Efficient atomic broadcast using deterministic merge

We present an approach for merging message streams from producers distributed over a network, using a deterministic algorithm that is independent of any nondeterminism of the system, such as the amount of time the messages are delayed by the network, or their arrival order. Thus, if this algorithm is replicated at multiple \mergers", then each merger will merge the message streams in exactly the same way. The technique is therefore a solution to atomic broadcast and global atomic multicast 12]. We assume that each producer has access to (approximately) synchronized clocks and can estimate the expected message rates of all producers. We propose an algorithm, called the Bias Algorithm. To measure the performance of the Bias Algorithm, we assume that messages are generated by memoryless processes operating at known message rates, and we measure the expected total merge delay at a given time L. For the case of two producer processes, we give optimal algorithms in this metric, and sho...

• Type: Conference
• Year: 2000
• Where: PODC
• Authors: Marcos Kawazoe Aguilera, Robert E. Strom
• Keywords: Algorithm, Bias Algorithm, Distributed And Parallel Computing, Message Streams, PODC 2000