Modelling Network Memory Servers with Parallel Processors, Break-downs and Repairs
- 1 January 2007
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 11-20
Abstract
This paper presents an analytical method for the performability evaluation of a previously reported network memory server attached to a local area network. To increase the performance and availability of the proposed system, an additional server is added to the system. Such systems are prone to failures. With this in mind, a mathematical model has been developed to analyse the performability of the proposed system with break-downs and repairs. Mean queue lengths and the probability of job losses for the LAN feeding the network memory server is calculated and presentedKeywords
This publication has 8 references indexed in Scilit:
- Modelling and Performability Analysis of Network Memory Servers2006
- Cashmere-VLM: Remote memory paging for software distributed shared memory2003
- Should I add a processor? (performance evaluation)2002
- OceanStore2000
- The Network RamDisk: Using remote memory on heterogeneous NOWsCluster Computing, 1999
- ETAQA: an efficient technique for the analysis of QBD-processes by aggregationPerformance Evaluation, 1999
- Spectral expansion solution for some finite capacity queuesAnnals of Operations Research, 1998
- A Queueing Model with Finite Waiting Room and BlockingJournal of the ACM, 1976