Skip to Main Content
Survival and revenue maximisation result in pressures to cut costs and avoid adequate reliability tests. Serious errors also arise from ignorance about the reliability behaviour of manufactured products. Desk-based "simulations" alarmingly use theoretical text-book "data" to predict reliability and allow release of non-proven products. Some current reliability "standards" deliver a very poor calibre of evidence. "Qualification" to these standards provides inadequate evidence for high reliability systems. Such products are time-bombs when installed in "high reliability" networks. The networks are therefore may more robust at added cost by system-level fault tolerance. Practical solutions to assuring the reliability of the products, even where Telcordia specifications are used are through practical reliability evidence from the "building block" elements that make up the product. Reliability tests are more informative than "Qualification" and generate information on reliability behaviour, FIT rates and the essential statistics needed by those deploying systems in the field. Typical Qualification tests generate only crude estimates of high wearout FIT rates, at levels around 17000, even if the product has a genuinely low FIT rate, whereas Reliability test data yields wearout FIT rates lower than 500.
Date of Conference: 7-11 May 2008