I'm a web developer with more than two years of experience, and I recently took on a complete project that also requires me to make the hosting and infrastructure decisions. The client's existing server is too old to run the application stack: React, Node.js, and PostgreSQL.
The current options are a new server costing about $4,500, or more than $6,000 after licenses and related expenses, versus two small PCs with Ryzen 5 3500U processors, 16 GB of DDR4 RAM, and 512 GB SSDs, plus an automated backup setup for roughly $1,000.
The application has fairly modest requirements: around 12 concurrent users, a database currently about 700 MB in size, and roughly 1.3 GB of data to process if all users run the most demanding map-related task at once. The map data may be reducible to under 100 MB by generating a static map file.
I'm concerned that spending $6,000 on a server would be excessive for this workload. Two small PCs seem powerful enough, and the second machine could act as a standby if the primary fails. My main concerns are reliable data recovery, tested backups, and avoiding a complete outage after a hardware failure. Would on-premises hardware make sense here, or should I use a VPS or another cloud-hosted option?
4 Answers
This workload is tiny for modern infrastructure. A basic VPS would likely handle it comfortably and cost only a few dollars per month, while also avoiding hardware purchases, power, cooling, and maintenance. You could run the application and database in containers, automate database and file backups, and keep copies with a separate storage provider. Compare the monthly hosting cost with the actual cost of downtime and the time you’ll spend maintaining physical machines.
You probably don’t need an expensive enterprise server for this capacity. A used business server or workstation running a stable Linux distribution may be more than enough. However, make sure the decision includes support, monitoring, patching, replacement hardware, and who is responsible for fixing it outside business hours. Consumer mini PCs may lack ECC memory, redundant storage, and easily replaceable parts, so the lower purchase price comes with more operational responsibility.
If the application must stay on-site, two small PCs can be sufficient for twelve users, but they are not automatically high availability. Use one as the production host and the other as a warm standby or backup target. Keep encrypted backups off-site, retain multiple versions, and perform a restore test before launch. Backups stored only on the same machine or in the same building won’t protect you from theft, fire, ransomware, or a failed drive.
Before choosing hardware, clarify the required recovery time and recovery point. Ask how long the client can tolerate an outage and how much data they can afford to lose. Those answers determine whether a simple VPS with automated snapshots is enough or whether you need a standby system and replicated backups. Also document who owns the infrastructure and who handles failures, updates, security, and recovery after the project is delivered.
That makes sense. I was focused on buying hardware, but I should compare the total cost and recovery process rather than just the initial price.