Workfile disk management for concurrent mergesorts in a multiprocessor database system

This paper studies workfile disk management for concurrent mergesorts ina m ultiprocessor database system. Specifically, we examine the impacts of work file disk allocation and data striping on the average mergesort response ti me. Concurrent mergesorts in a multiprocessor system can creat severe I/O i nterference in which a large number of sequential write requests are contin uously issued to the same workfile disk and block other read requests for a long period of time. We examine through detailed simulations a logical par titioning approach to workfile disk management and evaluate the effectivene ss of datastriping. The results show that (1) without data striping, the be st performance is achieved by using the entire workfile disks as a single p artition if there are abundant workfile disks (or system workload is light) ; (2) however, if there are limited workfile disks (or system workload is h eavy), the workfile disks should be partitioned into multiple groups and th e optimal partition size is workload dependent; (3) data striping is benefi cial only if the striping unit size is properly chosen; and (4) with a prop er striping size, the best performance is generally achieved by using the e ntire disks as a single logical partition.