Is MTTR More Important Than MTTF For Improving User-Perceived Availability ?

High availability of internet systems can be achieved either through high mean-time-to-failure (MTTF) or low mean-time-to-recovery (MTTR). Traditionally, system designers have focused on maximizing MTTF as a way of providing high availability. However, recent work in recovery-oriented computing has emphasized recovery from failures, rather than failure avoidance [1]. In this paper, we investigate the impact of MTTR and MTTF, as well as user retry rates, on the user-perceived availability of internet systems through simulation experiments. Previous work suggests that between two systems with a given availability, the one with lower MTTR always yields higher userperceived availability [3]. Our results show that while this is true for certain ranges of system parameters, it is not always the case. This leads us to propose a method for determining whether improving MTTR or MTTF will yield greater user-visible benefits for a given system, allowing a system administrator to invest in system improvements that yield maximum benefits.