QoS-Guaranteed Path Selection Algorithm for Service Composition

Manish Jain, Puneet Sharma, Sujata Banerjee
2008 Proceedings of the Fifth International ICST Conference on Heterogeneous Networking for Quality Reliability Security and Robustness  
Overlay networks have been employed as a popular solution to work around rigid and often suboptimal IP routing to provide better performance. Examples of services/applications that can use overlay networking include multicasting, end-toend QoS and secure overlay services [1], [2] . Service overlay networking (SON) is an emerging approach that builds on overlay networks to deliver advanced services to user. The main idea in the SON approach is that each overlay node can provide one or more basic
more » ... e one or more basic services, referred to as componentservices, in addition to application data forwarding service. These component-services act as building blocks for more advanced services, which can be created by combining these component-services in series or in parallel. Figure 1 illustrates the use of SON to provide media-content at user-specific rates by joining two component-services, content and transcoder, without requiring the content to be stored in multiple formats. One of the main challenges in SON is to, for a given user request, find a path between a source and a destination passing through a specific set of overlay nodes such that the path will satisfy one or more QoS requirement specified by user. This set of nodes depend on the user request which determines the sequence of services (and therefore nodes) that must be selected to satisfy user request. During the last few years, there have been several proposals for service path selection in SON [3], [4], [5] . The main limitation in these approaches is that they aggregate multiple constraints using a linear function. Jaffe [6] showed that the use of a linear function to aggregate multiple constraints can not satisfy all the QoS constraints. In this work, we propose a heuristic algorithm, K-Closest Pruning (KCP), to solve the problem of multi-constraint service path selection for SON in polynomial time. The main feature of this algorithm is that the path selected by this algorithm meets all the QoS requirements specified by the user/application. We model SON as a set of N overlay nodes O 1 , . . . , O N and N 2 overlay links connecting the N nodes. Each overlay node is characterized by the maximum resource capacity and average resource utilization. Similarly, the overlay link between any node O i and O j is characterized by three metrics: loss rate, delay and available bandwidth. The distinct type of component-services available from N overlay nodes is denoted by S 1 , . . . , S M , where M < N, such that each node supports at most one component-service. An end-user request has two components: advanced service A that can be mapped to a service template S 1 J , S 2 K , . . . , S n T , which specifies composition order of n services, and QoS constraints Q in terms of end-to-
doi:10.4108/icst.qshine2008.3959 dblp:conf/qshine/JainSB08 fatcat:ao7ah7ymcfc4hpf2usuwznuwge