Efficient distributed recovery algorithm using non-FIFO channel in distributed network

This paper presents a recovery algorithm using a vector clock for distributed networks. The main idea of our recovery algorithm started from two concepts. The first concept is the use of global local time for dependency tracking. We can get the maximal state with the global local time. The second concept is the snapshot algorithm. We can have a complete view of global state with the snapshot algorithm. In this algorithm, each process maintains a global local time for n processes using a vector clock. Since global local time is obtained by the snapshot algorithm, each process keeps its own maximal state. The maximum consistent state can be tracked using only global local time and it requires neither a unique timestamp nor any index tables. Also, it does not need O(n) additional information for dependency tracking in normal operation.<
>