Carrier grade high availability platforms are designed t o enable the development and deployment of highly available services in the telecommunications industry. In order to build-in high availability and compare availabilities that differ in the sixth decimal place during the design phase, fairly detailed stochastic models are needed t o evaluate the design and perform design trade offs. Thas paper describes an availability model for a high availability platform using three-level hierarchical decomposition that mixes reliability block diagrams and Markov chains. The model is built and evaluated using the SHARPE software package. Sensitivity analysis is performed to identify the effects of critical parameters.