You may find this SUGI paper useful:

http://www2.sas.com/proceedings/sugi28/151-28.pdf

HTH,

Ben.

On Fri, 11 Feb 2005 10:31:56 -0800, ptmanolakos@HOTMAIL.COM wrote:

>Hi, > >Our site is planning to migrate to SAS release 9.1.3. However, our >quality control policy requires us to test this release in order to >verify it meets certain performance standards. In particular, there >have been concerns raised regarding the use of multi-threaded reads of >Oracle views and tables. We consequently sought to test and document >the effects of multi-threading on our Oracle servers. > >We would like to learn if users can share their knowledge in this >area. The SAS release 9 online documentation clearly states how >the number of threads to the Oracle server are determined. If >the Oracle table is partitioned, then the number of threads is >directly determined by the number of partitions in the table subject >to the limitations imposed by dbsliceparm. My assumption is that >the relationship is linear but I plan to test this supposition. If the >Oracle table on the other hand has no partitions, the MOD method >is used. > >However, what happens in the case of an Oracle view? Suppose a view >is defined by 13 underlying Oracle tables, but one uses SAS Access to >query the view. Would the MOD method be applied, or would the number >of threads be determined by the partitions in each underlying table? > >Any help on this question is greatly appreciated. Also, does anyone >care to share their experiences of using SAS Access to connect to >Oracle with release 9 and above (i.e. with multithreading). > >Peter Manolakos