues. SCE design has been used to demonstrate the ideas. Finally, the experimental results are presented to illustrated that while it provides a flexible and portable environment, SCE can also provide a scalable, fast, and efficient cluster environment for users.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121269735","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004294
Tao Liu, Qiuyang Li, Yulu Yang
On the basis of the idea of the Delta network and the Baseline network, a new type of MIN, Advanced Baseline network, which using 4/spl times/4 switches, is proposed. In this paper it is proved that 4/spl times/4 switches have the better performance/cost ratios than 2/spl times/2 switches based on the current level of the VLSI technology. This paper expounds its topological properties and routing techniques and makes performance/ratios comparisons. And furthermore a fault-tolerance Advanced Baseline network is discussed in detail, which has a good reliability in complex parallel systems.
{"title":"Advanced Baseline: a new MIN with fault-tolerance characteristic","authors":"Tao Liu, Qiuyang Li, Yulu Yang","doi":"10.1109/ISPAN.2002.1004294","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004294","url":null,"abstract":"On the basis of the idea of the Delta network and the Baseline network, a new type of MIN, Advanced Baseline network, which using 4/spl times/4 switches, is proposed. In this paper it is proved that 4/spl times/4 switches have the better performance/cost ratios than 2/spl times/2 switches based on the current level of the VLSI technology. This paper expounds its topological properties and routing techniques and makes performance/ratios comparisons. And furthermore a fault-tolerance Advanced Baseline network is discussed in detail, which has a good reliability in complex parallel systems.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126942431","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004256
S. Iyengar, C. Wu
In recent years, the increasing sophistication of detection and tracking systems with multiple sensors has generated a great deal of interest in the development of new computational structures and strategies. The design of such spatially distributed systems involves the integration of solutions to the problems in data-association, hypothesis testing, data fusion, etc. More importantly, the design of efficient algorithms for these problems becomes all important issue as numerous distributed computational framework models are developed. In this paper, we present a survey of the recent research work conducted on the distributed sensor networks (DSNs) ranging from multi-sensor fusion methods, through target location, complexities of sensor deployment, probabilistic optimal surveillance to mobile agent based DSN (MADSN).
{"title":"Computational aspects of distributed sensor networks","authors":"S. Iyengar, C. Wu","doi":"10.1109/ISPAN.2002.1004256","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004256","url":null,"abstract":"In recent years, the increasing sophistication of detection and tracking systems with multiple sensors has generated a great deal of interest in the development of new computational structures and strategies. The design of such spatially distributed systems involves the integration of solutions to the problems in data-association, hypothesis testing, data fusion, etc. More importantly, the design of efficient algorithms for these problems becomes all important issue as numerous distributed computational framework models are developed. In this paper, we present a survey of the recent research work conducted on the distributed sensor networks (DSNs) ranging from multi-sensor fusion methods, through target location, complexities of sensor deployment, probabilistic optimal surveillance to mobile agent based DSN (MADSN).","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125505322","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004297
W. Jia, B. Han, Pui-on Au, Yonggang He, Wanlei Zhou
This paper proposes a set of novel efficient multicast routing algorithms on hypercube interconnections of cluster computers using a multicast shared-tree approach. We present some new algorithms for selecting an optimal core (root) for constructing the shared tree so as to minimize the average delay for multicast messages. Simulation results indicate that our algorithms are efficient in the sense of short end-to-end average delay, load balance and lower resource utilization over hypercube cluster connections.
{"title":"Optimal multicast tree routing for cluster computing in hypercube interconnection networks","authors":"W. Jia, B. Han, Pui-on Au, Yonggang He, Wanlei Zhou","doi":"10.1109/ISPAN.2002.1004297","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004297","url":null,"abstract":"This paper proposes a set of novel efficient multicast routing algorithms on hypercube interconnections of cluster computers using a multicast shared-tree approach. We present some new algorithms for selecting an optimal core (root) for constructing the shared tree so as to minimize the average delay for multicast messages. Simulation results indicate that our algorithms are efficient in the sense of short end-to-end average delay, load balance and lower resource utilization over hypercube cluster connections.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114888555","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004260
D. Taniar, Rebecca Boon-Noi Tan
Data-cube queries containing aggregate functions often combine multiple tables through join operations. We can extend this to "multi-join expansion-aggregate" data-cube queries by using more than one aggregate function in a "SELECT" statement in conjunction with relational operators. In parallel processing for such queries, it must be decided which attribute to use as a partitioning attribute, in particular the join attribute or "cube-by". Based on the partitioning attribute, we introduce three parallel multi-join expansion-aggregate data-cube query methods, namely the multi-join partition method (MPM), the expansion partition method (EPM) and the "early expansion partition with replication" method (EPRM). All three methods use the join attribute and "cube-by" as the partitioning attribute. A performance evaluation of the three parallel processing methods is also carried out and presented.
