Infrastructure
Provisioning and IP allocation, distributed storage, networking between sites, snapshots, rescue and the recovery path after a host fails — in one panel, with one set of permissions and one operation log. Nodekeeper does not replace your hypervisor or reinvent distributed storage; it connects the work around them.
On this page: Virtualization · Storage · Networking · Recovery · Hosts & databases · Plans
KVM / ESXi / PVE management: install an operating system, allocate an IP, open a console, take snapshots, power on and off, back up and restore. The backends are not identical, so the table lists what has been verified for each one.
| Workflow | KVM | Proxmox VE | VMware ESXi |
|---|---|---|---|
| Automatic OS install | Windows (18 versions on x86_64, 6 on aarch64), Linux (23 distributions on x86_64, 13 on aarch64), custom ISO | Reinstall verified (LXC containers) | One-click provisioning, unattended Windows install, EFI network boot |
| IP allocation | — | IP pool | IP guard |
| Web console | ✓ | ✓ | ✓ |
| Snapshots | ✓ | ✓ (multi-disk group snapshots for HA VMs on distributed storage) | ✓ |
| Rescue system | ✓ | ✓ | ✓ |
| Port forwarding, traffic limits | ✓ | — | — |
| Passthrough | GPU / disk | GPU / USB / disk | GPU / disk (RDM) / USB |
| System disk on distributed storage From Business | ✓ | ✓ | ✓ (NVMe/TCP) |
| High-availability failover From Business | ✓ | ✓ (VMs) | not supported |
— = not in the verified list for that backend yet; ask us before relying on it. The rescue system boots into memory and never wipes the customer disk; cloud PCs have their own rescue system.
Read more: Virtualization · VM docs
Distributed storage deployment and management uses container-based orchestration; new clusters start on version 20.2.4. The same panel then puts that storage to work — VM disks, office PC system disks, website storage and members’ own storage.
Where the storage is used: office PCs that boot from distributed storage · tour: shared base image and capacity · Storage
An encrypted underlay built on the Noise framework (Nodekeeper’s own implementation) carries BGP EVPN and VXLAN between your nodes, with relays, IPv6 and DHCP. VMs keep their addresses when they move between hosts.
iptables / frp / SSH forwarding and the system firewall are managed centrally, so published ports and access paths are kept in one list instead of scattered config files. HAProxy and nginx reverse proxies are in Free as well.
Read more: Networking · Firewall docs · Reverse proxy docs
Detect → confirm it is safe to take over → fence the old host → choose a healthy host → attach storage → start → check the service. VM HA restarts supported VMs (KVM, PVE) on another host with the same IP and MAC; measured 65 s (PVE) to about 4–8.5 min (KVM) in tests.
Cold restart elsewhere with fencing so two copies never write the same disk. Confirm mode or automatic mode; simultaneous failures pause instead of acting.
No automatic takeover yet. If the site node that serves iSCSI to Windows office PCs fails, those PCs lose their disk until it is back or replaced.
The user’s cloud PC can be started on another compatible PC after the old one is fenced. Unsaved work on the failed PC is lost.
Tour: automatic VM recovery, with timelines · High availability
Join nodes to one panel, open a terminal on any host you are authorised for, and start, stop or inspect systemd services without switching tools. Each node runs Nodekeeper under its own licence.
Set up and watch primary/replica pairs for MySQL, PostgreSQL and MongoDB from the same panel that runs the database instances.
Nodekeeper is not asking you to compare hypervisors feature by feature. It focuses on connecting the work around them: system provisioning, IP allocation, hosting services, storage, networking and recovery. The useful comparison is the complete task you need to finish — and which manual steps remain in your actual environment.
| From plan | Infrastructure capabilities |
|---|---|
| Free | Multi-host web terminal, system services, iptables / frp / SSH forwarding, system firewall, HAProxy, DD-image reinstall |
| Personal | Mesh networking between nodes |
| Pro | KVM / Proxmox VE / ESXi management, mesh client for remote staff |
| Business | Distributed storage deployment and management, snapshots, VM HA, cloud PCs, database replication, sync groups |
| Enterprise | Same software capabilities as Business; terms agreed with sales |
Licensed per node — per machine that runs Nodekeeper. How distributed storage hosts and site nodes are counted for a cluster: ask us.
The trial is a monthly subscription whose first month is $0: card required, once per account, cancel before day 30 and you are not charged. It covers up to 3 nodes; for a larger cluster, talk to us first.