When development cycles shrink and the number of projects grows exponentially, agility and speed become the primary competitive advantages. This case study illustrates how a rapidly growing software development company made a quantum leap by moving from legacy virtualization to a cloud-native ecosystem. By implementing a stack consisting of Harvester and Rancher, the company gained a powerful tool for the instantaneous deployment of Kubernetes clusters and traditional virtual machines within a single, resilient infrastructure.
Context
The client is a dynamic software development company based in Croatia, specializing in the design and support of products for European markets. Their portfolio includes several active projects ranging from enterprise web applications to high-load B2B platforms. For a long time, the team relied on Proxmox VE with standard KVM virtual machines.
However, as new projects emerged, the company hit architectural bottlenecks. New orders, especially during the development and testing phases, required the rapid provisioning of new Kubernetes clusters. Creating K8s environments on top of standard VMs was slow, required manual configuration, and complicated both automation and administration. The client needed a next-generation platform: a "single pane of glass" to manage both containerized and traditional KVM workloads, featuring deep API integration and Terraform support.
Goals and results
The Challenge
The client’s primary pain point was the "clunkiness" of their legacy infrastructure when handling cloud-native workloads. DevOps engineers were forced to spend hours on routine resource provisioning, network configuration, and K8s cluster deployment for new environments. Even with automation scripts and Terraform manifests, managing this fleet of instances presented significant hurdles, particularly regarding project isolation.
Furthermore, the company required robust high availability. Configuring clustered file systems on the old platform demanded niche expertise and significant time. The client needed a system with a low barrier to entry that provided excellent cluster management tools, simplified data replication, supported legacy VM-based projects, and was fully IaC-ready.
Infrastructure configuration
To ensure maximum performance and eliminate bottlenecks, a hyper-converged infrastructure architecture based on three dedicated servers was selected.
.jpg)
The use of a high-speed local network completely eliminated bottlenecks during Longhorn data replication and ensured near-instantaneous VM migration between nodes.
The Solution
The INTROSERV team proposed implementing a hyper-converged infrastructure based on Harvester HCI, integrated with the Rancher management platform.
- Tiered storage allocation
Installing Harvester on a dedicated SSD RAID ensured OS stability, while the NVMe pool for Longhorn guaranteed high IOPS for VM disks and Kubernetes Persistent Volumes. - Lag-free synchronous replication
The 25 Gbps internal network allowed Longhorn to mirror data across three nodes with minimal latency. In the event of a server failure, VMs automatically restart on adjacent nodes without data loss. - Rancher unified control plane
The client received a single graphical interface uniting the management of VMs and Kubernetes clusters – covering the full lifecycle: provisioning, monitoring, and updates. - Full IaC automation
The native Rancher Terraform provider and Harvester API enabled a true Infrastructure-as-Code approach.
Private infrastructure as a strategic advantage
The solution implemented by INTROSERV was not just a hypervisor swap, it was a transition to a fundamentally different way of managing infrastructure – convenient, automated, and scalable.
Notably, the client considered an alternative path: moving workloads to hyperscalers. However, preliminary calculations showed that equivalent compute power would cost 3 to 4 times more annually. As the project portfolio grew, this gap would only widen – even before accounting for egress traffic fees and unpredictable instance price hikes.
Dedicated hosted infrastructure also solves the issue of data sovereignty: all client data resides exclusively on rented servers within a secure perimeter controlled by the client. For a company serving European clients under strict GDPR requirements, this is a necessity, not just a convenience. An additional argument for the chosen stack was RKE2 – SUSE’s Kubernetes distribution, which Rancher deploys by default. RKE2 is secure by default, complying with CIS Benchmark and FIPS 140-2 standards, making it the preferred choice for government agencies and regulated industries.
Is your current virtualization environment failing to meet your business needs? Are you looking for a cloud-native approach to infrastructure management?
Let the INTROSERV team handle the transition. We will conduct an audit, select the optimal hardware, and develop a migration plan to the Harvester + Rancher ecosystem.
