6GSMART-ICC: A Cloud-Continuum ORChestrator (CORC)

Nowadays, to meet the demands of Industry 4.0, a seamless convergence between the different cloud domains (edge, edge and cloud) is key for industries to have a cloud-native intelligent orchestration layer capable of establishing the necessary mechanisms to ensure continuity of services as data moves between the different cloud domains. To this end, a meta-orchestrator on top of the orchestration layer of the different domains must enable the creation of a real computing continuum and granular control of the resources available in each network segment.

To this end, the Cloud-Continuum ORChestrator (CORC) implements capabilities for managing and processing data from manufacturing environments oriented to the industrial cloud continuum, which spans from intelligent devices on the shop floor through the public and private edge to the cloud.

The implemented Cloud-Continuum ORChestrator solution is responsible for providing the mechanisms for service continuity between domains and precise control of the resources available in the different segments of the network.

The following diagram reflects the infrastructure with the different computing environments or domains in the cloud, covering from the extreme edge to the public cloud. It represents the diversity of cloud computing and is capable of enabling the convergence of devices and environments across the cloud continuum.

The domains contemplated are:

• Extreme edge: It consists of low-power consumption and high-portability devices such as Raspberry Pi orchestrated with K3s.
• Edge: This consists of devices with higher capacity than the extreme edge but with some limitations.
• Public Cloud: This is a public cloud offered by large cloud providers such as AWS, Azure, or GCP.

Use case validation

To evaluate and demonstrate the capabilities of the Cloud-Continuum ORChestrator (CORC) in the creation, deployment and management of applications and services in the cloud continuum in an efficient manner, providing the required connectivity, performance, scalability, availability and security aspects, a use case based on a typical scenario of a multi-cluster and multi-domain infrastructure in an industrial environment has been created. The objective has been to enable the convergence of devices and environments across the different domains of the cloud continuum, providing mechanisms for cross-domain service continuity and precise control of the resources available in the different network segments.

To this end, a microservices-based distributed application has been developed that follows the cloud-native software architecture pattern, which will allow the remote visualisation of an ‘industrial process’ composed of four microservices/components with specific requirements for their placement in different domains or network segments (such as the extreme edge, the edge and the cloud):

• A micro-service called ‘producer’, that captures video via a camera and transmits it via streaming and that will be deployed at the extreme edge.
• A micro-service that consumes the streamed video and supervises the pre-processing and storage in near real-time, which we call the ‘consumer’, and which will be deployed at the edge.
• A player in charge of playing the streamed video, called ‘player’, which is also hosted at the edge.

Finally, we will have the observability stack deployed in the infrastructure, overseeing monitoring resources (energy, CPU, memory consumption), covering clusters, nodes, workloads, etcetera.

Monitoring is supported by a dashboard deployed in the public cloud, through which we will have a centralised view of the state of the infrastructure concerning resource use.

Author: Jesús Benedicto – IT Project Manager & SW Architect at Eviden