At present, common service systems are centralized. A centralized service system is attributed to a single service provider who controls the service correctness and availability as well as fairness. Decentralized computing tries to change this situation. The idea is that one service is provided by multiple service providers. In terms of system composition, the system is no longer composed of a single node, but of multiple nodes. Different nodes belong to different service providers. Therefore, everything is no longer decided by a single service provider, but by multiple service providers. Any one service provider is no longer essential to the service, but optional. This chapter focuses on the problems encountered in decentralized computing, as well as strategies and methods for solving them. The involved knowledge includes consensus protocol, consensus algorithm, Paxos, and Practical Byzantine Fault Tolerance (PBFT). This chapter will uncover the mysterious veil of Paxos, presenting a concise implementation of Paxos and PBFT.