X

Information, tips, tricks and sample code for Big Data Warehousing in an autonomous, cloud-driven world

  • March 24, 2015

Finding the Distribution Method in Adaptive Parallel Joins

Yasin Baskan
Director, Product Management

Parallel Execution(PX) uses a producer/consumer model which requires distribution of rows between producers and consumers (two PX server sets). Until 12c, the distribution method was decided at parse time based on optimizer statistics. 12c brings adaptive parallel distribution methods which enables the distribution method decision to be taken at runtime based on the actual number of rows. This functionality is explained in the "Optimizer with Oracle Database 12c" white paper in the "Adaptive Parallel Distribution Methods" section. This was also discussed by Randolf Geist here.

So, how can we find out which distribution method was chosen at runtime? One approach is to use the view V$PQ_TQSTAT as explained by Randolf in his post, or look at the rows processed by producers and consumers in SQL Monitor as explained in the above mentioned paper. But, the easiest way to find this information is to look at the OTHER column in SQL Monitor.

The OTHER column shows a binocular icon in the lines with PX SEND HYBRID HASH row source.

When you click on those icons you can see the distribution method used at runtime.

As the DOP downgrade reason I talked about in my previous post, this is also a cryptic number code as of 12.1. For the adaptive distribution methods there are three possible values. The mappings for these codes are; 6 = BROADCAST, 5 = ROUND-ROBIN, and 16 = HASH distribution. 

Any comments and feedback welcome. 

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.