Is it safe to run multiple SAS sessions in Window?

[somebody]

I have a large datasets so each time I run some code, it takes a while. So I open another SAS window(Start--> SAS--> SAS 9.4.exe) to do something else. Is this safe to do? would this affect my firstly opened SAS window? So far, both windows run fine, except for the second window, I got the following error when it is open.

WARNING: Unable to copy SASUSER registry to WORK registry. Because of this,
WARNING: you will not see registry customizations during this session.

Is this ok ? 

[Jagadishkatam]

Yes it is absolutely fine

[Kurt_Bremser]

This registry catalog is used to store your interface settings (among other things) and opened for update by the first SAS session, so it is locked for every other session you start while the first is active.

The WARNING tells you that any non-default settings you made will not be available, any changes made in the first session will not be recognized by the others, and any changes made in the additional sessions will be lost as soon as they terminate.

For batch executions, you can safely ignore this.

[Shmuel]

One point to consider:

As long each session updates different data-sets or create different external files, 

it will run fine without interfering each other.

In case you have common files, sometimes library.format may cause issues and if you try

update a same data-set you may loose updates done by one of the sessions, as only last closed

will be kept on.

[FreelanceReinh]

Hi @somebody,

To avoid this warning you can set the RSASUSER system option, which I normally use on my computer and only rarely switch to NORSASUSER (see this older post).

Using two or more SAS sessions in parallel can save a lot of run time (depending on the tasks) if the sessions access different datasets and use different threads/cores of a multi-core processor (check the CPU usage in Windows Task Manager).

Also, I like having a "test session" available (to test small code snippets) in parallel to the SAS session in which I develop production code. Thus I keep all the "dirty" testing code, log, output and WORK datasets out of the "clean" production environment.

[ballardw]

If the code for your "large datasets" doesn't need to be manually changed you might consider submitting it for a batch job. Right click on the program file in Windows Explorer and you should get a context sensitive menu to submit the code as a batch job.