Virtualisation and backup environment modernisation for an IT services provider
A growing infrastructure environment required a more structured approach to virtualisation, storage and backup, as well as preparation for further expansion. The project included an assessment of the existing environment, target architecture design, service migration, separation of the backup layer from production infrastructure and implementation of centralised monitoring.
- IT and hosting services provider based in Warsaw
- Hosting and Data Centers
Challenge
The environment had been developed gradually as the number of services and customers increased. Individual infrastructure components continued to fulfil their roles, but over time the number of dependencies between virtualisation, storage, networking and backup also grew.
Before further expansion, the architecture needed to be reorganised and the risks associated with single points of failure reduced.
Growing environment
The infrastructure needed to support an increasing number of services without introducing additional accidental dependencies.
Backup
The data protection layer needed to become more independent from the primary production environment.
Fault tolerance
The impact of a failure affecting a single host, storage component or other part of the infrastructure needed to be reduced.
Solution
We started the project with an assessment of the existing infrastructure, dependencies between services and the way compute, storage, networking and backup resources were being used.
Based on this analysis, we designed a target architecture with clearly separated roles for the individual infrastructure layers.
Virtualisation
We reorganised the compute layer and prepared a clustered environment designed to support further resource expansion.
Storage
Storage was aligned with workload characteristics, performance requirements and the required level of fault tolerance.
Backup
We separated the backup layer from production infrastructure and prepared recovery scenarios for systems affected by infrastructure failures.
The migration was divided into stages to reduce risk and allow the environment to be verified after successive services were moved.
Result
Following the modernisation, the infrastructure gained a more predictable structure, with clearly defined responsibilities for its individual layers.
Fewer dependencies
A structured architecture makes it easier to understand the impact of failures and reduces unnecessary dependencies between services.
Improved recovery
An independent backup layer makes it possible to plan recovery even when the production infrastructure itself is affected.
Ready for growth
The environment can be expanded by adding resources without redesigning the entire architecture at every stage of growth.
Technologies and services used
Proxmox VE Proxmox Backup Server ZFS Zabbix
Modernisation starts with architecture
In a larger virtualisation environment, replacing servers alone does not solve architectural problems. The key is understanding the dependencies between compute, storage, networking, backup and the services running on the platform.
Before migration, we therefore analysed workload placement, resource utilisation, storage requirements, network communication and the way backups were created and stored.
The objective was not to replace infrastructure simply for the sake of replacement. The goal was to build an environment that could be developed, monitored and recovered in a predictable way.
Compute, storage and backup as separate layers
One of the main assumptions of the project was to clearly define the responsibilities of individual infrastructure components.
Compute
Virtualisation hosts provide computing resources for workloads and can be expanded independently from the other infrastructure layers.
Storage
The data layer was selected according to actual performance, capacity and redundancy requirements.
Backup
Backup copies remain outside the primary production path and provide an independent data protection mechanism.
Backup is not High Availability
Clustering, RAID, ZFS or replication can improve infrastructure resilience against specific types of failure, but they do not replace an independent backup.
For this reason, backup was designed as a separate part of the architecture, with its own resources, retention policy and recovery scenarios.
When designing backup, we focus primarily on the ability to restore the service — not simply on whether a backup job finishes with a successful status.
Migration in stages
Instead of moving the entire environment at once, we divided the migration into stages. This allowed us to verify successive parts of the infrastructure and reduce the scope of any potential issue.
For critical changes, rollback options were included, while the migration order reflected dependencies between individual services.
Monitoring as part of the new architecture
After migration, the environment is monitored centrally. Monitoring covers not only CPU and memory utilisation, but also the condition of hosts, storage, networking, backup systems and key infrastructure services.
This makes it possible to identify problems at the infrastructure level instead of discovering them only after users report that a service has become unavailable.
Good infrastructure should not only tolerate failures. It should also make it possible to quickly identify when one of its components begins to behave incorrectly.
Architecture prepared for further growth
The result of the project is not a closed environment that will need to be redesigned from the beginning in a few years. Individual layers can evolve as the number of services, storage utilisation and demand for compute resources increase.
This approach reduces the need for ad hoc infrastructure additions and allows the environment to grow according to a previously defined architectural model.