We have a web application running on a single small instance. Most of the time this is fine. However, occassionally our application does intensive queries and utilizes more CPU than the small instance can handle.
What I am wondering is; is there a way to have a SPOT instance (C1 High CPU Medium) launch and run while the spot price is low? Ie, have it run always as the ‘main server’ unless the spot price goes up; and then just revert (seamlessly) back to our reserved small instance – for the rare times the spot price increases?
Basically – a way to get a high cpu instance on the cheap…and our small instance will suffice ‘Most of the time’ anyway, so a failsafe to it is okay.
You should take look at AWS Spot Labs. You can get access to some advanced features that will allow you to
In the meanwhile, you can use your reserved or on-demand instances in an auto-scaling group. Then while your Spot server is running, the load should be minimal and you will have a minimal number of on-demand instances. Once the spot instance is going down, you will start scaling up the on-demand (or reserved) instances to compensate for the lack of the main machine.
Actually you can even have your Spot Instances in a different (lower threshold) auto scaling group and launch several cheaper spot instances when they are available. See more details in: http://docs.amazonwebservices.com/AutoScaling/latest/DeveloperGuide/US-SpotInstances.html