Quantitative system performance: computer system analysis using queueing network models
Quantitative system performance: computer system analysis using queueing network models
Probability, statistics, and queueing theory with computer science applications
Probability, statistics, and queueing theory with computer science applications
Open, Closed, and Mixed Networks of Queues with Different Classes of Customers
Journal of the ACM (JACM)
Some Extensions to Multiclass Queueing Network Analysis
Proceedings of the Third International Symposium on Modelling and Performance Evaluation of Computer Systems: Performance of Computer Systems
| Hi-index | 0.00 |
Transaction (open) type workloads are often used in approximating computer system workloads which are actually closed because open workloads provide reasonable estimates in many cases and their solutions are straight-forward. We have found that their use can distort the results for many workloads in a multiclass queueing network model of a computer system. We have replaced transaction workloads with what we call fixed class workloads. We present an approximate algorithm based on MVA that represents a class with a given throughput by a corresponding terminal or batch class, which we call a fixed class workload. We solve for the closed population required to deliver the requested throughput. We also present techniques for overcoming problems encountered in the solution of some fixed class models.