Cloud technology plays an increasingly important role in our everyday digital lives. This increase has put a strain on developers and IT administrators to manually monitor, manage, scale, and maintain cloud software services and hardware infrastructure. Nikolaos Bezirgiannis, PhD student at CWI, explored an alternative solution of letting cloud software ‘manage itself’ and have automated control over its actual deployment in the cloud. He defends his PhD thesis ‘Abstract Behavioral Specification: unifying modeling and programming’ on 17 April 2018 at Leiden University. The results are interesting for, among others, software engineers.
Bezirgiannis says: “During the continuous race of software ‘catching-up’ with hardware, I believe that resource-aware as well as energy-aware software will be the new focus in the evolution of programming languages. In my research, I improved the resource-aware software.”
He adds: “In my thesis I unified application-centric code with the extra code of cloud configuration, provisioning and deployment. I used one combined high-level modeling language: an extension of ABS, the Abstract Behavioral Specification language. Software models written in this language can now be executed as multicore-enabled distributed applications. Because the models are ‘resource-aware’, they can monitor and control their own cloud deployment. The resource-aware language now comes packaged with a tool-suite for human-in-the-loop simulation of cloud services. Such a live simulation can be used for training DevOps engineers to the cloud environment of IT companies, targetting predefined Service Level Agreements (SLA).”
The research was carried out at Centrum Wiskunde & Informatica (CWI) in Amsterdam and Leiden Institute of Advanced Computer Science (LIACS). Bezirgiannis‘ promotor and co-promotor are Frank de Boer (CWI and UL) and Stijn de Gouw (OU). In 2017, Bezirgiannis, his promotor Frank de Boer and co-promotor Stijn de Gouw won the Best Paper Award in the 6th European Conference on Service-Oriented and Cloud Computing (ESSOC) for training DevOps engineers. The current work was partially funded by the European research project ENVISAGE – Engineering Virtualized Services. Necessary ICT infrastructure for this work was kindly provided by the Dutch national e-infrastructure foundation (SURF).