Constrained Component Deployment in Wide-Area Networks Using AI Planning Techniques

Component-based models represent a dominant trend in the construction of wide-area network applications, making possible the integration of diverse functionality contained in modules distributed across the network. Although linkages between modules have traditionally been specified statically, a growing number of frameworks are investigating approaches where appropriate components are dynamically selected and deployed in the network. Dynamic approaches enable flexible adaptation of the application to achieve load-balancing, satisfy QoS requirements, and in general, enable customization to changing client and network characteristics. A key element of dynamic component-based frameworks is the deployment plan, which describes the choice of components, their locations, and linkages. The problem of finding a valid component deployment is harder than traditional optimization problems because one needs to decide on the set of components while satisfying various constraints resulting from application semantic requirements, network resource limitations, and interactions between the two. Partly because of this complexity, existing component frameworks typically address only a limited case of the general deployment problem. In this paper, we propose a general model for the component placement problem (CPP) and present an algorithm for it, which is based on planning algorithms developed by the Artificial Intelligence community. These algorithms have benefited from several decades of research and offer substantial expressibility with good efficiency. We validate the effectiveness of our algorithm by demonstrating its scalability with respect to network size and number of components in the context of deployments generated for two example applications – a security-sensitive mail service, and a webcast service – in a variety of network environments.