Cloud Computing enables scaling of web services. A typical customer would expect that the performance of web services on the cloud will be directly proportional to the availability of rented resources. However, we obtained that achieved performance is not always directly proportional to the scaling, by realising series of experiments varying the server load by changing the message size and the number of concurrent messages. The goal is to analyse the performance of web services utilising different hardware resources on the cloud for the same server load. We set a hypothesis about expected performance behaviour and then analyse and discuss the results about optimal resources when scaling the load. Interestingly, the results show different behaviour in determined regions, and also that there is a region where web services hosted on the cloud achieve superlinear speedup (speedup greater than the number of scaled hardware resources), meaning that the customers will get more performance than expected. Moreover, a region where input parameters are smaller without scaling the resources, provides an even better performance compared to scaled resources.
Cloud Computing Performance Speedup