摘要
为了在局域网络中实现连续、高可用的数据流处理,提出一种基于自适应修复和节点复制的方法:self-recovery andreplication-based backup(SRRB)。通过使用节点复制机制,利用对等节点并行向下游节点发送数据,使下游节点能够使用最先到达该节点的输入数据,提高数据流的传输速率。同时通过使用自适应修复机制,增强系统的容错处理。为了平衡网络开销和低延时保障,提出中心节点算法(CLEA)。通过在网络模拟器ns-3平台上实现,结果表明了SRRB方法具有更好的连续性、稳定性和高可用性。
A self-recovery and replication-based backup （SRRB） approach is proposed to realize both continuous and highly-available data streams processing over local area networks. This approach uses process-pairs mechanism to send output tuples to each downstream replica so that it can use whichever data arrives first for enhancing data stream transmission rate. To further improve the robustness of system, automatical recovery mechanism is devised, Next, for balancing the cost of data stream processing and latency guarantee, central leader election algorithm （CLEA） is developed. It is demonstrated by experiments on network simulator ns-3 that SRRB has more continuity, stability and high-availability.
出处
《计算机工程与设计》
CSCD
北大核心
2011年第7期2302-2305,共4页
Computer Engineering and Design
