How to Map Two Tables To One Class in Fluent NHibernate?

Why Hive's Sort Merge Bucket Map Join does the join in just one mapper?

• I have a Hadoop cluster and I use Hive for querying. I have  two large tables, which are bucketed and sorted on the join key. So, I  think using "sort merge bucket map join" would be helpful. I set the required flags and then run the query. I see that it starts a job which has 1 mapper and no reducers. Question: Why Hive does the join in one mapper? Did I  miss anything? Because tables are bucketed on the join key, it seems  that join can be done in multiple mappers in parallel. Why Hive doesn't  do that?

• Answer:

  It's tough to say without taking a look at the explain plan and/or the mapreduce job's job.xml because the answer here may depend on a few things depending on your hive configuration, the format of your data set, the metadata associated with it among other things. 2 things that jump out of the top of my head are: 1. InputFormat: Set by hive.input.format property, it may be a format that combines the buckets when sending them to the mapper. 2. Size of your dataset and buckets: If the bucket sizes are small, they may be bundled together before being sent to mappers.