Our first goal is to provide a FCFS (first-come-first-served) queueing analysis sufficiently general to embrace the detailed structure of a large majority of existing systems. The parameters of the mathematical model include a stationary, discrete probability distribution describing the patterns by which requests address information on secondary storage devices, where successive addresses are also allowed to form a first-order Markov chain. Such patterns normally influence system performance, and they are determined by the mechanism that allocates specific storage locations to records (units of information). Thus, in the calculation of conventional performance measures we refer to essentially combinatorial problems in which the effects of different record allocations are studied.
KeywordsService Time Busy Period Idle Period Disk System Secondary Storage
Unable to display preview. Download preview PDF.