Pub Date : 2002-08-18DOI: 10.1109/ICPP.2002.1040916
Barry Lawson, E. Smirni, D. Puiu
We focus on non-FCFS job scheduling policies for parallel systems that allow jobs to backfill, i.e., to move ahead in the queue, given that they do not delay certain previously submitted jobs. Consistent with commercial schedulers that maintain multiple queues where jobs are assigned according to the user-estimated duration, we propose a self-adapting backfilling policy that maintains multiple job queues to separate short from long jobs. The proposed policy adjusts its configuration parameters by continuously monitoring the system and quickly reacting to sudden fluctuations in the workload arrival pattern and/or severe changes in resource demands. Detailed performance comparisons via simulation using actual supercomputing, traces from the parallel workload archive indicate that the proposed policy consistently outperforms traditional backfilling.
{"title":"Self-adapting backfilling scheduling for parallel systems","authors":"Barry Lawson, E. Smirni, D. Puiu","doi":"10.1109/ICPP.2002.1040916","DOIUrl":"https://doi.org/10.1109/ICPP.2002.1040916","url":null,"abstract":"We focus on non-FCFS job scheduling policies for parallel systems that allow jobs to backfill, i.e., to move ahead in the queue, given that they do not delay certain previously submitted jobs. Consistent with commercial schedulers that maintain multiple queues where jobs are assigned according to the user-estimated duration, we propose a self-adapting backfilling policy that maintains multiple job queues to separate short from long jobs. The proposed policy adjusts its configuration parameters by continuously monitoring the system and quickly reacting to sudden fluctuations in the workload arrival pattern and/or severe changes in resource demands. Detailed performance comparisons via simulation using actual supercomputing, traces from the parallel workload archive indicate that the proposed policy consistently outperforms traditional backfilling.","PeriodicalId":393916,"journal":{"name":"Proceedings International Conference on Parallel Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125789289","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 : 2002-08-18DOI: 10.1109/ICPP.2002.1040904
Wai-Keung Wu, S. Majumdar
Inter-operability in heterogeneous distributed systems is often provided with the help of CORBA compliant middleware. Many distributed object computing systems, however, are characterized by limited heterogeneity. Such systems often contain a subset of components that are written in the same programming language and run on top of the same p-Worm. We present engineering techniques that exploit such limited heterogeneity in systems for achieving high system performance. With these techniques components implemented using diverse programming languages and/or platform use a CORBA compliant middleware, whereas the similar components can use a "flyover" that employs a separate path between the client and its server and avoids a number of CORBA overheads. Insights into system behavior and performance gained from results of experiments with synthetic workload running on a network of PCs are presented.
{"title":"Engineering CORBA-based systems for high performance","authors":"Wai-Keung Wu, S. Majumdar","doi":"10.1109/ICPP.2002.1040904","DOIUrl":"https://doi.org/10.1109/ICPP.2002.1040904","url":null,"abstract":"Inter-operability in heterogeneous distributed systems is often provided with the help of CORBA compliant middleware. Many distributed object computing systems, however, are characterized by limited heterogeneity. Such systems often contain a subset of components that are written in the same programming language and run on top of the same p-Worm. We present engineering techniques that exploit such limited heterogeneity in systems for achieving high system performance. With these techniques components implemented using diverse programming languages and/or platform use a CORBA compliant middleware, whereas the similar components can use a \"flyover\" that employs a separate path between the client and its server and avoids a number of CORBA overheads. Insights into system behavior and performance gained from results of experiments with synthetic workload running on a network of PCs are presented.","PeriodicalId":393916,"journal":{"name":"Proceedings International Conference on Parallel Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114474401","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 : 2002-08-18DOI: 10.1109/ICPP.2002.1040870
Jinggang Wang, B. Ravindran
In this paper, we present a MAC-layer packet scheduling algorithm, called BPA, for real-time switched Ethernet networks. BPA considers a message model where trans-node application-level messages have end-to-end timeliness requirements that are specified using Jensen's benefit functions. The objective of BPA is to maximize the aggregate message-level benefit. The algorithm reasons that this objective can be achieved by maximizing aggregate packet-level benefit, where packets of messages are allowed to inherit benefit functions of their parent messages. BPA thus solves a non-preemptive packet scheduling problem. Since this problem is NP-hard, BPA heuristically computes packet schedules to maximize aggregate benefit, incurring a worst-case computational complexity of O(n/sup 2/). This is better than the O(n/sup 3/) complexity of the previously known best algorithm (called CMA) for the same problem. Further, our experimental studies show that BPA performs as good as CMA for a broad set of benefit functions, and significantly outperforms CMA for some benefit functions. Furthermore, we observe that BPA yields lower missed-deadline ratio than CMA when message arrival density increases.
