Web Services are considered a research and industry de-facto for providing functionality in a distributed manner that is usable by heterogeneous environments. Simply put, web services are packaged functionality that relies on a set of standards that facilitate the definition of the methods of the web services, its inputs numbers and formats, and its output numbers and formats. Combined with Replication, Web Services could provide performance optimization solutions to unlimited number of real-life business application. However, when discussing replication, one must ask, how to allocate (choose) replicas to provide the best performance achievable. This thesis is dedicated to answer this question. 
	The thesis is titled “A Broker-Based Web Service Allocation Algorithm” where an execution paradigm (Proteus) is presented and several allocation algorithms are presented, examined, and analyzed to show the optimum one. The thesis focuses on the Broker component of the execution paradigm. The broker has the allocation algorithm embedded in it. Furthermore, the thesis focuses on the Least Response Time (LRT) Algorithm, which is an algorithm that simply, yet not so simply, allocate web service replicas that provide faster response time. 
	The thesis provides all necessary background work and related research. It contains a simulated part, as well as a description of a real life system (Proteus). It also contains a section dedicated to analyze the results of execution logs and analyze different variations of environments (heterogeneous and homogeneous), number of replicas, and level of parallelism. Those logs are examined and conclusions were drawn.