XCP-ng Server Setup Best Practices
1. Hardware Planning
General Hardware Both older and newer hardware are fine, as long as it is x86 and supports virtualization.- Software RAID Boot XCP-ng supports mdadm mirror setup on install
- Hardware RAID Controller Good for local storage management
- HBA or passthrough — Allows software (like ZFS or external storage) to manage redundancy. RAID controllers are not recommended when using passthrough to ZFS
- Dual or more NICs — Separate management, VM, and storage traffic.
2. Install & Initial Config
- Use the latest stable XCP-ng ISO — Stable builds ensure compatibility and security.
- Set a static IP during install — Prevents DHCP surprises, especially for remote management and storage access.
- Have a Host Naming Scheme — Simplifies management and XOA integration.
3. Security
- The firewall is enabled by default — review and adjust if needed
- Disable SSH password authentication — Use SSH keys for access.
- Use VLANs or Separate Network — Limit exposure of management interfaces.
- Apply updates to XCP-ng and XOA — Keep the host secure, stable, and get new features
4. Storage Setup
Local Storage
- Use SSDs or NVMe for local SRs — Essential for high-performance workloads.
- Local ZFS this is supported but currently not managed via the XO interface
- Prefer EXT over LVM Thin provisioned and not block based so it’s easier to manage and recover
External Storage
- NFS is preferred over iSCSI
Simpler integration with XCP-ng- Easier to recover from because it is file based
- Use MC-LAG switch setup for redundancy
May be slightly lower performance than iSCSI in some edge cases
- iSCSI
- Multipath is supported
- Not thin provisioned
- Block not file based
- Hyper Converged XOSTOR which is based on DRBD
- Storage is replicated across nodes, allowing VMs to survive host failures.
- No need for a dedicated external SAN/NAS—storage
- Heavy network dependency to sync hosts
- Can have much higher latency on writes without performant hardware
- Only suitable when consistent, low-latency networking is available between nodes.
5. VM Management Best Practices
- Use rational naming conventions — e.g.,
prod-app-01,lab-win10-02 - Tag by role, department, and or environment — Useful for automation, filtering, and backups
- Storage Design — VMs are ideal for running compute workloads, but storage should be handled externally when practical. https://lawrence.video/storagedesign
Don’t Over-provision — Don’t overprovsion CPU or Memory
6. Backup Strategy
- Use XO backups — They are integrated and easily automated
- Keep backups on separate storage — Ideally not on the same NAS as the live NFS share
- Test your restores regularly — Because “successful backup” does not mean “tested restore”
7. Networking
- Have a clear Naming Scheme — Take advantage of the descriptions field
- Name The Unused Interfaces — I name them “Not In Use”
- Segment management, VM, and storage traffic — Use VLANs or separate NICs
- Management Interface use a dedicate NIC if possible, not a bonded interface
- Bond NICs — Fully supported make sure your network switch does as well
- Dedicated Backup, Storage, Migration Network — In XCP-ng, you can assign a default network for backup, storage, and migration traffic at the pool level.
8. Monitoring & Maintenance
- Setup XO Email — Set up email notifications in XO
- Use Syslog — Send data to a syslog server, I personally like Graylog
9. Documentation & Scaling
- Keep a config doc or runbook — Include IPs, hostnames, pool members, etc..
- Use tags — For groupings like environment, function, or backup policies
- Netbox Support — The Netbox plugin allows better IP and asset tracking for XCP-ng environments.
