Walton, C., et al., "A Taxonomy and Performance Model of Data Skew in Parallel Joins",<F3.17e+05> Proc. VLDB<F3.733e+05> Conf., Barcelona, Spain, September 1991.  Home/Search   Document Not in Database   Summary   Related Articles

....should allow for the use of flexible workload isolation mechanisms that can adapt to workload changes. 3 Assuming adequate network bandwidth to eliminate network contention. 3. 4 Skew The variance in response times across the multiple instances of a parallel relational operator is called skew ([Wolf90, Hua91, Kits90, Omie91, Walt91, DeWi92b]) Since the degree of declustering determines the degree of parallelism of operators like select, skew is an important factor when selecting the degree of declustering for a relation. A taxonomy of data skew in parallel databases was presented in [Walt91] Four types of skew were identified: skew ....

....( Wolf90, Hua91, Kits90, Omie91, Walt91, DeWi92b] Since the degree of declustering determines the degree of parallelism of operators like select, skew is an important factor when selecting the degree of declustering for a relation. A taxonomy of data skew in parallel databases was presented in [Walt91]. Four types of skew were identified: skew in initial tuple placement (tuple placement skew) skew due to variations in predicate selectivity across nodes (selectivity skew) skew in redistribution of tuples in preparation for a join (redistribution skew) and finally, imbalance in the number of ....

[Article contains additional citation context not shown here]

Walton, C., et al., "A Taxonomy and Performance Model of Data Skew in Parallel Joins",<F3.17e+05> Proc. VLDB<F3.733e+05> Conf., Barcelona, Spain, September 1991.

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