A Study of the Reliability of Internet Sites

Modeling the reliability of distributed systems requires a good understanding of the reliability of the components. Careful modeling allows highly fault-tolerant distributed data applications to be constructed at the least cost. Failure and repair rates of components are o f e n assumed to be exponentially distributed. This hypothesis is testable f o r failure rates, though the process of gathering and reducing the data to U usable form can be dificult. By applying an appropriate test statistic, some sumples were found to have a realistic chance of being drawn from an exponential distribution, while others can be confidently classed as non-exponential. Data were collectedfrom a large number of hosts via the Internet. Almost all of the visible Internet (over 350,000 hosts) were considered, and more than 68,000 of these that were judged likely to respond were queried. These hosts were sampled several times to obtain up-times, and finally to determine averuge host availability. Estimates of availability, mean-time-to-failure (MTTF) and mean-timeto-repair (MTTR) were derived. The results reported here correspond with those commonly seen in practice.