Pub Date : 1991-12-01DOI: 10.1109/PDIS.1991.183102
Carl D. Tait, D. Duchamp
Portable computers are now common, a fact that raises the possibility that file service clients might move on a regular basis. This new development requires rethinking some features of distributed file system design. The authors argue that existing approaches to file replica management would not cope well with the likely behavior of mobile clients, and they present their solution; a lazy server-based update operation. This operation facilitates fast, scalable, and highly fault tolerant implementations of both read and write operations in the usual case. To cope with the weak semantics of the update operation, the authors propose a new file system service interface that allows applications to opt for UNIX semantics by use of a slower, less fault-tolerant read operation.<>
{"title":"Service interface and replica management algorithm for mobile file system clients","authors":"Carl D. Tait, D. Duchamp","doi":"10.1109/PDIS.1991.183102","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183102","url":null,"abstract":"Portable computers are now common, a fact that raises the possibility that file service clients might move on a regular basis. This new development requires rethinking some features of distributed file system design. The authors argue that existing approaches to file replica management would not cope well with the likely behavior of mobile clients, and they present their solution; a lazy server-based update operation. This operation facilitates fast, scalable, and highly fault tolerant implementations of both read and write operations in the usual case. To cope with the weak semantics of the update operation, the authors propose a new file system service interface that allows applications to opt for UNIX semantics by use of a slower, less fault-tolerant read operation.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126095377","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183113
K. Hua, Chiang Lee, J. Peir
The most debated architectures for parallel database processing are Shared Nothing (SN) and Shared Everything (SE) structures. Although SN is considered to be most scalable, it is very sensitive to the data skew problem. On the other hand, SE allows the collaborating processors to share the work load more efficiently. It, however, suffers from the limitation of the memory and disk I/O band-width. The authors present a hybrid architecture in which SE clusters are interconnected through a communication network to form a SN structure at the inter-cluster level. In this approach, processing elements are clustered into SE systems to minimize the skew effect. Each cluster, however, is kept small within the limitation of the memory and I/O technology to avoid the data access bottleneck. A generalized performance model was developed to perform sensitivity analysis for the hybrid structure, and to compare it against SE and SN organizations.<>
{"title":"Interconnecting shared-everything systems for efficient parallel query processing","authors":"K. Hua, Chiang Lee, J. Peir","doi":"10.1109/PDIS.1991.183113","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183113","url":null,"abstract":"The most debated architectures for parallel database processing are Shared Nothing (SN) and Shared Everything (SE) structures. Although SN is considered to be most scalable, it is very sensitive to the data skew problem. On the other hand, SE allows the collaborating processors to share the work load more efficiently. It, however, suffers from the limitation of the memory and disk I/O band-width. The authors present a hybrid architecture in which SE clusters are interconnected through a communication network to form a SN structure at the inter-cluster level. In this approach, processing elements are clustered into SE systems to minimize the skew effect. Each cluster, however, is kept small within the limitation of the memory and I/O technology to avoid the data access bottleneck. A generalized performance model was developed to perform sensitivity analysis for the hybrid structure, and to compare it against SE and SN organizations.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134523994","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183079
K. Hua, S. Hannenhalli
Deals with parallel transitive closure computations. The sort-based approaches introduced sorts the tuples of the relation into topological order, and the sorted relation is then horizontally partitioned and distributed across several processing nodes of a message passing multiprocessor system. This data partitioning strategy allows the transitive closure computation of the local data fragments to be computed in parallel with no interprocessor communication. The construction of the final result then requires only a small number of joins. Extensive analytical results are included in the paper as well. They show that the proposed techniques leads to a speedup that is essentially linear with the number of processors. Its performance is significantly better than the recently published hashless parallel algorithm.<>
{"title":"Parallel transitive closure computations using topological sort","authors":"K. Hua, S. Hannenhalli","doi":"10.1109/PDIS.1991.183079","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183079","url":null,"abstract":"Deals with parallel transitive closure computations. The sort-based approaches introduced sorts the tuples of the relation into topological order, and the sorted relation is then horizontally partitioned and distributed across several processing nodes of a message passing multiprocessor system. This data partitioning strategy allows the transitive closure computation of the local data fragments to be computed in parallel with no interprocessor communication. The construction of the final result then requires only a small number of joins. Extensive analytical results are included in the paper as well. They show that the proposed techniques leads to a speedup that is essentially linear with the number of processors. Its performance is significantly better than the recently published hashless parallel algorithm.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126225914","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183068
A. M. Keller, S. Roy
