Lightweight Persistent Storage for Industrial Applications

Sammanfattning: Clouds are large computer centers that offer remote access to computing and storage resources, making them popular for business and web applications. They are now being considered for use in safety-critical applications such as factories, but lack sufficient time predictability, which makes it challenging to use them in these time-sensitive applications. To address this, an intermediate layer, called the Fog layer, is introduced to provide computational resources closer to the network edge. However, the fog computing paradigm faces new challenges in resource management, scalability, and reliability due to resource constrained nodes. Container-based virtualization is a technology that provides lightweight virtualization and fault tolerance mechanisms and can be applied to fog computing. By studying a robotic use-case, we realized the critical importance of persistent data storage for stateful applications. However, container-based solutions lack fault-tolerant persistent storage. In this thesis, we propose a persistent, fault-tolerant storage solution for container-based fog architectures that is designed for stateful applications and that provides scalability, auto recovery, and re-integration after failures at applications and nodes level. Key elements of our solution are a replicated data structure and a consensus protocol for distributed data consistency and fault tolerance in case of node failures. The fault tolerance and consistency of the solution are modeled and verified, and its timing requirements are evaluated. We use simulation to evaluate the timing performance of our solution in larger set-ups. The results of our study show that although adding a consistency protocol adds a timing overhead, the solution still meets timing requirements for the studied use-case even in presence of a set of considered faults. By leveraging a four-dimensional approach, we also conduct a comparative analysis of our solution with other approaches from various perspectives. Our observations indicate that our solution can be applied in a broader context than initially intended.