hi all,
as you know there some methods while distributing data between consumer and producer sets. for ex: partition, hash, range ...
if you use an order by statement in a parallel select statement you probably see
range method because consumer sets creates a range to order the rows and Whenever a producer parallel execution server (parallel process) wants to send a row, it look for consumer parallel process which has appropriate range.
but about Broadcast and Round robin methods I couldnt understand what they do and how hey work... I looked for documentation but I couldnt find a document which explain them well. for ex if I write a select like that
select /*+ parallel(sales, 4) */ * from sh.sales
union
select * from sh.sales2
then I can see
Round Robin in the execution plan.
so what does "Broadcast" do and what does "Round robin" do.
thank a lot for answers.
Edited by: elcaro on Sep 25, 2011 11:00 AM