AWS currently offers five levels of T2, Burstable Performance, instances on EC2. T2 instances “provide a base level of CPU performance with the ability to burst above the baseline.”
What’s not obvious from the general documentation page is that what’s the easy indicator to determine if you need to upgrade your T2 instance to the next level.
T2 Instance CPU Baseline Performance
What I found to be the best/quickest way to figure out when to upgrade T2 instance is to look at CPU Utilization (Percent) chart on the monitoring section for the instance. You need to compare the average usage of the CPU and the baseline of the T2 instance (above the baseline means the server is in the “burst” mode which uses the extra CPU credits).
Here are the baseline of each T2 instances:
|Instance Type||CPU Baseline Performance|
CPU Credit Usage and CPU Credit Balance
CPU credit usage chart and CPU credit balance chart are also good indicators. However, when the credit usage is too rapid, you might not be able to catch the issue in time to resolve the performance issue. The decrease in the credit balance is recorded only AFTER the server is performing beyond baseline performance. In other words, it’s not a good PREVENTATIVE indicator.
This is a simple way to measure the performance of the server reflected on CPU usage. There are other resources that causing the issue on your server(s). The most common issue I experience is the depletion of available memory. This happens if I am running a multiple servers (such as LAMP) under one instance. While CPU usage is within a normal range, not too uncommon to run out of available memory and your application stop running properly. I find that t2.nano instance and t2.micro instance aren’t sufficient to run a LAMP in most cases.
Resources: New Low Cost EC2 Instances with Burstable Performance
Resources: EC2 Update – T2.Nano Instances Now Available
Resources: Burstable Performance Instances