SAS Windows 7 VM on Parallels: available memory not used

[kubi8]

I'm running SAS 9.4 on Windows 7 64 as a VM on Paralles on a MacPro, 12 cores, 64GB RAM. I've assigned 8 cores and 48GB to the windows 7 VM.

I'm using a tool for linking data sets, which uses SAS as data base. The data are stored on an external 1TB SSD, connected with thunderbolt2 to the MacPro. The tool generates a new table using SQL scripts on two tables and starts 4 sas.exe processes, which only use about 100MB of RAM each. The data sets used are around 40-60MB with 400'000 records each. The tool splits the process into chunks and creates temporary sas-files of 300-500MB, one per running process. These files could easily be processed in RAM. Instead, SAS writes the files to the hard drive. I found a lot of I/O in the \temp\SAS Temporary Files\ folders.

How can I force SAS to use the existing RAM?

I've been playing around with

- memsize (between 4G and 40G)

- memmaxsz 4G

- sortsize 2G-12G

- buffsize 8k-32k

- bufno 100 - 500

Nothing has improved the over all performance and no setting motivated SAS to use the available RAM.

Any ideas how I could increase performance by using the existing RAM?

Thanks for any help.

[JuanS_OCS]

Hi,

a pity that you donlt use Unix-based OS, because you can easily define a RAM file system.

Anyway, SAS still provides you great functionality to improve your performance a lot. You may want to read the following links:

https://support.sas.com/documentation/cdl/en/hostwin/63285/HTML/default/viewer.htm#winperform.htm

http://support.sas.com/documentation/cdl/en/hostwin/67962/HTML/default/viewer.htm#p041tbb02reefnn1jm...

You can focus on using MEMLIB and MEMCACHE. For this, I love this paper: "Why Aren’t You Using MEMLIB?"

http://support.sas.com/resources/papers/proceedings10/070-2010.pdf

https://support.sas.com/documentation/cdl/en/hostwin/63285/HTML/default/viewer.htm#win-sysop-memlib....

Further more, if you would like to go for more advanced features, you can give a look to the LASR server licenses.

Let us know how it goes.

Best regards,

Juan