基于代理机制的自适应服务组合演化

A Daptive Service Composition Evolution Based on Agent Mechanism

在很多大型系统的,面临的一个关键就是异常的发生。在任何数据中心,异常都在不断的发生,或由磁盘故障、内存、网络连接,或软件错误引起的。较小的异常连锁反应可能造成大规模的异常。而异常的不可预测性使他们变得具有挑战而且难以应付。预测异常更好的选择是创建一个可以自主适应异常的系统。本文为自适应系统的服务组合探讨了一个分散的自适应方法。本文的做法是基于代理协调机制,而这适合于大规模的协调系统。我们通过详细的实验的复合服务评估这种方法。实验的结果表明,当面临大规模的异常时,它可以创建服务组合以应对这种异常。

In many large-scale systems, one of the key is the occurrence of anomalies. In any data center, anomalies are constantly occurring, or caused by disk failures, memory, network connections, or software errors. Smaller abnormal chain reactions may cause large scale anomalies. And the unpredictable nature of the anomalies makes them challenging and difficult to cope with. The better option to predict exceptions is to create a system that can adapt itself to the exception. In this paper, a decentralized adaptive method is discussed for the service composition of the adaptive system. The approach is based on the agent coordination mechanism, which is suitable for large-scale coordination systems. We evaluate this method through detailed experiments on a composite service. The results of the experiments show that when confronted with large scale anomalies, it can create a service composition in order to cope with this anomaly.