MYRTUS Reference Infrastructure
It is very well known that centralized solutions struggle to meet scalable and low latency scenarios, as well as privacy matters. Therefore, MYRTUS defines a reference infrastructure where a diversity of fog-level and edge-level devices converge with the cloud to form a computing continuum capable of addressing the needs of complex and dynamic systems.
Figure 1. MYRTUS Onion-Like Continuum
The MYRTUS reference infrastructure is a composable layered cloud-fog-edge continuum, integrating heterogeneous, federated, and collaborative computing components, whose generic architecture is drafted in Figure 2.
- The Edge Layer consists of commercial multicores, Heterogeneous Multi-Processor Systems-on-Chip (HMPSoCs) Field Programmable Gate Array (FPGA)-based accelerators, and adaptive RISC-V processors with custom computing units.
- The Fog Layer consists of Fog Micro Data Centers (FMDCs) and smart gateways to provide analytics services on medium to long-term data and to extend the capabilities of edge devices.
- The Cloud Layer provides intensive computing, long-term storage, subsystem monitoring, coordination, data mining, and historical analysis.
To establish a continuum of resources and to dynamically adjust the computation load over them: 1) all the components at each layer communicate with their layer-/component-specific MIRTO agent which, in turn, communicates with the other layer-/component-specific agents, and 2) all layers support Kubernetes as low-level orchestrator.
To create a continuum of information, all layers will share one ontological KB (logical view), which can be distributed in different layers (implementation view). This way, data remains close to their consumers but can be shared between layers for analysis and decision-making.
Figure 2. MYRTUS layered computing continuum infrastructure.
