Hive With tez

[saurabh <mp...@gmail.com>]

Hi,

We are in process of exploring TEZ for Hive 0.14.
Needed some pointers to start on Hive with Tez.
E.g. in Hive HDFS Block size plays a vital role in getting the number of
Mappers and later independent execution of mappers can accelerate
processing substantially.

I understand this is a very vast topic and cannot be described, however
some quick pointers will be helpful.

I am currently working on:
Query vectorization and COB with ORC tables.

Thanks,
Saurabh

Re: Hive With tez

[Jeff Zhang <zj...@gmail.com>]

Regarding the mapper task number, Hive on tez is very similar with Hive on
MapReduce. One difference is that hive on tez can group split together
which may use less tasks than mapreduce.  What issues did you see when you
use hive on tez ?

On Sun, Jul 5, 2015 at 10:39 PM, saurabh <mp...@gmail.com> wrote:

> Hi,
>
> We are in process of exploring TEZ for Hive 0.14.
> Needed some pointers to start on Hive with Tez.
> E.g. in Hive HDFS Block size plays a vital role in getting the number of
> Mappers and later independent execution of mappers can accelerate
> processing substantially.
>
> I understand this is a very vast topic and cannot be described, however
> some quick pointers will be helpful.
>
> I am currently working on:
> Query vectorization and COB with ORC tables.
>
> Thanks,
> Saurabh
>



-- 
Best Regards

Jeff Zhang