Structured derivation of semi-synchronous algorithms
The semi-synchronous model is an important middle ground between the synchronous and the asynchronous models of distributed computing. In this model, processes can detect (timeout) when other processes fail. However, since detection is done by timing out, it incurs a cost much higher than the typical delay of messages. The paper presents a new communication primitive, Timely Announced Broadcast (TAB), and uses it in algorithms for consensus and set consensus in the semi-synchronous model. Separate implementations of TAB, withstanding different types of failures, allow to derive algorithms for consensus and set consensus under crash and omission failures. The time bounds obtained by our algorithms asymptotically match, or improve, the previously known bounds.