Running mahout using hadoop on Amazon's EMR/EC2

Question

I want to migrate my current local hadoop cluster into amazon . In this hadoop cluster I am using services like mahout,hbase and hive . I have two option now in amazon either go for pure EC2 instances or Elastic map reduce cluster . I want some suggestion on what is better option to move the cluster which has these kinds of requirement .

Answer 1

I always suggest people to go for EMR, as that is managed and will be a bit more costlier than using pure ec2, but the headache and time you will spent in configuring the clusters and then managing them can be saved by running managed services like EMR.

• Mahout can easily be run like a custom jar.
• Hive cluster can also be launched within minutes.
• Similary for HBase, Amazon has recently added creating HBase cluster over EMR.

See other views here.