HIRSCHVOGEL GmbH & Co. KG, a German family-owned group of official Mercedes-Benz dealerships, has modernized its IT infrastructure with a hyperconverged Proxmox Virtual Environment (VE) and Ceph cluster spread across three buildings. Hardware renewal costs and the price of VMware licensing led the company to choose an architecture that combines compute and storage, removes dedicated external systems, and keeps the high availability its production services need.
Hirschvogel’s migration to Proxmox in 30 seconds
- The company operates seven Mercedes-Benz sales and service centers in Germany and needed to renew its virtualization platform.
- The new infrastructure uses five Fujitsu PRIMERGY RX1440 M2 systems with Proxmox VE and Ceph storage running entirely on NVMe drives.
- Two nodes sit in each of the two main fire compartments; a fifth system provides quorum from a third building.
- The storage network runs at 100 GbE in a direct mesh topology, with no dedicated switches for Ceph.
- The migration wrapped up in March 2026, and according to the company, production workloads have run since April without interruptions.
The rollout was carried out with Fsas Technologies, a Fujitsu group company and Proxmox Gold partner. The project included architecture design, technical validation, workload migration, and the addition of Proxmox Backup Server for backups.
The case illustrates one of the alternatives organizations with VMware environments are weighing when it’s time to renew: replacing a traditional server-plus-external-storage architecture with an integrated platform that manages resources from a single infrastructure.
Why Hirschvogel decided to change its VMware infrastructure
Hirschvogel’s previous infrastructure had reached end of life. Beyond renewing its servers, the company had to deal with VMware license renewal at a time when associated costs had gone up. Rebuilding the environment with the same architecture would have meant a significant investment.
The previous system used four virtualization hosts and mirrored external storage. Equipment and storage were distributed redundantly between two fire compartments, while a third location supplied the quorum the cluster needed to operate.
This design met the company’s availability requirements, but it also added physical complexity. Keeping compute and storage as separate systems required more cabling and additional components. The renewal offered a chance to preserve the geographic distribution while simplifying the architecture at the same time.
Christian Knoll, Hirschvogel’s IT lead, identified Proxmox VE as an alternative that fit the company’s technical and operational needs. His initial interest came from conversations with other industry professionals and led to contact with Fsas Technologies to explore a possible deployment.
The collaboration made it possible to design a hyperconverged platform with Proxmox VE and Ceph. Before ordering the hardware, Fsas Technologies’ competence center reviewed and validated the proposed design. According to the account published by Proxmox, the technical coordination also allowed the order to be placed ahead of an expected storage price increase.
Five nodes, NVMe storage and a 100 GbE network
The new infrastructure brings compute and storage resources together on a single platform. Instead of keeping virtualization servers separate from an external array, Hirschvogel uses five Fujitsu PRIMERGY RX1440 M2 systems running Proxmox VE that form a cluster with Ceph.
The layout preserves the three physical locations of the previous design. Two nodes sit in each of the two main fire compartments, while the fifth system, installed in a third compartment, provides quorum. The buildings are roughly 100 meters apart and connected by fiber optic.
Ceph handles the distributed storage and uses NVMe drives across every system, with four-way data replication. This setup makes it possible to do away with the mirrored external storage the previous architecture relied on.
The network has also been simplified. Ceph traffic runs over a direct 100 GbE mesh network, with no need for dedicated storage switches. The design cuts reliance on specific external equipment and concentrates the main functions on the cluster nodes.
The configuration published by Proxmox includes the following resources:
- Virtualization: five Fujitsu PRIMERGY RX1440 M2 servers running Proxmox VE.
- Physical layout: two nodes in each of the two main compartments, and a fifth system for quorum in the third.
- Storage: Ceph with NVMe drives and four-way data replication.
- Cluster network: direct 100 GbE mesh connection for Ceph traffic.
- Power protection: two uninterruptible power supply (UPS) systems.
- Compute capacity: 136 physical CPU cores.
- Memory: 5.96 TiB.
The company also added Proxmox Backup Server, Proxmox’s own backup management solution, which is already in production. So the project wasn’t limited to replacing the virtualization platform — it also covered protecting workloads within the renewed environment.
The migration finished in March 2026, and production keeps running
Hirschvogel completed the migration in March 2026 and powered down its previous infrastructure. Since April, the company has run its production workloads entirely on Proxmox VE. According to data reported by the company and included in the case study published by Proxmox, the environment hasn’t recorded any interruptions since then, and the IT team has seen a noticeable performance improvement over the previous platform.
The company doesn’t back those claims with comparative figures for latency, processing capacity, or response times. So the case documents an improvement the IT team perceived, but it doesn’t allow that improvement to be quantified or attributed to any specific component of the new architecture.
The spread across three fire compartments keeps the systems physically separated, while Ceph and the cluster configuration make it possible to bring compute and storage together in a shared infrastructure. The quorum at the third location is part of the availability design, though the published information doesn’t detail failure-test results or the exact recovery conditions for each scenario.
Fsas Technologies supported Hirschvogel through planning, technical review, deployment, and the final migration. That collaboration helped coordinate the platform change while preserving the distribution of systems between buildings.
The new architecture also offers a path to add resources as the company’s needs evolve. Hirschvogel expects to be able to increase compute and storage capacity within a hyperconverged model, without necessarily having to reproduce the separation between hosts and external storage from its previous environment.
Knoll sums up the goal of the change as greater control over infrastructure and budget, along with the digital sovereignty, availability, and flexibility the company needs. That’s an assessment from Hirschvogel’s IT lead, not an independent cost comparison between VMware and Proxmox.
The project offers a concrete example of migrating from a VMware architecture to Proxmox VE and Ceph at a company with several physical locations. Hirschvogel’s reported experience highlights storage simplification and production continuity, though the economic results and performance gains haven’t been published with detailed metrics.
Frequently asked questions
How many servers does Hirschvogel use in its Proxmox cluster?
The infrastructure uses five Fujitsu PRIMERGY RX1440 M2 systems running Proxmox VE. Combined, they provide 136 physical CPU cores and 5.96 TiB of memory.
How does Hirschvogel distribute its systems across the three buildings?
Two nodes are installed in each of the two main fire compartments. A fifth system, located in the third site, provides the cluster’s quorum.
What role does Ceph play in the new infrastructure?
Ceph provides distributed storage using NVMe drives with four-way data replication. The architecture eliminates the mirrored external storage and uses a direct 100 GbE mesh network with no dedicated switches for Ceph.
When did Hirschvogel complete its migration to Proxmox VE?
The migration finished in March 2026, and the company shut down its previous environment. According to the published information, production workloads have run on Proxmox VE since April with no recorded interruptions.
