The MATILDA architecture provides a contribution to bridge the gap between application-level orchestration and the Network Providers’ mechanisms that have to guarantee end-to-end connectivity, making applications network-aware – to the sole extent needed by their requirements – and, at the same time, making the network application-aware.
The architecture is divided into three distinct layers; namely, a) Development Environment and Marketplace, b) 5G-ready Application Orchestrator and b) Programmable 5G Infrastructure Slicing and Management. In a nutshell, the development environment is responsible to package a cloud-native component in a proper format, so as to be usable by the Control Plane architectural components. Beyond that, the combination of the components in the form of complex graphs is performed by editors that are provided in this layer. Cloud-native components and application graphs are persisted in a marketplace, so as to be searchable by developers. On the other hand, the logically centralized service mesh control plane is the layer that is in charge of the orchestration, monitoring and policy enforcement of a 5G-ready application. The MATILDA programmable 5G infrastructure slicing and management is the interface toward the Network Operators’ domain for the specification of configuration and management information of all underlying resources based on the requirements of the active policy.
Application developers will be requested to deal only with the application graph metamodel; notwithstanding this, they should be aware of capabilities offered by the 5G network layer in terms of connectivity and added-value services, obviously without being bothered by the knowledge of the architectural and implementation details of this offering. This is in line with the “as-a-Service” paradigm but, for the the 5G concept to really make a difference, it is required that this network awareness goes beyond the datacentre-confined vision intrinsic in the cloud environment to embrace the end-to-end 5G-enhanced capabilities, to permit the shift from cloud-native to truly 5G-ready applications. In this respect, the description of the networking requirements of the application needs to be specified only in terms of requested KPIs and constraints; likewise, the capabilities exposed by the network must match the same representational constructs. MATILDA has been devised with the ambition to provide the necessary tools to achieve these goals, which, on one hand, can contribute to ease the task of micro-service based cloud-native application developers and, on the other hand, can enhance the capabilities of Network Service Operators, particularly the small ones, in their market opportunities.