K. A. Hua, W. Tavanapong, and H. C. Young. A Performance Evaluation of Load Balancing Techniques for Join Operations on Multicomputer Database systems. In P. S. Yu and A. L. P. Chen, editors, Proceedings of the 11th International Conference on Data Engineering, pages 44-51, Los Alamos, CA, USA, 1995. IEEE Computer Society Press.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Parallelism. 2 PDBMS : Parallel Database Management Systems. 3 OLAP : On Line Analytical Processing. load among processors while minimizing local computation and communication costs inherent to multi processors machines. Data skew can have a disastrous effect on parallel performance [4, 13, 2, 9, 7] and parallelism can only maintain acceptable performance through efficient algorithms realizing complex queries on dynamic, irregular and distributed data. Such algorithms must dynamically reduce or eliminate data skew at the lowest possible cost. Our contributions to materialized view ....

K. A. Hua, W. Tavanapong, and H. C. Young. A Performance Evaluation of Load Balancing Techniques for Join Operations on Multicomputer Database systems. In Proc. of the 11th International Conference on Data Engineering, CA, USA, 1995.

....algorithm is more appropriate. A dynamic load balancing technique is proposed in [ZJM94] where periodic checkpoints are done during the evaluation process to detect that actual hash function output. If needed, some of the tasks assigned to overloaded sites are redistributed to the others. In [HTY95], both previous strategies are implemented in a nCUBE 2 parallel computer and practical results are compared, where the method in [DNS 92] is shown to be superior. In [WDY 94] some of the previous results for sort merge and hash join algorithms proposed in previous works are reviewed so to ....

K.A. Hua, W. Tavanapong and H.C. Young, "A Performance Evaluation of Load Balancing Techniques for Join Operations on Multicomputer Database Systems", Procs. IEEE Intl. Conf. Data Engineering, 1995, pp 44--51.

....tree by sampling the relations. The optimization and the practical query could be done in parallel under an MPP environment. Although the optimization based on sampling is also utilized by conventional relational database systems, it is for the distribution of the hash buckets during the hash join[5]. Parade adopts right deep linear tree for the TREAT network. Because of the exponential increases in combinations are required for a bushy tree, it could not be optimized within a reasonable time period in dealing with a large query tree. Our optimization strategy is shown as follows: Phase 1: ....

Kien A. Hua, Wallapak Tavanapong, and Honesty C. Young. A Performance Evaluation of Load Balancing Techniques for Join Operations on Multicomputer Database Systems. In Proc. of ICDE '95, pages 44--51, March 1995.

....this cost is insignificant compared to the savings in the execution time overhead from load balancing. 6 Earlier Approaches to Load Balancing Most of the earlier approaches can be classified into two categories Conventional and Sampling Based. Based on earlier analyses of these algorithms [HY95] and our own studies, we have identified the best proposed solutions in both classes. Their split table computation phases are described below. 6.1 Conventional : Extended Adaptive Load Balancing Parallel Hash Join (ABJ ) This algorithm was proposed in [HL91] Each node scans and partitions ....

Kien A. Hua and Honesty C. Young. A performance evaluation of load balancing techniques for join operations on multicomputer database systems. Proc. of IEEE Conf. on Data Engineering, 1995.

....In the former, a task is statically assigned to one of the processor for the entire computation. The latter allows for immigration of the tasks to other processors during the join process. Several parallel join algorithms have been proposed to alleviate the presence of the data skew; i. e [10, 11, 7, 18, 26, 14, 27, 9, 20, 5, 21]. Most of the proposed algorithms are hash based algorithms. In the rest of this section, we review a few of the proposed algorithms. Bucket tuning was introduced in [10] In this strategy, the number of the buckets of one of the relations is chosen to be very large. In the later phases, smaller ....

....tuning strategies use the cardinality of the partitions buckets as a load metric. The above algorithms are sensitive to the output skew and expected to perform poorly in the presence of mild or high output skew. An incremental hash based algorithm has been proposed and improved in [27] and [9], respectively. In this approach, the join process proceeds in several steps. A check point strategy is used after each step to either evaluate the load degree (at the end of the first step) or apply partition tuning using the cardinality of the partially full buckets to change the buckets ....

K. A. Hua, W. Tavavapong, and H. Young. A Performance Evaluation of Load Balancing techniques for Join Operations on Multicomputer Database Systems. IEEE 11th Int'l Conference on Data Engineering, 1995.

No context found.

K. A. Hua, W. Tavanapong, and H. C. Young. A Performance Evaluation of Load Balancing Techniques for Join Operations on Multicomputer Database systems. In P. S. Yu and A. L. P. Chen, editors, Proceedings of the 11th International Conference on Data Engineering, pages 44-51, Los Alamos, CA, USA, 1995. IEEE Computer Society Press.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC