-Re: Does Hive 0.11 have Query Flattening optimizations?
Aniket Mokashi 2013-06-27, 21:30
I am almost certain that both the queries will compile to the same MR plan
in hive (I haven't checked though).
In fact, having clause is a syntactic sugar to achieve whats done in the
nested query. Take a look at:
On Wed, Jun 26, 2013 at 10:46 PM, Mihir Kulkarni <[EMAIL PROTECTED]>wrote:
> Does hive support Query Flattening? For example a query like this:
> *SELECT alias.a0, alias.a1*
> * (SELECT COUNT(b) AS a0, c AS a1*
> * FROM test*
> * GROUP BY c) alias*
> *WHERE alias.a0 > 2;*
> would be flattened into:
> *SELECT COUNT(b), c*
> *FROM test*
> *GROUP BY c*
> *HAVING COUNT(b) > 2;*
> Does Hive (0.11) have such kind of optimizations or are they even useful
> considering all queries are ultimately converted into MapReduce jobs? At
> Informatica Corp we rely on Hive a lot and hence are interested to support
> such optimizations.
> Thanks in anticipation.
> *Mihir Kulkarni
> **Software Engineer | Data Engine
> Informatica Corporation*