The rule that matters most
For the vast majority of web projects, RAM runs out before the CPU does. A server short of memory starts swapping, and swap on an NVMe SSD remains hundreds of times slower than RAM — this is the number one cause of "slow" sites on an otherwise correctly sized VPS.
Guidelines by use case
| Use case | Recommended configuration |
|---|---|
| Brochure site, blog, portfolio | 1 GB RAM, 1 vCPU |
| CMS with caching and a few extensions | 2 GB RAM, 1 vCPU |
| Online store, lightweight ERP, API | 3 to 4 GB RAM, 2 vCPU |
| Business applications, active database, multi-site | 5 GB RAM, 3 vCPU and more |
Signs that your VPS is undersized
- Response time collapses under load but returns to normal at rest.
free -hshows swap being used persistently.- The OOM killer process is killing your services (visible with
dmesg | grep -i oom). - The load average sustainably exceeds the number of available vCPUs.
Signs that you are overpaying
A CPU that stays below 10% utilization at peak and memory that never goes beyond half mean you can move down a tier. Check before renewing.
Can you change size later?
Yes. Migrating to a higher tier is done without reinstalling: your data comes with you. That is why we recommend starting small and scaling up when a real indicator justifies it, rather than oversizing as a precaution.
Storage
All our plans use NVMe SSD, from 30 GB to 100 GB. Monitor space with df -h: a full disk causes more brutal failures than a lack of CPU. Logs and local backups are the first two culprits.