{"title":"BPA: a fast packet scheduling algorithm for real-time switched Ethernet networks","authors":"Jinggang Wang, B. Ravindran","doi":"10.1109/ICPP.2002.1040870","DOIUrl":"https://doi.org/10.1109/ICPP.2002.1040870","url":null,"abstract":"In this paper, we present a MAC-layer packet scheduling algorithm, called BPA, for real-time switched Ethernet networks. BPA considers a message model where trans-node application-level messages have end-to-end timeliness requirements that are specified using Jensen's benefit functions. The objective of BPA is to maximize the aggregate message-level benefit. The algorithm reasons that this objective can be achieved by maximizing aggregate packet-level benefit, where packets of messages are allowed to inherit benefit functions of their parent messages. BPA thus solves a non-preemptive packet scheduling problem. Since this problem is NP-hard, BPA heuristically computes packet schedules to maximize aggregate benefit, incurring a worst-case computational complexity of O(n/sup 2/). This is better than the O(n/sup 3/) complexity of the previously known best algorithm (called CMA) for the same problem. Further, our experimental studies show that BPA performs as good as CMA for a broad set of benefit functions, and significantly outperforms CMA for some benefit functions. Furthermore, we observe that BPA yields lower missed-deadline ratio than CMA when message arrival density increases.","PeriodicalId":393916,"journal":{"name":"Proceedings International Conference on Parallel Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116134846","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 : 2002-08-18DOI: 10.1109/ICPP.2002.1040905
D. Manivannan, M. Singhal
Knot detection in a distributed graph is an important problem and finds applications in several areas such as packet switching, distributed simulation, and distributed database systems. The paper presents a distributed algorithm to efficiently detect the existence of a knot in a distributed graph. The algorithm requires 2e messages and a delay or 2(d+1) message hops to detect if a node in a distributed graph is in a knot (e is the number of edges in the reachable part of the distributed graph and d is its diameter). A significant advantage of this algorithm is that it not only detects if a node is in a knot but also finds exactly which nodes are involved in the knot.
{"title":"A distributed algorithm for knot detection in a distributed graph","authors":"D. Manivannan, M. Singhal","doi":"10.1109/ICPP.2002.1040905","DOIUrl":"https://doi.org/10.1109/ICPP.2002.1040905","url":null,"abstract":"Knot detection in a distributed graph is an important problem and finds applications in several areas such as packet switching, distributed simulation, and distributed database systems. The paper presents a distributed algorithm to efficiently detect the existence of a knot in a distributed graph. The algorithm requires 2e messages and a delay or 2(d+1) message hops to detect if a node in a distributed graph is in a knot (e is the number of edges in the reachable part of the distributed graph and d is its diameter). A significant advantage of this algorithm is that it not only detects if a node is in a knot but also finds exactly which nodes are involved in the knot.","PeriodicalId":393916,"journal":{"name":"Proceedings International Conference on Parallel Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131985966","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 : 2002-08-18DOI: 10.1109/ICPP.2002.1040880
Jie Wu, Dajin Wang
We propose a deterministic fault-tolerant and deadlock-free routing protocol in 2D meshes based on Wu's fault-tolerant odd-even turn model (2000) and Wang's rectilinear-monotone polygonal fault block model. The fault-tolerant odd-even turn protocol, also called extended X-Y routing, was originally proposed to achieve fault-tolerant and deadlock-free routing among traditional, rectangular fault blocks. It does not use any virtual channels. The number of faults to be tolerated is unbounded as long as nodes outside fault blocks are connected in the mesh network. The recently proposed rectilinear-monotone polygonal fault blocks (also called minimal-connected-components or MCCs) are of the polygonal shapes, and is a refinement of rectangular fault blocks. The formation of MCCs depends on the relative locations of source and destination, and they include much fewer healthy nodes in resultant fault blocks. In this paper, we show that with a simple modification, the extended X-Y routing can also be applied to 2D meshes using extended MCCs.
{"title":"Fault-tolerant and deadlock-free routing in 2-D meshes using rectilinear-monotone polygonal fault blocks","authors":"Jie Wu, Dajin Wang","doi":"10.1109/ICPP.2002.1040880","DOIUrl":"https://doi.org/10.1109/ICPP.2002.1040880","url":null,"abstract":"We propose a deterministic fault-tolerant and deadlock-free routing protocol in 2D meshes based on Wu's fault-tolerant odd-even turn model (2000) and Wang's rectilinear-monotone polygonal fault block model. The fault-tolerant odd-even turn protocol, also called extended X-Y routing, was originally proposed to achieve fault-tolerant and deadlock-free routing among traditional, rectangular fault blocks. It does not use any virtual channels. The number of faults to be tolerated is unbounded as long as nodes outside fault blocks are connected in the mesh network. The recently proposed rectilinear-monotone polygonal fault blocks (also called minimal-connected-components or MCCs) are of the polygonal shapes, and is a refinement of rectangular fault blocks. The formation of MCCs depends on the relative locations of source and destination, and they include much fewer healthy nodes in resultant fault blocks. In this paper, we show that with a simple modification, the extended X-Y routing can also be applied to 2D meshes using extended MCCs.","PeriodicalId":393916,"journal":{"name":"Proceedings International Conference on Parallel Processing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126166692","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}