Compatibility is a decisive and necessary factor which impacts the successfulness and correctness of establishing connections among services. Reliable service parameter predictions on compatibility may not only provide assurance on fulfilling workflow, but also reduce the risk of unexpected wasting on time and expenses. Recent research on service's parameter prediction focus on two aspects: Collaborative Filtering and Matrix Factorization. Both of them faces challenges from data-sparse problem, lack of solid mathematical support, and weakness on tracking dynamic services. In this paper, we propose Link-ability, a parameter to reveal the compatibility of a service. Firstly, we propose an architecture called Topology-Retrievable Service Oriented Architecture (TSOA), which is capable of collecting successful invocation records from service requesters and generate the whole topology of the service pool in a specific period of time. Secondly, we bring up the Link-ability Generation Algorithm (LGA) and Markovian mathematical strategies to generate Link-ability value. The algorithm is supported by solid mathematical proof and capable of solving datasparse problems. Both TSOA and LGA are upgraded to adapt the circumstances of services exiting the service pool (we define it as a Drop-out). Lastly, we design and conduct two series of experiment to strengthen the reliability and correctness of Link-ability. The result shows that Link-ability considers comprehensive information of the whole topology and bring 54.18% reduction in average error rate against Drop-out scenarios.