My initiation internship project at SNRT (Morocco's national broadcasting company), during the first year of the engineering cycle (IRIC, ENSA Khouribga). The goal: design and deploy a high-availability, fault-tolerant Proxmox VE cluster on physical servers, backed by shared NAS storage — so a virtual machine keeps running even if a whole host fails.
⬇ Download the full report — PDF, 94 pages (French)
The report is written in French. This is a short English summary of what it covers.
What the project covers
The report starts from the fundamentals and builds up to a working cluster:
- Virtualization — virtual machines, hypervisors (type 1 vs type 2), and why they matter.
- Clustering & high availability — what a cluster is, how HA works, and the role of quorum in keeping a cluster consistent.
- Shared storage — SAN vs NAS, and RAID levels for redundancy.
- Proxmox VE — the open-source virtualization platform, its features, and installation.
The build
The practical core deploys a two-node Proxmox VE cluster on physical servers with a shared NAS over NFS, plus everything needed to run it in production:
- Physical architecture & network plan — two servers (managed through a KVM switch) and the IP addressing scheme.
- Cluster configuration — installing Proxmox on both nodes and joining them into a cluster.
- The two-node quorum problem — a 2-node cluster can't form a majority on its own, which makes HA unstable; the report explains why and configures a fix so the surviving node keeps running after a failure.
- Shared NAS via NFS — configuring the NAS and adding it as cluster storage, so VM disks live on shared storage and can move between nodes.
- Networking & firewall — a private NAT network and the Proxmox VE firewall.
- Users & permissions — groups, user accounts and role-based access.
Tests & validation
The cluster is validated against the scenarios that matter for a broadcaster that can't afford downtime:
- High Availability — kill a node and watch its VMs restart automatically on the other.
- Live migration — move a running VM between nodes with no downtime.
- Firewall and NAT network tests.
Tech stack
Proxmox VE · KVM · NAS / NFS · RAID · Clustering / HA · Quorum · NAT · firewall