Auto Scaling is a concept which came into existence because of the wide spread use of AWS Amazon Web Service platform. It allows the infrastructure to scale elastically, or words as needed. We can create auto-scaling groups from scratch and use the existing production instance to scale up as needed. Auto scaling gives an extra layer of fault tolerance, high availability and flexibility, at a very effective cost. If we are able to predict or plan the traffic pattern, then we can do predictive scaling on top of auto scaling.
In case of HBase we can create a Auto Scaling groups of region servers, namenode, secondary namenode, use Amazons Elastic Map Reduce (EMR), most of the know tools are already integrated with Amazon EMR, like Pig, Hive, Ganglia, DistCp and so on.