Hi,
I have 50 million obs output data set to be written back to Oracle table.
Using Table Loader transformation and can't ignore the fact that it is incredibly slow.
Table has 5 attributes, indexes dropped.
Using default proc append.
If someone can suggest some tuning tips for Table Loader, I would really apretiate it.
Thanks!
First thing I'd check: Option settings for INSERTBUFF and DBCOMMIT
If changing these options to something better than the installation default doesn't help: Consider using bulk load (also managed via options).
The following discussion might also be useful for you (as well as the link to a tuning paper in there).
https://communities.sas.com/t5/SAS-Data-Management/The-SQL-tuning-checklist/m-p/438922#M13536
First thing I'd check: Option settings for INSERTBUFF and DBCOMMIT
If changing these options to something better than the installation default doesn't help: Consider using bulk load (also managed via options).
The following discussion might also be useful for you (as well as the link to a tuning paper in there).
https://communities.sas.com/t5/SAS-Data-Management/The-SQL-tuning-checklist/m-p/438922#M13536
Adding INSERTBUFF=16000 helped a lot.
I also set BUFNO to 50.
That reduced load time from 40min to 1min25sec.
Thanks!
Best regards,
If there isn't any special reason to do otherwise then I normally set buffno to zero. This way there's only a single commit once all the data got loaded.
Thanks Patrick! Will set it to MIN (default) and re-run the load to see if the load time is not significally worse.
Not to min but literally to zero: dbcommit=0
Besides of performance this also ensures that you'll never end up with a partially loaded table but it's either all or nothing - which in my opinion makes it easier to fix things and rerun as you don't have to care about backing out partially loaded data.
SAS Innovate 2025 is scheduled for May 6-9 in Orlando, FL. Sign up to be first to learn about the agenda and registration!
Need to connect to databases in SAS Viya? SAS’ David Ghan shows you two methods – via SAS/ACCESS LIBNAME and SAS Data Connector SASLIBS – in this video.
Find more tutorials on the SAS Users YouTube channel.