Presents an algorithm for parallel hash-join computation on main-memory databases that adapts to data skew, and its implementation on the IBM RP3 multiprocessor. The algorithm exploits the random access capabilities of main memory databases to detect and counteract skew on the fly. Data skew is detected at run time by monitoring the observed frequencies of values of the join attribute and applying to them a threshold function that takes account of the distribution of workload among processors. If and when this threshold is reached for certain values of the join attribute, the computation corresponding to it is fragmented among an appropriate number of processors. Fragmentation requires some replication of input tuples-modestly increasing the total workload, but reduces the completion time significantly by reducing workload at the overloaded processor. A simplified analysis is supplemented by experiments. The description and analysis of the algorithm are based on the shared-nothing model. The implementation uses hierarchical shared memory providing non-uniform memory access.<>
{"title":"Adaptive parallel hash join in main-memory databases","authors":"A. M. Keller, S. Roy","doi":"10.1109/PDIS.1991.183068","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183068","url":null,"abstract":"Presents an algorithm for parallel hash-join computation on main-memory databases that adapts to data skew, and its implementation on the IBM RP3 multiprocessor. The algorithm exploits the random access capabilities of main memory databases to detect and counteract skew on the fly. Data skew is detected at run time by monitoring the observed frequencies of values of the join attribute and applying to them a threshold function that takes account of the distribution of workload among processors. If and when this threshold is reached for certain values of the join attribute, the computation corresponding to it is fragmented among an appropriate number of processors. Fragmentation requires some replication of input tuples-modestly increasing the total workload, but reduces the completion time significantly by reducing workload at the overloaded processor. A simplified analysis is supplemented by experiments. The description and analysis of the algorithm are based on the shared-nothing model. The implementation uses hierarchical shared memory providing non-uniform memory access.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126841290","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183103
A. Leff, Philip S. Yu, J. Wolf
The high performance networks in modern distributed systems enable a site to obtain cached objects from the main memory of other sites more quickly than the time needed to access local disks. In this environment, efficient mechanisms can be devised to support rapid request/response exchanges for objects that reside on remote sites. As a result, it becomes possible to use remote memory as an additional layer in the memory hierarchy between local memory and disks. The paper studies the performance potential of remote memory caching. A key issue is managing the of object replicas maintained by different sites. Although exploiting remote memory improves performance over a wide range of cache sizes as compared to a distributed client/server caching architecture, efficient use of remote memory can be difficult to achieve. The effect of several system parameters, and several cache policies are examined; their effect on both overall and local system performance is described.<>
{"title":"Policies for efficient memory utilization in a remote caching architecture","authors":"A. Leff, Philip S. Yu, J. Wolf","doi":"10.1109/PDIS.1991.183103","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183103","url":null,"abstract":"The high performance networks in modern distributed systems enable a site to obtain cached objects from the main memory of other sites more quickly than the time needed to access local disks. In this environment, efficient mechanisms can be devised to support rapid request/response exchanges for objects that reside on remote sites. As a result, it becomes possible to use remote memory as an additional layer in the memory hierarchy between local memory and disks. The paper studies the performance potential of remote memory caching. A key issue is managing the of object replicas maintained by different sites. Although exploiting remote memory improves performance over a wide range of cache sizes as compared to a distributed client/server caching architecture, efficient use of remote memory can be difficult to achieve. The effect of several system parameters, and several cache policies are examined; their effect on both overall and local system performance is described.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134435859","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183099
Ronald F. DeMara, D. Moldovan
The authors have been experimenting with knowledge representation and reasoning mechanisms suitable for parallel computation. Their primary domain is natural language understanding (NLU) using a semantic network representation and parallel marker-passing techniques. They have implemented a multiprocessor called the Semantic Network Array Processor (SNAP-1) for real-time processing of moderately-sized knowledge bases.<>
{"title":"Marker-passing on a parallel knowledge processing testbed","authors":"Ronald F. DeMara, D. Moldovan","doi":"10.1109/PDIS.1991.183099","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183099","url":null,"abstract":"The authors have been experimenting with knowledge representation and reasoning mechanisms suitable for parallel computation. Their primary domain is natural language understanding (NLU) using a semantic network representation and parallel marker-passing techniques. They have implemented a multiprocessor called the Semantic Network Array Processor (SNAP-1) for real-time processing of moderately-sized knowledge bases.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114130048","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183092
R. Ladin, M. Mazer, A. Wolman
Summary form only given. The authors are designing and implementing a tool for making certain types of distributed services highly available through replication. The project has three separable goals: (1) to provide a tool for developing highly available services by exploiting semantics to get good performance. (2) to understand the issues in building a correct, easy-to-use, generic replication tool; and (3) to show that such a tool can indeed yield excellent performance in terms of availability and response time (relative to a nonreplicated service and relative to a hand-crafted replicated service).<>
