Sizing a Small Proxmox VE Deployment for Internal Business Applications

[mohsin045]

I'm setting up a small Proxmox VE environment for an office deployment with roughly 120 to 150 users. The main workload will be a couple of Linux VMs, a PostgreSQL database, and one internal application that staff use daily.

At the moment I'm trying to decide whether it's worth separating the application server and the database into different VMs or if keeping both on the same VM is still a reasonable approach for a deployment of this size.

I'm also interested in hearing how others handle the following:

• CPU and RAM allocation for business applications
• ZFS vs LVM-Thin for reliability
• Snapshot schedules
• Backup strategy with Proxmox Backup Server
• Database performance after Proxmox updates

The application itself is an attendance management system, and the data will eventually be synchronized with Vertex-HCM, but my main focus right now is building a stable Proxmox setup before moving it into production.

I'd appreciate any recommendations or real world experiences from people running similar office workloads on Proxmox VE.

 

[FrankList80]

My 2 cent: no of concurrent user is the main driver and what they are doing. It sounds like occasional time/entry but no heavy lifting. I would stay with the same VM for both PGSQL and and the Appserver. Give them a lot of RAM and run them not on spinning disks. A dedicated PBS i would use with dedicated NIC only for Backup. Should it be a HA setup which adds complexity? This will be argued here, i guess. I prefer put of simple no HA setups. Solid Servers with a lot of RAM and you good for 3-6 years. I would expect no stress from your requirements - "easy sailing".

The main "thing" is imo - "Uptime" - sounds like a 24/7 setup, so i would think of UPS, AC installed stable network from the Users etc. - the "Hygienic" factors so to say.

"Restore" of your main application is imo the 2nd "main" thing - if i were you i would test thoroughly with a TEST instance how usual snapshotting and restore even under strange circumstances "works" (or not). Legally wise with HCM loosing time-events could be big thing. Dependent on your results, you could think of using a more sophisticated Backup architecture with a COTS vendor supporting PVE. An offsite (but secure) PBS would imo the "Cherry" on the cake ... Good luck!