Running highly available virtualization at the Edge with just two data nodes

Running highly available virtualization at the Edge with just two data nodes Edge computing always presents a familiar puzzle. You need your applications to be highly available because out there in the real world, things break. But you also have to fit your hardware into cramped, power-constrained cabinets. And, frankly, you need to stay within a reasonable budget. Traditionally, if you wanted production-grade, distributed storage that could survive a node failure without losing its mind or your data, you had to deploy at least three full-sized storage servers. Multiply that by hundreds or thousands of retail stores or telecom towers, and your infrastructure costs skyrocket. Our latest Technical Reference tackles this exact headache. It proves you can get enterprise-grade resilience at the Edge. Better yet, you do not need that expensive third heavy server. The Power of two nodes and a lightweight witness The solution lies in a validated reference architecture. This combines SUSE Virtualization, Portworx Enterprise by Everpure, and Supermicro Compact Edge Servers using a Two-Node with Arbiter (TNA) topology. Instead of three massive data-storing servers, this architecture uses only two worker nodes to run your virtual machines and containers, backed by synchronous data replication. The magic third component is a lightweight arbiter node. The arbiter doesn’t store any workload data. It only holds cluster metadata so it can act as a definitive tie-breaker if a network partition or a node failure occurs. Because the arbiter is computationally lightweight, it can live on a much smaller, less expensive server or even be virtualized. For distributed enterprises, swapping a heavy NVMe storage server for a lightweight witness can slash per-site storage infrastructure costs by 25 to 35 percent. Real-world resilience under fire We don’t just write architectures on paper; we also test them. The engineering teams at SUSE and Everpure put this TNA configuration through real failure scenarios to see how it handles disaster. - Sudden Storage Node Outage: With active virtual machines actively logging data every second, one of the primary storage nodes was abruptly shut down. The surviving node and the arbiter maintained a majority quorum. Kubernetes automatically triggered live migrations, and all virtual machines were back up and running on the surviving node within about two minutes. - Cascading Double Failure: Next, they pulled the plug on the arbiter first, leaving the two storage nodes running perfectly fine on their own. Then, they abruptly shut down one of those remaining storage nodes. Even during this cascading crisis, the sole surviving node established leadership. Total observed I/O interruption for the workloads throughout this extreme double-outage scenario was a mere eight seconds. Part of an expanding Enterprise Storage ecosystem This new guide expands our growing library of validated solutions with Everpure. Depending on your specific edge or data center design, you might also want to explore our previous documents: - The deployment path for combining SUSE Virtualization with an external, high-performance back-end via the Everpure FlashArray Integration Guide. - The foundational blueprints for running container-native storage directly on physical infrastructure with the SUSE Virtualization and Portworx Bare-Metal Reference Configuration. An architecture like this is the result of focused technical collaboration and thorough validation. We want to extend our gratitude to the authors and experts who designed, built, and tested this solution: - Suresh S, Partner Solution Architect at SUSE - Gopala Krishnan, Partner Solution Architect at SUSE - Terry Smith, Partner Solution Innovation Director at SUSE (who also helped me write this blog article) - Bhumitra Nagar, Member of Technical Staff at Everpure Thank you for your contribution, your technical expertise, and your work in making two-node edge virtualization a practical reality for organizations. Get the technical details If you are ready to stop choosing between high infrastructure costs and high availability at the Edge, this architecture is worth a close look. You can find step-by-step implementation procedures—including disk partitioning, OS immutability workarounds, and sample Kubernetes manifests—by browsing the SUSE Technical Reference Documentation. Efficiency at the Edge isn’t about cutting corners; it’s about engineering smarter layouts. By eliminating redundant hardware stacks and shrinking the quorum footprint, you can reliably run legacy VMs alongside modern microservices right where your customers need them.