Proxmox multi-cluster management: which console to pick

Three open-source consoles manage several Proxmox clusters from one interface, and the first thing that sets them apart is who builds them. Proxmox Datacenter Manager comes from the company behind Proxmox VE: it is free and live-migrates VMs between clusters, but it only knows Proxmox VE and Proxmox Backup Server. PegaProx and ProxCenter are third-party consoles started in 2026, and both measure themselves against vCenter. PegaProx also drives XCP-ng pools as a tech preview, and its docs recommend handing it each cluster's root login. ProxCenter keeps migration, load balancing and SSO for its Enterprise edition, billed per node.

The vendor's console or a third-party one

Proxmox Datacenter Manager (PDM) calls itself a high-level control plane. It collects metrics and task logs from the installations you register with it, which it calls "remotes", runs everyday operations on them, and hands you off to each cluster's own interface for detailed configuration. Guest configuration beyond snapshots, power state and migration is still a roadmap item. The docs stress that loose coupling: if the console goes down, the remotes keep running.

ProxCenter's repository describes it as an alternative to VMware vCenter for Proxmox, and it borrows vCenter's landmarks: a vSphere-style inventory tree (clusters, nodes, VMs, storage, pools, tags) and a load balancer named DRS. PegaProx's website also lines its features up against vCenter and the native Proxmox interface, yet the project describes itself as a layer on top of Proxmox VE, "without replacing it". It edits VM settings from its own interface (CPU, memory, disks, network, EFI, Secure Boot), and its self-service portal is built for a hosting provider's customers.

Proxmox VE, PBS and XCP-ng: what each console connects to

PDM accepts Proxmox VE 8.4 or later and Proxmox Backup Server (PBS) 3.4 or later, according to its 1.0 announcement. The only new remote type on its roadmap is Proxmox Mail Gateway, and no other hypervisor appears there. ProxCenter connects to Proxmox VE and PBS; VMware ESXi and vCenter, Hyper-V, Nutanix AHV and XCP-ng only show up as migration sources, in the Enterprise edition.

That leaves PegaProx as the only one of the three that manages a hypervisor other than Proxmox. On an XCP-ng 8.2 or later pool, reached over XAPI, it powers VMs on and off, opens their console, manages disks and network interfaces, and puts a host into maintenance once its VMs are evacuated. The README files this integration under "Tech Preview", and the docs warn that some features may be limited or change.

Moving a VM from one cluster to another

All three consoles move guests between clusters through the same piece: the Proxmox VE remote-migrate API. PDM's code (server/src/api/pve/qemu.rs) and PegaProx's (pegaprox/core/manager.py) call it, and ProxCenter's docs name it. The Proxmox VE documentation still flags qm remote-migrate as an "EXPERIMENTAL feature!".

PDM migrates live, and its 1.0 announcement says the two clusters need no shared cluster network. PegaProx also migrates live by default, but its code switches to offline migration on its own once the disks add up to more than 100 GB, unless you force it: a comment there explains that the Proxmox WebSocket ticket expires before the memory sync finishes (pegaprox/api/vms.py). ProxCenter only moves a VM between clusters offline, and only in the Enterprise edition. Proxmox refuses such a move for a guest that is HA-managed, replicated or carrying snapshots, so the migration dialog has you clear all three, and ProxCenter restores HA and replication on the source if the move fails.

To move off another hypervisor, PegaProx migrates from ESXi to Proxmox VE or XCP-ng, and between Proxmox VE and XCP-ng in either direction, without snapshots. Its 1.3.0 notes leave one issue open: the "Snapshot-Iterative" ESXi migration can report success while data is missing on the target. ProxCenter Enterprise imports into Proxmox VE from ESXi, vCenter, Hyper-V, Nutanix AHV and XCP-ng, with warm (CBT) migration from ESXi, vCenter and XCP-ng when the target node runs Proxmox VE 9. PDM's docs plan no import from another hypervisor.

On load balancing, PegaProx moves VMs across the nodes of a cluster and across clusters, ProxCenter's DRS balances the nodes of a single cluster in the Enterprise edition, and neither PDM's docs nor its roadmap mention automatic balancing.

What the consoles show of your backups

PDM has handled Proxmox Backup Server as a remote since 1.0: datastore contents and usage, plus metrics on the dashboard. Tracking backup jobs, with the status of the last run, is still on its roadmap. ProxCenter's Community edition manages several PBS servers and lists their backups; reports are an Enterprise feature. PegaProx schedules backups itself, tracks the age of the last backup per cluster and datastore against the RPO you set, and can verify a backup by restoring it and booting the VM, a feature it labels experimental.

