What Proxmox VE is: KVM virtual machines, LXC containers and a web UI

Proxmox Virtual Environment (Proxmox VE, often just “PVE”) is a complete server operating system for running virtual machines and containers. You install it directly on a physical server (“bare metal”), and it turns that server into a hypervisor host, called a node. It is based on Debian GNU/Linux and its source code is released under the GNU AGPL, version 3.

Two kinds of guests

Proxmox VE calls everything that runs on a node a guest. There are two kinds:

• Virtual machines (VMs) use KVM and QEMU. Each VM emulates a full computer with its own kernel, so it can run Linux, Windows or other operating systems. VMs are managed with the qm command.
• Containers (CTs) use LXC. A container shares the host’s Linux kernel, so it only runs Linux, but it starts in seconds and uses very little overhead. Containers are managed with the pct command.

Every guest has a numeric ID (the VMID), such as 100. IDs below 100 are reserved, and in a cluster each ID must be unique across all nodes. Configuration files live in /etc/pve/qemu-server/<VMID>.conf for VMs and /etc/pve/lxc/<CTID>.conf for containers.

Ways to manage it

• Web interface: after installation, open https://<node-ip>:8006 in a browser and log in as root with the Linux PAM realm. From here you can create guests, open consoles, manage storage, networking, backups, the firewall and the cluster.
• Command line: tools such as qm, pct, pvesm (storage), pvecm (cluster), vzdump (backup) and ha-manager, each with a man page.
• REST API: everything the web UI does goes through a JSON REST API. pvesh lets you call it from a shell on the node, for example pvesh get /nodes.

A first look around from the node’s shell:

qm list                 # virtual machines on this node
pct list                # containers on this node
pvesm status            # storages and free space
pvesh get /nodes        # nodes, through the API

Clusters without a master

You can run Proxmox VE on a single node or join several nodes into a cluster. There is no central management server: every node can manage the whole cluster. Configuration in /etc/pve is stored in the Proxmox Cluster File System (pmxcfs), a database-backed file system that Corosync replicates to all nodes in real time. A cluster enables live migration of VMs between nodes and optional high availability (HA), where guests are restarted on another node if one fails.

Storage and backup

Disk images can live on local storage (a directory, LVM, LVM-thin, ZFS) or on shared storage (NFS, CIFS, iSCSI, Ceph RBD and others). Shared storage is what makes fast live migration and HA practical, because every node can reach the same disks. Backups are built in through vzdump, and Proxmox also makes a separate product, Proxmox Backup Server, for deduplicated, incremental backups.

Subscriptions and updates

The software is free to use. A paid subscription gives access to the enterprise package repository and support. Without one, you switch to the public no-subscription repository to receive updates; see the entry on repositories and updates.

When to choose a VM or a container

• Choose a VM for non-Linux systems, for workloads that need their own kernel or kernel modules, for strong isolation, and when you need live migration.
• Choose a container for lightweight Linux services such as DNS, a reverse proxy or a small web app, where low overhead and fast start matter. Running containers cannot be live-migrated; they are moved with a short restart instead.

Official documentation

• Proxmox VE Administration Guide: Introduction
• Installing Proxmox VE (accessing the management interface)