Читать книгу From Traditional Fault Tolerance to Blockchain - Wenbing Zhao - Страница 37

In such a distributed system, a failure could occur at a process. However, it is assumed that when a process fails, it simply stops execution and loses all its volatile state (i.e., the fail-stop model [18] is used). In addition, it is assumed that any two processes can establish a reliable connection (such as a TCP connection) for communication. Even though the network may lose messages, the reliable channel can effectively mask such losses. Naturally, the reliable connection ensures the first-in first-out (FIFO) property between the two endpoints of the reliable connection. This assumption also implies that the network does not partition, i.e., it does not prevent two or more processes in the system from interacting with each other for extended period of time.