As your website or application grows, you may need more CPU, RAM, or storage. Cloud servers make it easy to scale - this guide explains when and how to do it.
Vertical Scaling (Upgrading Your Server Plan)
Vertical scaling means upgrading your existing server to a larger plan - more CPU cores, more RAM, more storage - up to a dedicated-CPU or bare-metal server for heavy, sustained workloads.
When to Scale Up
Watch for these signs:
- CPU consistently above 80% - your server is underpowered for current load
- RAM usage consistently above 85% - risk of slowdowns and OOM (out of memory) crashes
- Disk nearly full - time to upgrade storage or add a volume
- Response times increasing - even though traffic hasn't changed dramatically
How to Scale Up
- Log in to your control panel
- Navigate to your server's Resize, Upgrade, or Scale section
- Choose a larger plan
- Confirm - the server will typically reboot briefly
Note: Most providers can scale up with a simple plan change. Scaling down may require a reinstall or manual migration, depending on the provider.
Horizontal Scaling (Adding More Servers)
Horizontal scaling means running multiple servers rather than one large one. This is for high-availability, large-scale applications.
Common Architecture
[Load Balancer]
/ \
[Server 1] [Server 2]
\ /
[Database Server]
- A load balancer distributes traffic across servers
- Each server runs the same application
- The database runs separately and is accessed by all app servers
This is more complex but provides redundancy - if one server fails, the others keep running.
Monitoring Resource Usage
Before scaling, confirm what's actually consuming resources:
# CPU and RAM usage
htop
# Disk usage
df -h
# Top memory consumers
ps aux --sort=-%mem | head -10
# Top CPU consumers
ps aux --sort=-%cpu | head -10
Resource Usage Reference
| Metric | Healthy | Warning | Critical |
|---|---|---|---|
| CPU | < 60% avg | 60–80% | > 80% sustained |
| RAM | < 75% | 75–90% | > 90% |
| Disk | < 70% | 70–85% | > 85% |
| Load avg | < # of CPUs | = # of CPUs | > # of CPUs |
Questions? Email us at [email protected] - we reply in under 2 hours, 7 days a week.