Virtualization for ISPs

One physical server per service is idle hardware, failure risk, and unnecessary rack cost. Proper virtualization consolidates without creating a single point of failure — and VM backup is different from file backup.

What virtualization is in an ISP context

In the ISP, virtualization consolidates critical services — RADIUS, DNS, management system, monitoring, CGNAT — into hypervisors with consistent snapshots and backup. When a service breaks, it restores from a snapshot in minutes, not hours of bare-metal reconfiguration.

What RASYS does

  • XCP-ng cluster design and deployment — open source hypervisor with Xen Orchestra for visual management and API.
  • KVM on Linux for environments that prefer a fully open stack without licensing costs.
  • VMware ESXi when the customer already has a license and needs support for the existing infrastructure.
  • P2V migration (physical to virtual) with a controlled window and rollback plan.
  • Shared storage — iSCSI, NFS, or local depending on cluster size.
  • VM backup via custom in-house scripts + Xen Orchestra — consistent snapshot, policy-based retention, off-site copy in the cloud.
  • Documented DR plan — defined RTO and RPO, restoration runbook written and validated on first execution.

Platforms we work with

XCP-ng with Xen Orchestra, KVM with libvirt/Virt-Manager, VMware ESXi/vSphere. Storage: iSCSI (TrueNAS, StarWind), NFS, Ceph for larger clusters.

When it makes sense to talk to us

Server room with many physical machines that could be VMs; a critical service runs on bare metal without consistent backup; you inherited VMware without documentation and the license is expiring; you need a documented DR plan.

Talk to us — initial conversation, no commitment. See also: Configuration Backup.

FREQUENTLY ASKED QUESTIONS

XCP-ng, KVM, or VMware — which do you recommend?

For a new ISP without an existing license, XCP-ng is the starting point: open source, mature web interface (Xen Orchestra), backup via API. KVM is great for teams that want a 100% Linux stack and are comfortable with the CLI. For customers already on VMware: we evaluate migration to XCP-ng to integrate with our monitoring and management stack — the migration pays off when it eliminates an isolated tool and centralizes operations on the same standard as the rest of the network.

Does P2V migration take the service offline?

It depends on the service. For most: we clone the disk live, bring the VM up in parallel, validate, and cut over with a minimal window (usually less than 30 minutes). For services with highly dynamic state (active production database), we plan a maintenance window with incremental state synchronization before the cutover.

Is VM backup the same as backing up the service inside the VM?

No. VM backup (full disk snapshot) restores everything at once, but can be inconsistent if the service was not quiesced beforehand. For databases, we add a logical dump on top of the VM snapshot. Both layers have different RPOs and different use cases — we implement both.