Amazon EMR, is made for data analytics on big data clusters. It comes pre-packaged with tools perfect for big data specialists. It can be made of hundreds of EC2 instances. EMR also supports autoscaling, and spot instances. EMR can be used for data processing, machine learning, web indexing, etc. In your EMR cluster you have the master node, core node, and task node. The master node is responsible for managing the EMR cluster and the health of the cluster. The core nodes run tasks and store data. The task nodes only run tasks. You have the option of on-demand, reserved and spot. For your master and core, it would be best to use on-demand or reserved for cost savings. Spot instances could be useful for task nodes.