The FLUIDOS Approach. The architecture of TerraviewOS is designed with a focus on efficient and secure operation across multiple customer on-premise sites leveraging the support services operated upon centralised cloud infrastructure. At the core of this architecture lies the concept of the FLUIDOS domain, which encompasses both customer sites and centralised services. In this use case, each customer site will host a FLUIDOS node, responsible for running the majority of TerraviewOS services along with the user interface. These nodes are designed to ensure security, with data encrypted both at rest and in transit, and will provide access to necessary centralised services. The backend services of TerraviewOS will run in the cloud, with a crucial design choice being the avoidance of one backend instance per FLUIDOS node. This approach aims to facilitate the secure sharing of workloads and data across FLUIDOS nodes that reside in the cloud. The FLUIDOS nodes will be designed to ensure that different nodes cannot interfere with each other, maintaining a high level of operational integrity and security. This is further enhanced by the requirement for workloads to run on Trusted Execution Environments (TEEs), which can be specified in the application descriptors through intents. Centralised cloud services of TerraviewOS, operating on a FLUIDOS node, will be multi-tenant, ensuring a clear separation between different customers. This separation is pivotal in maintaining data integrity and privacy for each customer. The key objective of this architecture is to minimise any duplication of service functionality, as such redundancy could lead to inefficiencies, technical debt, and unnecessary costs. All deployments for customer sites will be managed centrally, streamlining the operational process and ensuring consistency across the FLUIDOS domain. This centralised management approach allows for better control and oversight of the services provided, ensuring that each customer site operates effectively and securely within the overarching TerraviewOS ecosystem. Therefore, the expected advantages can be summarized in the following: • Business continuity: adapt to network changes, operate without uplink, and replicate for failover. This technical use case focuses on maintaining business continuity through network adaptability, offline operation, and failover replication. The system should adjust to network fluctuations, ensuring functionality without a primary network connection, and use altern...
The FLUIDOS Approach. Traditionally, PDCs were monolithic applications running on dedicated hardware; however, with the increasing computational power available at lower costs, this is changing in recent years. Experimental efforts are underway to virtualize applications and utilize Kubernetes for orchestrating the deployment of PDCs and real-time analysis applications at the edge. This is aimed at reducing latency issues and improving resiliency, avoiding the need of operator physical assistance in case of outages, and paves the way for their usage within a FLUIDOS-based environment. In fact, FLUIDOS creates a continuum of resources from the edge to the cloud and enables the displacement of workloads, such as data collection and analysis processes, based on specific scenarios (faults, reconfiguration, maintenance). The main features of the approach enabled by FLUIDOS are: • Computing Continuum FLUIDOS would enable PDCs and analysis applications to continue functioning even if communication with control centers is interrupted by migrating PDC services to an adjacent node in case of fault. • Intent-Based Orchestration FLUIDOS can automatically orchestrate PDCs based on the latency between the node and PMUs, thereby improving the power grid state estimate or responding to faults. • Cybersecurity FLUIDOS ensures service isolation from other applications on the hosting node with different usage permissions. It also leverages logging and anomaly detection capabilities and provides survival capabilities in case a portion of the grid is disconnected from the main network, hence preserving its operations in case of a cyber-attack.
The FLUIDOS Approach. The FLUIDOS continuum has the capability to transparently use computing resources nearby, increasing the overall system's productivity by intelligently and dynamically externalizing the robotics workload to other devices (e.g., a server at the factory premises) and/or using the robot's idle time (i.e., when the robot is docked in the battery charging station) to increase the entire system's computational capabilities. Each robot can leverage this approach when it is well- connected to the network, while it can rely on its sole onboard computing capabilities (which are turned on only upon necessity) when moving in a poorly connected area. This approach will lead to a significant decrease in battery usage and an increase in the robot's computational capabilities beyond its onboard limits, with the capability to dynamically adapt its computing behavior (i.e., onboard or offloaded) based on the actual operating conditions. With FLUIDOS, we can apply the cloud continuum computing approach by considering each robot as an edge device and intelligently and dynamically outsourcing robotics workloads to other robots or devices depending on the environment. This can be achieved without interfering with the robotics task, so robot developers will be able to run their applications without changing their way of working. Instead of using monolithic bare-metal workloads, the robots will use cloud-native technologies like containerization and Kubernetes to split and dynamically place the workloads in different devices. All robots will be treated as edge devices that can accept or externalize workloads, instead of being isolated devices. Given the potential capability of the FLUIDOS intent-based orchestrator to pursue different objectives, workload distribution among the different available systems can be optimized to achieve diverse goals such as: • Maximize the battery life of the robots. • Minimize the time it takes for the robots to complete their tasks. • Ensure that all of the robots are evenly utilized. • Avoid overloading any individual robot. Highly dynamic decisions can be envisioned as well. For example, if a robot is low on battery, FLUIDOS might move its workloads to other robots with more battery power. Or, if a robot is overloaded, FLUIDOS might move some of its workloads to other robots that are less busy. In a nutshell, the FLUIDOS approach to robot workload orchestration enables to improve the performance, efficiency, and reliability of your entire system.
