A class of computer systems whose primary task is the massive processing of batch transitions and which are called transaction-driven computer systems (TDCSs) is modeled and analyzed. A generic queuing model for transaction-driven computer systems is presented, and the mean sojourn time experienced by the different transactions is calculated. The approach is to model a TDCS by a cyclic polling system with bulk arrivals, deterministic service times, limited-one service, and zero switch-over periods. Since the performance analysis of this polling model has not been provided before, the emphasis is on deriving mean delay approximations for it. The analysis is carried out for models with general switch-over periods, and a special case of it (zero switch-over periods) is suitable for analyzing a TDCS.< >

The analysis of waiting times in polling systems in which the stations are polled according to a general service-order table is discussed. Such systems can be used to represent token-bus local area networks in which the routing of the token is fixed. Stations are given higher priority by being listed more frequently in the table, or by receiving service according to the exhaustive service discipline. The polling system is modeled by a single-server multiqueue system in discrete time. Nonzero switchover times between the queues are assumed. An extension of the principle of work conservation to systems with nonzero switchover times leads to an exact expression for a weighted sum of the mean waiting times at the various queues. By using a limiting procedure, the discrete-time results are translated to continuous-time results. The special case of polling in a star network is discussed and compared to polling in a corresponding network with strictly cyclic service order.>

Single-served, multiqueue systems with cyclic service in discrete time are considered. Nonzero switchover times between consecutive queues are assumed; the service strategies at the various queues may differ. A decomposition for the amount of work in such systems is obtained, leading to an exact expression for a weighted sum of the mean waiting times at the various queues.>