Internet-of-things (IoT) generally refers to the interconnection of things (e.g. sensors, smart phones, home appliances, etc.). Several IoT platforms like IoT-A [1], BUTLER [2], iCORE [3], etc., have been proposed addressing issues like security, performance, energy-efficiency, etc. The evolution of these platforms, however, created isolated systems with limited inter-operability. A new generation of IoT projects (i.e RERUM [4], [5], FIESTA [6], INTER-IoT [7]) provided solutions to security-by-design, inter-operability, and several as-a-service capabilities like platform-as-a-service, testbed-asa-service, etc. Despite these advances, and by considering the rapid proliferation of the IoT platforms and the smart devices, there is the need for proper storage and processing of vast amount of data generated and collected by distributed IoT networks. Fog Computing (FC) is an environment where data are stored and pre-processed before transmitting them to the cloud, so that computation, storage and networking services can be performed locally. FC has a number of advantages, within the IoT concept, like scalable real-time services, fault detection and isolation, enhanced security and privacy, mobility support, geo-distribution, location awareness, low latency, etc. In this paper, we present a three-layer IoT architecture that employs fog-enabled gateways (GWs). More specifically, the presented architecture consists of: (i) sensor nodes (SNs) that lie on the network edge, (ii) fog-enabled GWs that aggregate sensory data and perform several other operations like SN registration, etc., and (iii) an IoT middleware used as backend cloud. The proposed architecture is capable of collecting not only sensory data but management data as well, like the device uptime, the energy consumption of the SNs, etc. This can be used as the first step towards the implementation of a Management-Platform-as-a-Service with inter-operability capabilities. This service will support the collection of various management-related data like network statistics, energy consumption, health status of the SNs, aiming to detect and predict potential failures at all layers of the infrastructure, thus making feasible a self-healing IoT platform.