Conventional approaches to system availability assurance usually revolve around reliability-redundancy allocation and spare parts provisioning. This paper proposes a new scheme called redundancy sharing to enhance the availability of k-out-of-n:G systems. Under the new scheme, hot standby components in one system can be transferred to other systems for which the number of functional components drops below k. Although redundancy sharing is frequently practiced in private and defense industries, the underlying theory is rarely investigated in literature. This paper makes an early attempt to estimate the system availability considering redundancy sharing, and further discuss its potential benefits to the overall fleet availability. We strive to achieve two objectives: 1) estimating system availability by jointly considering redundancy sharing, redundancy allocation, and spare parts inventory; and 2) demonstrating the effectiveness and benefit of redundancy sharing in a variety of operating conditions. Continuous-time Markov model and multi-echelon inventory theory are employed to characterize the operation and maintenance of small and large fleets, respectively.