In the ever-evolving networking landscape, the demand for efficient and adaptable Virtual Private Network (VPN) solutions is growing. Software-Defined Networks (SDNs), particularly Peer-to-Peer (P2P) overlay VPNs, offer a practical approach for networks spanning various edge and cloud providers. However, existing decentralized VPNs, while resilient and scalable, typically utilize a single tunnel type and overlook data plan costs and link performance in their selection processes. This oversight can lead to cost and performance inefficiencies, especially in edge-to-cloud networks where diverse nodes have unique needs that generic solutions fail to meet effectively. Although SDN facilitates the integration of multiple link types in overlay VPNs, existing systems lack efficient policies for selecting favorable tunnels. To bridge this gap, we introduce PolyNet, a Multi-Criteria approach designed to make cost- and performance-aware policy decisions in hybrid-link overlay networks. PolyNet employs a dynamic link selection policy during runtime that evaluates latency using Vivaldi network coordinates and considers cost, and integrates with SDN-based P2P overlays to enhance link management capabilities and support multiple link types. This paper presents the design of PolyNet and evaluates its performance through simulations and prototype testing. Results demonstrate that PolyNet achieves up to a 19.1% cost reduction and a 14.1% latency improvement over traditional methods in Symphony P2P topologies. Additionally, tests with a software prototype confirm the advantages of hybrid links, showing that kernel-layer GENEVE tunnels can increase throughput by up to 8.9 times compared to user-layer Nebula and WebRTC tunnels in edge clusters.

The management of drinking water quality is critical to public health and can benefit from techniques and technologies that support near real-time forecasting of lake and reservoir conditions. The cyberinfrastructure (CI) needed to support forecasting has to overcome multiple challenges, which include: 1) deploying sensors at the reservoir requires the CI to extend to the network’s edge and accommodate devices with constrained network and power; 2) different lakes need different sensor modalities, deployments, and calibrations; hence, the CI needs to be flexible and customizable to accommodate various deployments; and 3) the CI requires to be accessible and usable to various stakeholders (water managers, reservoir operators, and researchers) without barriers to entry. This paper describes the CI underlying FLARE (Forecasting Lake And Reservoir Ecosystems), a novel system co-designed in an interdisciplinary manner between CI and domain scientists to address the above challenges. FLARE integrates R packages that implement the core numerical forecasting (including lake process modeling and data assimilation) with containers, overlay virtual networks, object storage, versioned storage, and event-driven Function-as-a-Service (FaaS) serverless execution. It is a flexible forecasting system that can be deployed in different modalities, including the Manual Mode suitable for end-users’ personal computers and the Workflow Mode ideal for cloud deployment. The paper reports on experimental data and lessons learned from the operational deployment of FLARE in a drinking water supply (Falling Creek Reservoir in Vinton, Virginia, USA). Experiments with a FLARE deployment quantify its edge-to-cloud virtual network performance and serverless execution in OpenWhisk deployments on both XSEDE-Jetstream and the IBM Cloud Functions FaaS system.