topic Re: sas Basic question for optimization in SAS Programming

sas Basic question for optimization

Maaz — Thu, 02 Nov 2017 16:26:35 GMT

Can converting a proc sql code in sas make some difference .My concern is time

as I have a long code in proc sql it takes more than half an hour to run.

if not how can I save my time?

Re: sas Basic question for optimization

PeterClemmensen — Thu, 02 Nov 2017 16:27:40 GMT

What does your code look like?

Re: sas Basic question for optimization

RW9 — Thu, 02 Nov 2017 16:51:27 GMT

Without seeing any code its hard to say. Certainly in some circumstances SQL can be more resource intensive than a datastep or procedure variety. There are also other options however which may speed things up, for instance filtering data earlier so you only take in the rows you need, indexing data etc. Its quite hard to say though without any information.

Re: sas Basic question for optimization

hollandnumerics — Thu, 02 Nov 2017 17:24:44 GMT

Maaz,
There is a simple correlation you should understand when posting a code-related question:

The probability of an answer is proportional to the amount of SAS code, log and output you post with your question!

Re: sas Basic question for optimization

PGStats — Thu, 02 Nov 2017 18:20:46 GMT

Try the advice given in

http://support.sas.com/documentation/cdl/en/sqlproc/63043/HTML/default/viewer.htm#n1wue1kaft1rlsn1v0y5m569besp.htm

before considering converting to a SAS program.

Re: sas Basic question for optimization

Maaz — Fri, 03 Nov 2017 09:23:20 GMT

this is a small sample of code like this I have to run many more code which are more lenthy than this . the data is in Teradata and creating the table in tera data as well

Re: sas Basic question for optimization

hollandnumerics — Fri, 03 Nov 2017 16:05:20 GMT

Maaz,

As the whole program is SQL, why can't you recode it to run completely inside Teradata? Teradata is designed to run SQL efficiently, so why not let it get on with it.

Currently you are doing the following using PROC SQL:

extracting 8 (potentially vast?) tables into SAS
merging 4 tables each into 2 other temporary SAS tables
joining the 2 temporary SAS tables with UNION into another temporary SAS table
storing the final table back into Teradata

Doing the same steps all in Teradata using Pass-Thru will take SAS out the equation altogether.