The high-level difference is that anything Hadoop-based is oriented around "give me a job to run, I'll go crunch over the data, and spit the result back to you.

The high-level difference is that anything Hadoop-based is oriented around "give me a job to run, I'll go crunch over the data, and spit the result back to you." In "realtime" analysis, you tell the system "these are the queries I want you to run" and it continuously updates those with answers as data arrives.