Automated Configuration of System Infrastructure for SOA-Based Enterprise Computing

In SOA-based enterprise computing environments, the deployment of high-levelservices onto the underlying servers is a first-class priority. Much work has focused on routing composite service workflow requests to deployed servers; however, the turn around time of the requests and the utility of the servers are dependent on how well the requested service types match the capacity and performance of the servers. In this paper we consider the scenario where service workload patterns are available to system administrators. The resulting problem is then how to automatically introduce new servers to scale up to meet the computing requirements of that workload. We state the problem as a combinatorial search through a variety of configuration assignments of services onto the servers. We show that the optimal or near-optimal combinatorial selection can be found with an AI-based search heuristic.