Azure or Proxmox?

The two offerings meet different needs, but their eligibility follows the same common non-negotiable rules. A need that breaks one of them is outside the lab scope, whatever the platform. For a recommendation in two minutes: what do I need?

Two offerings, two processes

Before the technical criteria, a practical difference that often weighs just as much: who creates the environment, and therefore how long you wait.

Azure lab

You create the lab yourself, a simple e-mail declares it.

Who creates it

You: you have the rights to create your Resource Group.

What you send

A declaration e-mail, 3 pieces of information.

Lead time

Immediate: no approval to wait for.

Create an Azure lab →

Proxmox lab (on-premise)

The administrators create the lab, you request it.

Who creates it

The administrators: Pool, VLAN and permissions.

What you send

A detailed request.

Lead time

Processing time on the administrators’ side.

Request a Proxmox lab →

When to choose which

Azure lab

• Azure training: upskilling on a service or an architecture.
• Need for managed services: AKS, Databricks, PaaS, serverless.
• POC or client demo requiring PaaS/IaaS services.
• Environment shared between several consultants on the same engagement.

Proxmox lab

• Classic VM or LXC needs, with no dependency on managed cloud services.
• Self-hosted solutions to test or evolve.
• Internal network constraint: access to resources that can’t be exposed on the Internet.
• Low latency, data sovereignty, tests close to the existing infrastructure.

Two options that avoid a lab

An identity? Users, groups and app registrations, self-service in Entra ID. A place to share? Sites, lists and libraries in SharePoint. No lab, no e-mail, no cost.

Detailed comparison

Criterion	Azure lab	Proxmox lab
How to get it	You create the Resource Group yourself, then declare the lab by e-mail. No approval to wait for.	You send a request: the administrators create the Pool, the VLAN and the permissions.
Recommended for	Azure training, managed services (AKS, Databricks, PaaS), cloud testing, shared labs.	VMs, LXC, self-hosted solutions, internal network constraints, sovereignty, on-premise testing.
Location	Four open regions: Switzerland North, Switzerland West, West Europe, Sweden Central. The service catalogue differs from one to the next.	Dedicated on-premise cluster pve-lab-cluster.
Network isolation	One dedicated VNet per lab. Peering allowed between labs, never to a production VNet.	One dedicated Pool and VLAN per lab, no routing to production.
Mandatory tags	owner, projectName, creationDate (denied if missing). expirationDate recommended, Client optional.	lab, owner-{trigramme}, env-lab + Pool YAML metadata.
Indicative costs / quotas	≈ 200 CHF / month / lab.	8 vCPU, 16 GB RAM, 200 GB disk.
Default security	NSG per subnet, SSH/RDP restricted to the Chancy IP only (194.148.102.82).	Firewall denies inbound / allows outbound, no direct public IP, MFA required.
End of life	VM auto-shutdown at 7 pm on weekdays and over the weekend. Resource Group deleted on expiry.	Snapshots deleted after 7 days at most. Pool deleted by the administrators.
Maximum duration	3 months, unless an extension is approved.	3 months, unless an extension is approved.