Data Distribution Service Based on Symbol Bloom Filter for Large-Scale Distributed Computing

Data distribution service (DDS) has been widely used for the data communication of distributed computing system. But the existing DDS is based on the Simple Discovery Protocol (SDP). Its high network data transmission cannot meet the real-time requirement of the large-scale distributed computing. For this problem, this paper proposes a lightweight automatic discovery algorithm for DDS. In this algorithm, a new Bloom Filter with symbol is presented to compress the information transmitted between the distributed computing nodes, which is called Symbol Bloom Filter (SBF). Then it is combined with the SDP to reduce the consumption of the automatic discovery process in DDS. Experiments show that the proposed SDP_SBF can present a lower fault positive ratio while reducing data transmission, and therefore improves the real-time performance of DDS in the large-scale distributed computing.