How to Plan Server Capacity

jkirkwood's picture

Who wants to be able to tell the future? That’s what a lot of people want in terms of server capacity planning and this sets their expectations as to how hard it might be to get right. 

However, it’s not as hard as you may think. Servers do not require your entire budget for them to work properly. Having a solid deployment plan, as well as an understanding of what that application/machine is doing, can save time and money. In order to accurately allocate resources, you just need to keep in mind the current load and its growth. I’ll explain below:

I’m going to start by assuming you have monitoring. You’re reading a blog about it, so it’s a fair assumption!

Take a baseline average. Look at your average usage over a period (let’s say a day) and capture it. Record the relevant information such as CPU utilization, load average, network utilization and memory use. 

After this, you need to either compare it to historical data or wait things out. If you have the data stored in your Opsview data warehouse (data archive). you can look over it using our Graph Center (excel or a graphing/reporting engine) and then look at how its growing over any given time period. Your graph should look like this:

Server Capacity Planning


Take this line and draw a line like what is shown above. That’s all you need to do! You can extrapolate the growth rate from that and end up knowing ‘roughly’ how much resources you will need at a given point.