Re: PX Deq Credit: send blkd

Hi Mark,

Bobak, Mark schrieb:
> I'm not in a position to test at the moment, but how about:
> create table new_tab parallel 32 as select /*+ parallel(t 32) */ * from tab1 t;
> ??

I tested with and without ALTER SESSION FORCE PARALLEL DDL|DML PARALLEL 32; If it or the suggested hint is not set, the producers don't parallelize at all.

> Also, you may want to add 'nologging' to make it fly even faster. (But, consider the recovery implications.)
> create table new_tab nologging parallel 32 as select /*+ parallel(t 32) */ * from tab1 t;
The wait event has nothing to do with loggining, there isn't enough IO for that, on behalf of the only CPU that's used to write.

> Of course, this assumes you have enough CPU and I/O available to support 32-way parallelism, else you may be in for a disappointment.
It is :)

Starting 4 Statements of the same kind, updating different tables/partitions are producing 4x250MB/s write speed without any harm.

Thanks
Martin

-- 
Usn's IT Blog for Linux, Oracle, Asterisk
http://www.usn-it.de

--
http://www.freelists.org/webpage/oracle-l