{"title":"A general tool for replicating distributed services","authors":"R. Ladin, M. Mazer, A. Wolman","doi":"10.1109/PDIS.1991.183092","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183092","url":null,"abstract":"Summary form only given. The authors are designing and implementing a tool for making certain types of distributed services highly available through replication. The project has three separable goals: (1) to provide a tool for developing highly available services by exploiting semantics to get good performance. (2) to understand the issues in building a correct, easy-to-use, generic replication tool; and (3) to show that such a tool can indeed yield excellent performance in terms of availability and response time (relative to a nonreplicated service and relative to a hand-crafted replicated service).<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125197462","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183061
Jon A. Solworth, C. Orji
The authors introduce distorted mirrors, a mirroring system which combines write anywhere semantics with traditional database-specified block locations. This technique radically reduces the cost of small writes; making it attractive for random access applications such as OLTP, while retaining the ability to efficiently perform large sequential accesses. Distorted mirror also scale better than traditional mirrors in terms of both disk caching and large mirrored sets. The effectiveness of distorted mirrors is shown on the TP1 benchmark.<>
{"title":"Distorted mirrors","authors":"Jon A. Solworth, C. Orji","doi":"10.1109/PDIS.1991.183061","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183061","url":null,"abstract":"The authors introduce distorted mirrors, a mirroring system which combines write anywhere semantics with traditional database-specified block locations. This technique radically reduces the cost of small writes; making it attractive for random access applications such as OLTP, while retaining the ability to efficiently perform large sequential accesses. Distorted mirror also scale better than traditional mirrors in terms of both disk caching and large mirrored sets. The effectiveness of distorted mirrors is shown on the TP1 benchmark.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126025174","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183093
Y. Yemini, O. Wolfson
Summary form only given. The overall architecture of NETMATE is the following. Observation and Control Points (OCPs) are supported via managing agents placed in networked devices. These agents can monitor local device behaviors and affect device control functions, when instructed by a manager. Managers communicate observation/control requests to agents via a manager-agent protocol. Two protocol standards are currently evolving: SNMP, used to manage Internets, and CMIP, proposed by OSI. The NETMATE manager is centered around a network Modeler database. The role of the modeler is to maintain a network Management Information Base (MIB) and provide active access services to dynamic operational network data. The manager presents network status data to users in a visual form. Users can navigate within a very large and complex network data space and control complex behaviors in a simple visual form. Finally, a collection of interpretation and diagnosis tools is used to analyze observed data, and automatically identify/classify fault conditions. The authors briefly describe the key problem domains for long term NETMATE research efforts.<>
{"title":"NETMATE: management of complex distributed networked systems","authors":"Y. Yemini, O. Wolfson","doi":"10.1109/PDIS.1991.183093","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183093","url":null,"abstract":"Summary form only given. The overall architecture of NETMATE is the following. Observation and Control Points (OCPs) are supported via managing agents placed in networked devices. These agents can monitor local device behaviors and affect device control functions, when instructed by a manager. Managers communicate observation/control requests to agents via a manager-agent protocol. Two protocol standards are currently evolving: SNMP, used to manage Internets, and CMIP, proposed by OSI. The NETMATE manager is centered around a network Modeler database. The role of the modeler is to maintain a network Management Information Base (MIB) and provide active access services to dynamic operational network data. The manager presents network status data to users in a visual form. Users can navigate within a very large and complex network data space and control complex behaviors in a simple visual form. Finally, a collection of interpretation and diagnosis tools is used to analyze observed data, and automatically identify/classify fault conditions. The authors briefly describe the key problem domains for long term NETMATE research efforts.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116167736","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 : 1991-12-01DOI: 10.1109/PDIS.1991.183074
J. Shao, D. Bell, M. Hull
There are two approaches to processing Datalog programs in parallel. One is to decompose the rules of a program into concurrent modules, and then assign them to processors. The other is to partition data between processors, so that each processor evaluates the same program, but with less data. The authors propose a third approach which combines the two methods in a single framework. In this approach, rules are decomposed into segments and data is partitioned among the segments. There are a number of advantages of this approach. Most importantly, it provides good focus on processing the tuples that are relevant to queries, and allows data to be partitioned and balanced dynamically at different levels. An analytic performance study is also presented to illustrate the usefulness of the proposed approach.<>
{"title":"Combining rule decomposition and data partitioning in parallel datalog program processing","authors":"J. Shao, D. Bell, M. Hull","doi":"10.1109/PDIS.1991.183074","DOIUrl":"https://doi.org/10.1109/PDIS.1991.183074","url":null,"abstract":"There are two approaches to processing Datalog programs in parallel. One is to decompose the rules of a program into concurrent modules, and then assign them to processors. The other is to partition data between processors, so that each processor evaluates the same program, but with less data. The authors propose a third approach which combines the two methods in a single framework. In this approach, rules are decomposed into segments and data is partitioned among the segments. There are a number of advantages of this approach. Most importantly, it provides good focus on processing the tuples that are relevant to queries, and allows data to be partitioned and balanced dynamically at different levels. An analytic performance study is also presented to illustrate the usefulness of the proposed approach.<<ETX>>","PeriodicalId":210800,"journal":{"name":"[1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1991-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117175768","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：