Difference B/n Proc sql joins and datastep joins

[deleted_user]

Hi All,
I think that Performance wise Proc sql joins take the higher priority when compare to datastep joins.Otherthan this there is any difference

Appriciate great help

Regards,
Govind

[Cynthia_sas]

Hi:
As this paper points out:
http://www2.sas.com/proceedings/forum2007/042-2007.pdf (page 4)

"In most cases, a DATA step MERGE statement generally outperforms an SQL outer join, even taking sort resources into account. One exception is a very sparse match join when you only want the observations with matching key values. Also, keep in mind that the SQL procedure and the DATA step MERGE do not provide the same results if you have a many-to-many match."

It really depends on your data and the type of join you want to do. Also, I don't know what you mean by "take the higher priority" -- you have to decide whether you want to optimize your programs and processes for clock time; for CPU resources; for I/O resources; for programmer resources; etc, etc. One technique may be more efficient on one set of data and join conditions, but less efficient on another set of data and join conditions.

You may need to benchmark different techniques to see which programming method works better for your data. Other resources for information are the SAS documentation and user group papers such as these:
http://www.ats.ucla.edu/stat/sas/library/nesug99/bt150.pdf
http://www2.sas.com/proceedings/sugi28/096-28.pdf
http://analytics.ncsu.edu/sesug/2007/IS02.pdf

cynthia

[deleted_user]

Hi Cynthia,

In DI STUDIO Generally which library we are store our taget datasets.

Appreciate Great Help

Regards,
Govind Message was edited by: govind_sas@yahoo.co.in

[deleted_user]

Hi Cynthia,

In DI STUDIO Generally which library we are store our taget datasets.Is There any specific library to store our taget datasets.

Appreciate Great Help

Regards,
Govind