We have created different sets of test CESs and QoS configurations (with varying response times) using the Web service generation tool GENESIS[9]. A Gaussian distribution is used for specifying the response time of all CESs. The number of invocations is increased while decreasing the number of CESs to show the effect of scaling on the overall performance (i.e. an increasing number of replicate instances required for CESs with high invocation rate.).