>Want to confirm why we have implemented it in this way and if any >correction in understanding is required please update
This is not implemented intentionally - the issue is that delta files show up as 0 sized splits. You should be getting exactly as many splits as you have buckets x partitions in Hive-2.0. This fix can't be backported to hive-1.2.1 without a corresponding Tez upgrade to 0.7.x. <https://github.com/apache/hive/commit/76961d1f67a5d5e3614d3d81c417684fab92 c6c2> Cheers, Gopal