The development team likes to call this feature the "Big Red Button."
You can use CloudWatch's Auto Scaling feature to scale your EC2 capacity up and down according to conditions that you define. You do this by creating an AutoScaling Group and attaching a trigger to it. A trigger uses CloudWatch Alarms to reference a CloudWatch metric that you choose, and is activated when the metric goes outside of a given range for the amount of time that you specify. For example, you can activate a trigger when the CPUUtilization metric exceeds 50% for a period of two minutes.
In certain cases you may want to prevent an AutoScaling Group from activating its triggers. You might want to do this if you are debugging your group and you don't want to worry about having all of the instances shut down. Or, a configuration error might make all of your instances appear to be unhealthy when they are actually doing just fine. Perhaps an external condition has caused a temporary drop in traffic to your site,and you want to maintain instances at the ready when it resumes.
This feature gives you very fine-grained control over which Auto Scaling behaviors are suspended. You can, for example, suspend all launches or all terminates. Alternatively, you can suspend just scheduled activities, or just health checking, and so on. Each behavior can be suspended and resumed independently
Next feature: Auto Scaling Follow the Line.