| Service Oriented Architectures (SOA) are a mainstay of enterprise computing and there is now a growing interest in services for systems of connected devices in a variety of environments, ranging from industrial manufacturing equipment to home automation, and other highly heterogeneous environments. In fact, the current trend in connected devices is expected to accelerate as the vision for the Internet-of-Things (IoT) becomes a reality. The IoT embodies the seamless discovery, configuration, and interoperability of networked devices in various settings, and in a sense, it has extended the application range of Enterprise Application Integration (EAI) to non enterprise environments. For instance, EAI in manufacturing environments with highly demanding dependability and timeliness requirements, must leverage closed proprietary middleware solutions that incorporate some fault tolerance techniques to fulfill such requirements, since transactional processing does not satisfy those requirements completely. But as non enterprise applications become increasingly critical, the middleware coping with Machine-to-Machine (M2M) communication and coordination, such as the Devices Profile for Web Services (DPWS), has to deal with fault tolerance and increasing complexity, while still abiding to resource constraints of target devices. Fault tolerant service integration in such scenarios can then be broken down into reliable communication and service correctness. These two features can be provided by gossip protocols, to ensure reliable message exchanges in different communication patterns, and consensus protocols, to ensure the normal behavior of intervening services. In this dissertation, we address these challenges by proposing a DPWSbased framework containing a gossip service and a consensus service, and evaluate its effectiveness on providing fault tolerance capabilities to existing services. |
