The increasingly popular cloud services provide on-demand resource usage and high scalability. In the case of the Platform as a Service (PaaS) cloud model, the users (who run their applications on the cloud) upload their applications to the system, set the parameters of the runtime environment via the configuration interface provided by the platform, and the system runs the application using the selected amount of resources. The parametrization options vary from provider to provider, entail different costs, thus leading users to complex questions.
The most striking advantage of the cloud based service model is the efficient and simple scalability. It lets the user successfully prevent usage overloads, while helping them to achieve significantly reduced costs under moderate loads. This thesis aims to provide examples of the problematic scenarios that may arise while operating PaaS based applications, taking the aforementioned premises into account.
This thesis encompasses the documentation of the measurements I implemented. It contains the description and the required tools of those measurements, including the recorded results. I present several PaaS services alongside the frameworks and extension modules required to implement applications for those services. The thesis focuses on the analysis of performance, scalability and reliability parameters of the various cloud services, with emphasis on the PaaS services of Heroku.