Complex software-intensive applications can be built with commercially available systems for cluster systems. To improve availability of (n,k)-way cluster systems, we develop self-configuring algorithm that not only determines the number of primary and backup nodes for meeting the requirement of availability and waiting time deadline, but also uses software rejuvenation for dealing with dormant software faults. Availability modeling of (n,k)-way cluster systems with software rejuvenation has a view of fault tolerance and switchover states with a semi-Markov process. According to the operating parameters, steady-state probabilities and availability are calculated, which are used for self-configuring algorithm.
展开▼