This paper proposes a novel analytical framework to evaluate the quality of component-based software systems. This approach uses CEG graph (which is capable of modeling sequential, parallel, selective and iterative composition patterns) to model the execution process of component-based software system. Based on a homogeneous continuous-time Markovian process, analytical methods to evaluate the make-span, cost and reliability of CEG (the graph of component execution) are developed. Through a case study, we show that our framework is capable of modeling real-world component-based software systems. Also, the comparison between analytical and Monte-carlo simulative results indicates that our analytical model is consistent with simulation. The paper concludes with a comparison with related work