Cloud-native applications: authoring and evaluation of two deployment patterns

Abstract

We live in the era of cloud computing today. Many companies are moving their legacy applications to the cloud or building cloud-native applications from scratch. What makes it more interesting is the fact that 80% of IT budget will be spent on cloud technology by the year 2025 [J. Soat, Mark Hurd Predicts the Future of IT, 2016]. However, there is no clear single approach on how to migrate or deploy these applications to the cloud, since various platforms provide different options. The diverse deployment approaches bring along different aspects in terms of both flexibility and responsibility of the deployment. However, one can ask: What are the main differences between these methodologies? What are the pros and cons? Is there a best or more robust methodology? When to use which technique and why? Are there any factors, which may influence your decision? Consequently, this thesis is introduced in order to answer such kind of questions and help developers choose the most suitable approach for their application. This work uses a systematic literature review in order to investigate the state of the art approaches and technologies used to enable platform-as-a-service deployments. Furthermore, considering the fact that there is still no scientific documentation for the varied approaches as patterns out there, new patterns for the various approaches are authored, described and documented. In addition, comparison matrices, which indicate the pros and cons of the multiple approaches using different application types are constructed, and a sample distributed application is deployed using the introduced patterns, in order to evaluate and validate them. Ultimately, a final discussion of the results is conducted and a decision tree that guides the choice of the appropriate pattern and the suitable platform is built.