Hybrid, but Confident – IT Architectures in Real-world Operation

Hybrid infrastructures are widely regarded as the contemporary answer to the requirements of modern IT: flexible, scalable, combinable. But how does this work in practice – without loss of control, without disruptions, without excessive complexity?

This article shows how hybrid architectures can be implemented technically, what requirements are crucial for this – and what is important for in-house operation, integration, and strategic controllability. With a real-world project example from the media industry.

Hybrid Reality – Strategically Obvious, Technically Challenging

Few organizations today rely solely on cloud infrastructures. The workloads are too diverse, the requirements too complex, and the demands for control, integration, and cost management too high.

Hybrid infrastructures – i.e., the interaction of on-premises systems with public or private cloud resources – are therefore the most realistic option in many cases. They enable:

• technical flexibility while continuing to operate critical components in-house,
• integration of existing systems into modern operating models,
• and graduated control of data sovereignty, performance, and operating costs.

However, setting up such architectures is technically challenging. Anyone who wants to operate cloud and on-premises not only in parallel but also in an integrated manner needs:

• a consistent virtualization or container platform,
• coordinated access concepts and identity management,
• a reliable network infrastructure with low latency,
• and monitoring that brings all components together.

Making your own infrastructure "cloudy"

Modern on-premise environments are no longer static, but highly automated and cloud-like in design. Many companies are building private clouds that are functionally indistinguishable from public cloud environments. The building blocks include:

Virtualization & containerization

• Classic hypervisor platforms such as VMware vSphere, Proxmox, or Hyper-V
• Container orchestration with Kubernetes, e.g., with Rancher, OpenShift, MicroK8s
• Use of infrastructure-as-code (e.g., with Terraform, Ansible) for provisioning resources
• Self-service portals through which specialist departments can independently request virtual machines, storage, or database instances

Identity management & security

• Integration of Azure AD Connect, SAML, OAuth2 to enable single sign-on across all systems
• Implementation of the zero trust principle: granular permissions, MFA, network segmentation
• Monitoring, logging, and auditing via platforms such as Prometheus, ELK Stack, Splunk

Storage & Performance

• Use of all-flash storage, NVMe-oF, hyperconverged infrastructures (HCI)
• Combination of primary storage (e.g., U.3-NVMe) with tiered storage for long-term data
• Use backup-to-cloud or disaster recovery-as-a-service for additional security

Case Study: Virtualization Cluster for a Media Company – Powerful, Scalable, Sovereign

A practical example of a modern, secure on-premises infrastructure with cloud-like functionality is the project of an international media company, which was implemented in 2024 in collaboration with Memorysolution and Mustang Systems.

The goal was to build a virtualized platform that would meet the high performance requirements of the editorial and production environment while remaining scalable, fail-safe, and integrable in the long term – without external dependencies.

Initial Situation & Requirements

The media company's IT department was looking for a platform to virtualize around 200 productive workloads, including editorial systems, databases, collaboration services, and internal tools for video and media data processing.

The requirements at a glance:

• High availability and redundancy at all levels
• Fast NVMe access for latency-sensitive applications
• Large-volume HDD storage for long-term archiving
• Scalability for growing production requirements
• Seamless integration into existing infrastructure (including Active Directory, monitoring, backup)

In addition, the platform should be operated entirely locally in order to maintain maximum control over data, configurations, and security mechanisms.

Technical Implementation

The solution is based on a three-node Proxmox cluster, supplemented by powerful storage infrastructure and a sophisticated network design. The configuration in detail:

Compute layer (virtualization cluster)

• 3× Supermicro A+ Server 2125HS-TNR
• Each equipped with
• 2× AMD EPYC™ 9554 (64 cores each)
• 1,536 GB DDR5 RAM per node
• 10× 7.68 TB NVMe SSDs (Samsung PM9A3) per node

Storage infrastructure

• Connection via Broadcom 9500-8e HBAs
• 3× Supermicro 847E1C-R1K23JBOD with 44 bays each
• Total storage capacity:
• 230 TB NVMe primary storage
• 1.5 PB HDD archive storage (tiered, energy-optimized)

Network

• Fully redundant 100 Gbit/s backbone with multiple uplinks
• Separate VLAN segments for management, storage, VM traffic, and backup
• Connection to central firewall cluster and monitoring system

Result: Cloud Convenience with On-premises Sovereignty

The solution was delivered as a turnkey complete system—including rack integration, storage tier configuration, network design, and documentation. The platform has been in productive operation ever since and meets the following objectives:

• Minimal latency for media-related applications, including video editing and cutting systems
• Maximum reliability thanks to dual redundancy at the compute and storage level
• Flexibility for future expansions – both for new nodes and container platforms
• Sovereignty through complete in-house operation without external cloud providers
• Compatibility with DevOps and automation approaches, including via Terraform modules and Ansible

Why this Project is Exemplary

This scenario illustrates how a modern IT platform can be designed to combine the strengths of the cloud (elasticity, self-service, scalability) with the advantages of on-premises solutions (data sovereignty, predictable costs, regulatory security). Instead of an "either/or" approach, a "both/and" approach was implemented here—with clear strategic control.

The decision to use Proxmox as a virtualization platform, combined with high-quality server and storage hardware, allows the company to flexibly integrate future developments such as containerization, edge workloads, or backup-to-cloud strategies.

for the complete case study

Conclusion: Hybrid is More than a Technical Connection—it is a Strategic Operating Model

The reality of many IT landscapes is not binary. There is no such thing as "the cloud" and "the data center," but rather a variety of requirements, risks, expectations, and technologies. Those who understand hybrid as a strategic operating model can master this complexity—and turn it into an advantage.

The following applies:

• On-premises is not a legacy system, but a strategic tool.
• The cloud is not a standard, but one building block among many.
• Hybrid is not a half-measure, but an intelligent design – if you plan it correctly.

Outlook for Part 5

In the final article in this series, we summarize the factors that determine sustainable infrastructure strategies today – with a practical checklist for technical decision-makers.