In this KB:
- Important metrics to consider in deciding to scale up
- Explanation of what to look for in the graph data
In this tutorial, we will show you when you need to scale up your Cloudways server.
There are three important monitoring graphs which you need to consider when trying to decide if it is time to scale.
These are “Idle CPU“, “Free Memory” and “Auto-healing Restarts”.
Step 1: Server Management
Tap on “Servers” on the top menu bar. Then, click on your target server from the list.
Step 2: Server Management
Click on “Monitoring” button under “Server Management.”
Step 3: Metrics to watch
Choose the Idle CPU graph.
- Idle CPU speaks to the rate of FREE (not utilized) CPU.
- So the lower the value, the busier your server is.
- This is the MOST IMPORTANT metric while examining your server wellbeing. If you observe that your server has reliably (i.e. over a time of a few hours) under 10/20 % free CPU, you have to scale up your server. On the off chance that free CPU reliably touches zero as in the screenshot below, you have to scale critically as most likely site execution is enduring, clients/guests will encounter timeouts and procedures may not finish.
Choose the “Free Memory” graph:
- Free memory (in MB) on your server.
- Higher is better.
- Free memory is not as great a marker of server health as Idle CPU. In general, the working framework will tend to utilize the greater part of the accessible memory, and this is particularly valid for smaller servers (512MB to 2GB). So you can see low free memory, yet server can be running superbly fine. As a general guideline, for larger servers (4GB and up) there ought to dependably be above 100MB of free memory. For smaller servers, the server can be working splendidly fine with free memory somewhere around 50MB and 15MB.
- What will be a better indicator of memory issues is if you start receiving monitoring alerts emails. These are generated when one of the basic services in the web stack (Nginx, Varnish, MySQL) crash and can’t be automatically recovered. This will usually happen when there is insufficient memory to begin the service.
- You may want to read this very interesting Reddit thread where we discussed free memory and server health.
Choose the “Auto-healing Restarts” graph:
- Some automatic core services restart due to resource (CPU, memory) issues.
- It should be zero.
- Auto-healing restarts is another extremely good indicator of your server health. Even a few scattered restarts mean that you server is low on resources and that the operating system is killing vital processes due to it. In the image below, you see multiple daily restarts indicating an absolute need for scaling the server.
Step 4: Vertical Scaling
Once you have decided that you need to increase your server size, here is the process:
- Click on “Vertical Scaling”.
- Increase your server as per your need.
- Press the “Scale Server” button.
You can read this more detailed KB outlining this process.