Proactive and reactive computer network maintenance policies: accounting and risk

摘要

This study analyses factors affecting investment decisions concerning end-user support based on information on accounting and failure statistics. Bayesian tests have been developed to evaluate the vendor mean time between failure figures of communication concentrators. Bayesian posterior failure distributions have been estimated both in the deterministic and random failure intensity cases, the last of which results in a simple binomial distribution with an annual mean of 16 communication concentrator failures and a standard deviation of five failures. The cost parameters and failure statistics suggest that proactive costs of k?92 compared with average reactive costs of k?96 annually support a network monitoring investment of k?35, but no additional ad hoc preventions. Results for the day-to-day decision-making in the data processing management of a mobile phone company have been analysed. These can also be generalised for other equipment and even more comprehensive problem analyses enabled and limited by accounting and failure statistics information support.