How Hirschvogel built a highly available Proxmox VE and Ceph cluster across 3 buildings

Facing a costly hardware and licensing renewal for its existing infrastructure, HIRSCHVOGEL GmbH & Co. KG chose a different approach. By migrating to a hyper-converged environment based on Proxmox Virtual Environment and Ceph, the automotive group avoided rebuilding its previous architecture and significantly simplified an infrastructure distributed across 3 physical locations.

HIRSCHVOGEL GmbH & Co. KG is a family-run Mercedes-Benz dealership group with 7 certified sales and service locations. High availability has long been a key requirement for its virtualization platform. The previous environment reflected this approach, with 4 virtualization hosts and a mirrored external storage distributed redundantly across 2 fire compartments, while a third location provided quorum.

When that existing environment came up for renewal, the company decided not to rebuild the same way. Together with Proxmox Gold Partner Fsas Technologies, Hirschvogel designed a hyper-converged cluster based on Proxmox Virtual Environment and Ceph, added Proxmox Backup Server, migrated its production workloads while retaining the distributed architecture of the original setup.

Replacing the existing architecture no longer made sense

The virtualization hosts had reached end of life, the existing VMware licenses were up for renewal, and the associated costs had increased. Rebuilding the same architecture with separate virtualization hosts and mirrored external storage would have required a substantial investment. Hirschvogel therefore began looking for an alternative that could meet its availability requirements while simplifying the infrastructure.

The existing architecture also introduced considerable physical complexity. Running compute and storage as separate systems across the different fire compartments required extensive cabling and additional infrastructure. Reducing this complexity became another key objective of the modernization project.

For IT manager Christian Knoll, Proxmox VE had become a strong candidate that met Hirschvogel’s technical and operational requirements. He first learned more about the platform through an exchange with industry colleagues, which led to a discussion about deploying Proxmox VE together with Fsas Technologies. The companies established contact shortly afterwards.

A hyper-converged cluster across 3 buildings

Instead of rebuilding separate compute and storage systems, Hirschvogel chose a hyper-converged architecture that combined both on a single platform. The new environment uses a Proxmox VE cluster with Ceph, distributed across the existing buildings and connected through a direct-mesh cluster network. Hirschvogel developed the architecture together with Fsas Technologies, whose competence center reviewed and validated the design before the hardware was ordered.

The project team coordinated the design and technical review quickly, enabling Hirschvogel to place the hardware order before an announced storage price increase took effect.

High availability across 3 fire compartments

Hirschvogel placed 2 cluster nodes in each of the 2 primary fire compartments. A fifth system in the third compartment provides the additional quorum required to maintain cluster availability if one of the primary locations becomes unavailable.

Ceph provides the storage layer on full-NVMe drives with 4-fold data replication. By integrating compute and storage on the same platform, Hirschvogel no longer requires the mirrored external storage system used in the previous environment. Ceph traffic runs over a 100 GbE direct-mesh network, eliminating the need for dedicated storage switches. Hirschvogel also implemented Proxmox Backup Server as part of the project, which is now in production.

The final architecture includes:

• Virtualization & compute: 5 PRIMERGY RX1440 M2 systems running Proxmox Virtual Environment.
• Site distribution: 2 cluster nodes in each of the 2 primary fire compartments, with a fifth system providing quorum from a third fire compartment. The buildings are approximately 100 m apart and connected via fiber.
• Storage architecture: Full-NVMe Ceph storage with 4-fold data replication.
• Cluster network: 100 GbE direct-mesh network for Ceph traffic.
• Power protection: 2 UPS systems.
• Cluster resources: 136 physical CPU cores and 5.96 TiB of memory.

Stable production and improved performance

Hirschvogel completed the migration in March 2026 and shut down the previous environment. Since April, the company has been running its production workloads entirely on Proxmox VE without a single outage. The IT team has also seen a noticeable improvement in performance compared with the previous platform.

Throughout the project, Hirschvogel worked closely with Fsas Technologies, from architecture planning and implementation through to the final migration. This close coordination helped the IT team manage the platform transition in a structured and smooth manner.

Preparing for future growth

With the new platform in production, Hirschvogel can scale compute and storage together as requirements grow. The hyper-converged architecture reduces the need for separate external systems while preserving the distributed design across the 3 fire compartments. At the same time, the company benefits from a more predictable infrastructure model for future expansion.

Christian Knoll
Head of IT, Hirschvogel GmbH & Co. KG

About Hirschvogel
HIRSCHVOGEL GmbH & Co. KG is a family-run authorized Mercedes-Benz partner operating 7 certified sales and service locations across Lower Bavaria and the Upper Palatinate, Germany. The company provides vehicle sales and a broad range of automotive services for private and business customers. Learn more: https://www.hirschvogel.eu/

About Fsas Technologies
Fsas Technologies, a Fujitsu company, provides IT infrastructure and hybrid IT services, including consulting, implementation, professional services, and support for enterprise IT environments. Learn more: http://www.fsastech.com/en-eu