Efficient QoS-aware Service Composition
Efficient QoS-aware Service Composition
复制标题
DOI：
10.1007/978-3-0346-0104-7_5
复制
发表时间：
2008
期刊：
影响因子：
--
通讯作者：
Mohammad Alrifai;T. Risse
中科院分区：
文献类型：
--
作者：
Mohammad Alrifai;T. Risse
Web服务组合请求通常与端到端QoS需求结合在一起，这些需求是根据非功能属性（例如响应时间、吞吐量和价格）指定的。支持QoS的服务组合的目标是找到服务的最佳组合，使它们的聚合QoS值满足这些端到端需求。局部选择技术非常有效，但在处理全局QoS约束方面存在不足。另一方面，全局优化技术可以处理全局约束，但其较差的性能使其不适合具有动态和实时需求的应用程序。在本文中，我们解决了这个问题，并提出了一种将全局优化与局部选择技术相结合的解决方案，以获得更好的性能。该方法分为两步：首先利用混合整数线性规划（MILP）将全局QoS约束分解为局部约束；其次，我们使用本地搜索来查找满足这些本地约束的最佳web服务。与现有的基于MILP的全局规划解决方案不同，在我们的案例中，MILP模型的大小要小得多，并且与可用服务的数量无关，从而产生更快的计算速度和更高的可扩展性。已经进行了初步的实验来评估所提出的解决方案的性能。
Web service composition requests are usually combined with endto-end QoS requirements, which are specified in terms of non-functional properties (e.g. response time, throughput and price). The goal of QoS-aware service composition is to find the best combination of services such that their aggregated QoS values meet these end-to-end requirements. Local selection techniques are very efficient but fail short in handling global QoS constraints. Global optimization techniques, on the other hand, can handle global constraints, but their poor performance render them inappropriate for applications with dynamic and real-time requirements. In this paper we address this problem and propose a solution that combines global optimization with local selection techniques for achieving a better performance. The proposed solution consists of two steps: first we use mixed integer linear programming (MILP) to find the optimal decomposition of global QoS constraints into local constraints. Second, we use local search to find the best web services that satisfy these local constraints. Unlike existing MILP-based global planning solutions, the size of the MILP model in our case is much smaller and independent on the number of available services, yields faster computation and more scalability. Preliminary experiments have been conducted to evaluate the performance of the proposed solution.