As I recall Hive 2 now officially recommends using Spark or Tez as execution engine instead of Map-Reduce (MR)
hive> set hive.execution.engine=mr;
Hive-on-MR is deprecated in Hive 2 and may not be available in the future versions. Consider using a different execution engine (i.e. tez, spark) or using Hive 1.X releases.
I still run my Hive 2 on Spark 1.3.1 (I know some state that it works with 1.4.1 but that is still an older version of Spark) and even with my 1.3.1 version of Spark engine things have improved in terms of stability and performance.
Now as a consumer I can verify that for larger tables, Hive 1.2.1 queries (even simple things like COUNT(*) or INSERT/SELECT) used to crash before and I had to switch to MR as the execution engine. On Hive 2.0 it all works fine.
However, I guess the 60K question is what has been done to make Hive 2 work with newer versions of Spark like 1.5.2 and 1.6, given that Hive 2 encourages using Hive on Spark and/or Tez.