Computational algorithms for closed queueing networks with exponential servers
Communications of the ACM
Dynamic space-sharing in computer systems
Communications of the ACM
Analysis of system bottlenecks using a queueing network model
Proceedings of the SIGOPS workshop on System performance evaluation
Network models for large-scale time-sharing systems
Network models for large-scale time-sharing systems
Performance monitors for multi-programming systems
SOSP '69 Proceedings of the second symposium on Operating systems principles
Determining an Optimal Secondary Storage Service Rate for the PASM Control System
IEEE Transactions on Computers
Bibliography on paging and related topics
ACM SIGOPS Operating Systems Review
SIGMETRICS '82 Proceedings of the 1982 ACM SIGMETRICS conference on Measurement and modeling of computer systems
Hi-index | 0.00 |
In this paper, we shall use the same analytical model to investigate the behaviour of the paging drum when the degree of multiprogramming is set to its optimal value. The main conclusion, which corroborates the general feeling on performance of virtual memory systems [3, 4, 5], is that drum utilization remains in the 50% range whenever CPU utilization is maximized, if no resource is saturated. The results are validated through simulation experiments in order to relax some theoretical assumptions used in the analytical model and to take into account some detailed mechanisms such as CPU overheads. As an application, an adaptive control algorithm based on the activity of the paging device is proposed. The algorithm has been simulated and comparisons with results reported in [2] are presented. Moreover, after investigations we show also that optimal performance can only be achieved through a balanced use of the different resources of the system. This conclusion cross-checks and extends Buzen's theoretical results [6] and Wulf's practical ones [7].