| Summary: | Cloud computing came to make computing resources easier to access thus helping a faster deployment of applications/services benefiting from the scalability provided by the service providers. It has been registered an exponential growth of the data volume received by the cloud. This is due to the fact that almost every device used in everyday life are connected to the internet sharing information in a global scale (ex: smartwatches, clocks, cars, industrial equipment’s). Increasing the data volume results in an increased latency in client applications resulting in the degradation of the QoS (Quality of service). With these problems, hybrid systems were born by integrating the cloud resources with the various edge devices between the cloud and edge, Fog/Edge computation. These devices are very heterogeneous, with different resources capabilities (such as memory and computational power), and geographically distributed. Software architectures also evolved and microservice architecture emerged to make application development more flexible and increase their scalability. The Microservices architecture comprehends decomposing monolithic applications into small services each one with a specific functionality and that can be independently developed, deployed and scaled. Due to their small size, microservices are adquate for deployment on Hybrid Cloud/Edge infrastructures. However, the heterogeneity of those deployment locations makes microservices’ management and monitoring rather complex. Monitoring, in particular, is essential when considering that microservices may be replicated and migrated in the cloud/edge infrastructure. The main problem this dissertation aims to contribute is to build an automatic system of microservices management that can be deployed in hybrid infrastructures cloud/fog computing. Such automatic system will allow edge enabled applications to have an adaptive deployment at runtime in response to variations inworkloads and computational resources available. Towards this end, this work is a first step on integrating two existing projects that combined may support an automatic system. One project does the automatic management of microservices but uses only an heavy monitor, Prometheus, as a cloud monitor. The second project is a light adaptive monitor. This thesis integrates the light monitor into the automatic manager of microservices.
|
|---|