The access each console asks for on your clusters

PegaProx's docs "strongly recommend" root@pam with each cluster's root password, which its database stores encrypted. With an API token or a limited user, four features stop working: rolling updates, SMBIOS auto-config, two-node HA and the node shell. ProxCenter's docs take the opposite line: an API token where possible, and in production a dedicated service account rather than root@pam. ProxCenter only uses SSH for disk imports, rolling updates, CVE scans and network flow collection. PDM connects with an API token, either an existing one or one it creates from a login when you add the remote (ui/src/remotes/wizard_page_info.rs), and opens remote shells through a token on Proxmox VE 9.1 and PBS 4.1 or later.

What each console needs to run

Read from the repositories and documentation on 7 October 2026.

PDM is best installed from its ISO, and the 1.0 announcement keeps installing it next to another Proxmox product for expert use only. PegaProx fits in one Python process and one SQLite file, which SQLCipher encrypts on Linux x86_64 only; its docs count 1 core and 1 GB of RAM for up to five clusters, and 8 cores and 32 GB beyond 200. Its install script pulls the main branch rather than a tagged release, and for production the README points to an APT repository run by a contributor, gyptazy. ProxCenter ships an installer that sets up Docker, Compose and PostgreSQL under /opt/proxcenter; without Docker, only the Community edition builds, with Node.js 22.12 and PostgreSQL 16 or later. On supported Proxmox versions, its README asks for 8.x or 9.x, while its connection guide accepts Proxmox VE 7.0 and PBS 3.0.

An outage of the console itself plays out differently in each project. The clusters don't need PDM to keep running, and its roadmap is evaluating an active-standby setup, noting that two instances side by side already cover the case at the cost of collecting metrics twice. Since 1.3.0, PegaProx can group up to four instances around a leader, and an admin promotes a standby by hand. ProxCenter keeps high availability for Enterprise: three nodes behind a Keepalived virtual IP, with PostgreSQL replicated by Patroni.

Where the projects stand

Proxmox Datacenter Manager took about a year to leave alpha: 0.1 on 19 December 2024, 0.9 beta on 11 September 2025, 1.0 on 4 December 2025, then 1.1 on 28 May 2026. The repository's Debian changelog stops at 1.1.7 from 15 July 2026, and commits are still landing in October. The code lives on git.proxmox.com, and the GitHub repository is only a read-only mirror.

PegaProx and ProxCenter opened their repositories seventeen days apart, on 18 January and 4 February 2026, and both ship far more often. PegaProx has published 57 releases since 0.6.0 on 25 January 2026. Its 1.0 ended the beta on 1 August 2026, and 1.3.0, out on 7 October, lists 38 behaviour changes; it fixes a silent data loss in the ESXi "Live Clone" migration and several authorization flaws, some of them introduced by the 1.2.0 fixes. The README describes a three-person team, and two developers account for 1,095 of the 1,296 commits. ProxCenter has shipped 25 releases, from 1.0.0 on 1 March 2026 to 1.4.11 on 5 October, ten of them in April. The adminsyspro account wrote every commit apart from Dependabot's, and 1.4.0 on 12 May 2026 swapped SQLite for PostgreSQL with no automatic data migration.

Licences and paid plans

Proxmox Datacenter Manager is under AGPL-3.0 or later, with no subscription of its own. The enterprise repository and support come with Basic or higher subscriptions on your remotes, provided at least 80% of the configured nodes hold one. Below that, you get the pdm-no-subscription repository, which the docs reserve for testing and non-production use.

PegaProx is under AGPL-3.0 only, with an extra term under section 7(b) in its NOTICE file: the "Powered by PegaProx" line in the client portal and the copyright line in the interface must stay visible on any instance users can reach, which rules out white-labelling. Every feature is free, and commercial support comes from independent companies that sponsor the project.

ProxCenter splits into two editions. Community, under AGPL-3.0 or later, has no node limit and covers inventory, storage and Ceph, backups, events and users. Enterprise, under a proprietary EULA governed by French law, adds migration, DRS, rolling updates, site recovery, RBAC with LDAP or OIDC, reports and multi-tenancy, all running on a closed-source orchestrator written in Go. The README still lists migration and "RBAC & SSO" without marking them Enterprise. The subscription is annual and counted per managed node, with no published price. A connected licence holds a 30-day lease, renewed by a daily sync that sends the key, the version and the node count; an offline mode has been available since 1.4.11. Past the node quota, the install turns read-only. The EULA also bars providing managed services to third parties without a separate written agreement, while the MSP mode is aimed at hosting providers.