Proxmox VE Infrastructure Design Report
Prepared by: Jacob Freeman
Date: 5-11-25
Client: Butler Community College
1. Physical Infrastructure
We are making our physical infrastructure around the idea of minimizing downtime and
maximizing protection in case of attack or failure, with things such as UPSs and Multiple NICs,
so that in the event of an accident, data is safe and uptime can be prioritized. This information is
sourced from the Proxmox Requirements and Standalone Research.
● Number of Servers: 3 physical nodes to support clustering and high availability
● CPU: Intel 64 or AMD64 with Intel VT/AMD-V CPU flag
● RAM: Minimum 2 GB for OS and Proxmox VE services. Plus, designated memory for
guests. For Ceph or ZFS, additional memory is required, approximately 1 GB of memory
for every TB used storage
● Storage: Recommendation for Storage is 1TB SSDs with PLP ( Power Loss Protection)
○ For VM Storage, one may also consider setting up a RAID with cache protection.
● Network: 2x 10GbE NICs per server for redundancy and traffic separation.
● Redundancy: Servers will need a UPS ( Uninterruptible Power Supply), and all physical
setups should have multiple gigabit network cards to prevent downtime.
2. Network Design
In Accordance with the information in Step 1, we are aiming to minimize downtime. By using
Promox’s High Availability Feature, we can ensure that we minimize downtime to just 3.65 days
with 99% uptime. This information was sourced from the Proxmox Wiki.
Network Segmentation: We will be organizing multiple VLANs to sort and maintain different
data streams. This allows higher security, as you can have sensitive information move through its
own channel.
● VLAN 01: Core Network Traffic
● VLAN 02: Management
● VLAN 03: Guest
● VLAN 04: Backups
● VLAN 05: DMZ
● VLAN 06: Client
Physical Network: As mentioned in section 1, we will be maintaining multiple gigabit speed
NICs to maintain redundancy, alongside UPs as well. This means both power and network access
are maintained.
Firewall: Proxmox provides a Firewall that can rule for all nodes in a cluster, and also define
further rules for containers. We will be using this firewall to secure the nodes and data from
external attacks.
3. Storage Configuration
Storage will be kept simple, redundant, and very accessible. By utilizing Shared Storage,
we can configure as many storage pools as needed, and it also allows migrating machines to live
without downtime, and without the need to copy; it is very fast to do so. (Data sourced from
Proxmox Documentation)
● Primary Storage: NFS/iSCSI Shared storage.
● Capacity Planning: We will keep enough storage in pools to be able to account for
expansion for storage and be able to loop new storage into the shared pool.
4. Virtual Configuration
We will be using Cluster Nodes for our virtual configuration, in order to maintain High
Availability. This also comes with the benefits of being able to utilize Promox’s GUI to manage
all nodes at once. Source
5. High Availability (HA)
High Availability is the feature we designed our core network principles around. We will be
using it to its full efficiency, to maintain as much uptime as possible.
In Order to use HA, we will need the following, which have also been detailed in the Previous
Steps:
● shared storage for VMs and containers
● at least three cluster nodes
● hardware redundancy (everywhere)
● Use reliable “server” components
● Hardware watchdog - if not available, we fall back to the Linux kernel software
watchdog (softdog)
● optional hardware fencing devices
● Fencing: We will employ Fencing to make sure nodes that begin to have errors are
prevented from spilling over into other working nodes, and are removed from the shared
resources, so that as little resource gets used up while the error node is fixed.
● Start Failure Policy: This Policy will set how many times a service can attempt to start
on a node until it is moved to another node. This is to maintain stability and not lock
services onto a node experiencing an error, which would bring about downtime.
6. Backup Strategy
We will be using Promox’s in-built Full Backup system to maintain secure backups in case of
needed restoration after an incident or failure. We will be backing up to the PBS (Proxmox
Backup Server). This also gives us access to single-file restoration options as well. We will be
storing these backups on separate storage media as well to maintain them, as if kept on the same
storage, and if we suffer a storage failure, the backups would not serve their purpose. We will be
using Stop Mode Backups, which will incur a small downtime cost, but will allow the most
effective backups.
● Daily Backup
○ Domain Controller
○ Email Server
○ Perform automatically at low-usage times to minimize the impact of
downtime.
● Weekly Backup
○ DVR
○ Other non-critical data