Virtual Machine Host

Refurbished Server

#

Here’s the first example of a refurbished server I found:

The specs are:

• Dell 1U Chassis
• Dual 16-core Xeons from 2016
• 256 GB of RAM in 8 DIMMs to use all the memory channels
• No disks, but 4 disk trays
• 4x 1GB Ethernet Ports
• iDRAC Enterprise for remote hardware management
• Dual power supplies

Refurbished disks are slightly overrpriced, so we’d want to add pair of 2TB SATA SSDs from Amazon.

• Server ($750) + Disks ($250) = $1000

Refurbished PowerEdge R630 @ ServerMonkey

Crucial MX500 2TB SATA SSD @ Amazon

Software Setup

#

Realistically, any modern community Linux distro would work for the host. Personally, I’d prefer the latest Debian Stable (currently 12) or Ubuntu LTS (currently 22.04) due to stability, reliability of automatic updates, and personal familiarity.

We don’t want the ET&S approved Oracle Enterprise Linux - the license cost for two sockets would be similar to the cost of the hardware.

For managing VMs at this scale there’s no need for any software beyond what’s available in the repos for a modern Linux distro (e.g. virsh, virt-manager, cockpit).

Networking

#

To be able to deploy virtual machines, we’d need a delegated IP block. This would be easiest to deal with with a separate router/gateway device.

To continue the 100% open source theme, this can be an OPNsense box. These are reasonably priced and come pre-installed:

https://protectli.com/product/fw4b/

A reasonable router/gateway config is about $300.

Where does this go?

#

Optimally, somewhere on the PSU campus where CS faculty can have access and students can have at least supervised access.

If there’s nowhere this can go in D&M, is there an existing rack in another building?

What would this let us do?

#

For the price of cloud hosting two small virtual servers for a year, we could instead host between 30 and 60 similar virtual servers for several years.

This would also allow us to run an Inkfish instance with several CPU cores. That would allow for automatic testing parallel speedup on student assignments that use multiple threads or processes, which would have been useful on a recent Data Structures lab.