{"title":"Parallel processing of multi-join expansion-aggregate data cube query in high performance database systems","authors":"D. Taniar, Rebecca Boon-Noi Tan","doi":"10.1109/ISPAN.2002.1004260","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004260","url":null,"abstract":"Data-cube queries containing aggregate functions often combine multiple tables through join operations. We can extend this to \"multi-join expansion-aggregate\" data-cube queries by using more than one aggregate function in a \"SELECT\" statement in conjunction with relational operators. In parallel processing for such queries, it must be decided which attribute to use as a partitioning attribute, in particular the join attribute or \"cube-by\". Based on the partitioning attribute, we introduce three parallel multi-join expansion-aggregate data-cube query methods, namely the multi-join partition method (MPM), the expansion partition method (EPM) and the \"early expansion partition with replication\" method (EPRM). All three methods use the join attribute and \"cube-by\" as the partitioning attribute. A performance evaluation of the three parallel processing methods is also carried out and presented.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"50 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132939726","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004262
M. Palis
Investigates the job scheduling problem within the context of reservation-based real-time systems that provide quality of service (QoS) guarantees. In such a system, each incoming job specifies a rate of progress requirement on the job execution that must be met by the system in order for computation to be deemed usable. A new metric, called granularity, is introduced that quantifies both the maximum slowdown and the variance in execution rate that the job allows. This metric generalizes the stretch metric used in recent research on job scheduling. An online pre-emptive scheduling algorithm is presented that is shown achieve a competitive ratio of g(1-r) for every set of jobs with maximum rate r and granularity g. This result generalizes a previous result based on the stretch metric that showed that a competitive ratio of 1-r is achievable for the case when g=1.
{"title":"Online real-time job scheduling with rate of progress guarantees","authors":"M. Palis","doi":"10.1109/ISPAN.2002.1004262","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004262","url":null,"abstract":"Investigates the job scheduling problem within the context of reservation-based real-time systems that provide quality of service (QoS) guarantees. In such a system, each incoming job specifies a rate of progress requirement on the job execution that must be met by the system in order for computation to be deemed usable. A new metric, called granularity, is introduced that quantifies both the maximum slowdown and the variance in execution rate that the job allows. This metric generalizes the stretch metric used in recent research on job scheduling. An online pre-emptive scheduling algorithm is presented that is shown achieve a competitive ratio of g(1-r) for every set of jobs with maximum rate r and granularity g. This result generalizes a previous result based on the stretch metric that showed that a competitive ratio of 1-r is achievable for the case when g=1.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115851145","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004293
K. Hiraki, M. Inaba, Junji Tamatsukuri, Ryutaro Kurusu, Yukichi Ikuta, Hisashi Koga, Akira Zinzaki
We propose a very fast file sharing facility "data reservoir", which can transfer huge data files between far distant places, and can be used as an ordinary NFS server in a local area network. We use a low-level protocol and a hierarchical disk striping technique. We show the experimental results of a prototype model, which transfers a single huge data file with nearly 4 Gbps speed.
{"title":"Data reservoir: a new approach to data-intensive scientific computation","authors":"K. Hiraki, M. Inaba, Junji Tamatsukuri, Ryutaro Kurusu, Yukichi Ikuta, Hisashi Koga, Akira Zinzaki","doi":"10.1109/ISPAN.2002.1004293","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004293","url":null,"abstract":"We propose a very fast file sharing facility \"data reservoir\", which can transfer huge data files between far distant places, and can be used as an ordinary NFS server in a local area network. We use a low-level protocol and a hierarchical disk striping technique. We show the experimental results of a prototype model, which transfers a single huge data file with nearly 4 Gbps speed.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133195415","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004299
J. Rahayu, D. Taniar
Index is an important element in databases, and the existence of index is unavoidable. When an index has been built on a particular attribute, database operations (e.g. selection, join) on this attribute will become more efficient by utilizing the index. We focus on parallel algorithms for selection queries involving index that is data searching on indexed attributes. We propose two categories of parallel selection queries using index: parallel exact match and range selections; depending on the type of selection conditions. As parallel algorithms for these selection queries are very much influenced by indexing schemes, we also describe various index partitioning methods for parallel databases, and discuss their efficiency in supporting parallel selection query processing.
