Inventory and dependencies
We first identify host roles, storage, networks and the workloads that actually require high availability.
- hosts and storage
- management / VM / backup networks
- workload criticality
- service dependencies
CASE / PROXMOX / HA
REAL ENVIRONMENT / ARCHITECTURE
Proxmox HA and backup — from standalone hosts to failure-aware architecture.
A practical example of a Proxmox environment where virtualization is only one layer; quorum, backup, recovery and operational clarity are equally important.
CONTEXT
A multi-node environment with growth and high-availability requirements.
The starting point was an environment with multiple Proxmox hosts, a separate Proxmox Backup Server and a plan to evolve toward HA.
The key design decision was to separate virtualization from backup and create an architecture that could grow without losing control of dependencies.
Quorum and high availability
HA only works as intended when quorum and node-loss behaviour are designed explicitly.
- 3+ nodes
- quorum
- VM priorities
- failure scenarios
Independent backup layer
PBS is a separate recovery layer rather than a replacement for HA. Backup and HA solve different failure classes.
- retention
- verify jobs
- restore testing
- offsite / secondary copy
Operations after deployment
The platform must remain maintainable: monitoring, change records, restore tests and failure procedures are part of the architecture.
- monitoring
- change log
- failover testing
- documentation
ENGINEERING TAKEAWAYS
Key engineering takeaways.
- HA does not replace backup, and backup does not replace HA.
- Untested recovery procedures can be a larger risk than the hardware failure itself.
- Quorum, storage and cluster networking should be designed together.
- Not every workload needs HA; prioritization reduces cost and complexity.
CASE / NEXT
SIMILAR ENVIRONMENT?
Discuss the environment