{"title":"Parallel selection query processing involving index in parallel database systems","authors":"J. Rahayu, D. Taniar","doi":"10.1109/ISPAN.2002.1004299","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004299","url":null,"abstract":"Index is an important element in databases, and the existence of index is unavoidable. When an index has been built on a particular attribute, database operations (e.g. selection, join) on this attribute will become more efficient by utilizing the index. We focus on parallel algorithms for selection queries involving index that is data searching on indexed attributes. We propose two categories of parallel selection queries using index: parallel exact match and range selections; depending on the type of selection conditions. As parallel algorithms for these selection queries are very much influenced by indexing schemes, we also describe various index partitioning methods for parallel databases, and discuss their efficiency in supporting parallel selection query processing.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132298473","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004292
S. See
The Internet is quickly evolving into a global computing platform. Internetworking of computers means that computing resources available for personal use need not be confined to the users' local environment. Instead, the Internet provides a delivery channel where remote computing resources are easily accessible. Computer simulations often require High Performance Computers (HPC) for fast number crunching. However, HPCs remain inaccessible to ordinary users due to the high cost involved. The aim of this project is to develop a Web-based simulation architecture where users can input a simulation model, based on the European Space Agency Simulation Language (ESL), through a Java graphical front-end application, and letting backend HPCs to perform intensive computation on behalf of the user. We have successfully implemented this system within an intranet environment, with the possibility of actual implementation on the World Wide Web. In this report, we discuss the architectural merits of this framework, our design philosophy and proposals for improvements and future development.
{"title":"A distributed simulation management engine for ordinary differential equations","authors":"S. See","doi":"10.1109/ISPAN.2002.1004292","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004292","url":null,"abstract":"The Internet is quickly evolving into a global computing platform. Internetworking of computers means that computing resources available for personal use need not be confined to the users' local environment. Instead, the Internet provides a delivery channel where remote computing resources are easily accessible. Computer simulations often require High Performance Computers (HPC) for fast number crunching. However, HPCs remain inaccessible to ordinary users due to the high cost involved. The aim of this project is to develop a Web-based simulation architecture where users can input a simulation model, based on the European Space Agency Simulation Language (ESL), through a Java graphical front-end application, and letting backend HPCs to perform intensive computation on behalf of the user. We have successfully implemented this system within an intranet environment, with the possibility of actual implementation on the World Wide Web. In this report, we discuss the architectural merits of this framework, our design philosophy and proposals for improvements and future development.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133902314","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004300
Cheng-Fa Tsai, Hang-Chang Wu, Chun-Wei Tsai
Clustering is the unsupervised classification of patterns (data item, feature vectors, or observations) into groups (clusters). Clustering in data mining is very useful to discover distribution patterns in the underlying data. Clustering algorithms usually employ a distance metric-based similarity measure in order to partition the database such that data points in the same partition are more similar than points in different partitions. In this paper, we present a new data clustering method for data mining in large databases. Our simulation results show that the proposed novel clustering method performs better than a fast self-organizing map (FSOM) combined with the k-means approach (FSOM+k-means) and the genetic k-means algorithm (GKA). In addition, in all the cases we studied, our method produces much smaller errors than both the FSOM+k-means approach and GKA.
{"title":"A new data clustering approach for data mining in large databases","authors":"Cheng-Fa Tsai, Hang-Chang Wu, Chun-Wei Tsai","doi":"10.1109/ISPAN.2002.1004300","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004300","url":null,"abstract":"Clustering is the unsupervised classification of patterns (data item, feature vectors, or observations) into groups (clusters). Clustering in data mining is very useful to discover distribution patterns in the underlying data. Clustering algorithms usually employ a distance metric-based similarity measure in order to partition the database such that data points in the same partition are more similar than points in different partitions. In this paper, we present a new data clustering method for data mining in large databases. Our simulation results show that the proposed novel clustering method performs better than a fast self-organizing map (FSOM) combined with the k-means approach (FSOM+k-means) and the genetic k-means algorithm (GKA). In addition, in all the cases we studied, our method produces much smaller errors than both the FSOM+k-means approach and GKA.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133221205","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：