Connect to SAS Grid with SAS ODBC Driver

[MTSlaughter]

Is it possible to query SAS datasets on a SAS Grid with the SAS ODBC Driver? I only have experience using the SAS ODBC Driver with PC SAS. If it is possible I would appreciate any configuration or connection tips.

Thanks!

[gwootton · Accepted solution]

It sounds like you're talking about accessing SAS data sets that have been stored on the compute tier of a grid deployment remotely from that environment and SAS, using the SAS ODBC driver. To do that, I think you would need a server to be running on that deployment to answer that ODBC connection (i.e. a SAS/SHARE server that has loaded a library with the datasets you want to read).
So I think you'd need a SAS/SHARE server configured on your grid deployment to be able to access it using the SAS ODBC driver.

Documentation on SAS/SHARE:
https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/shrref/titlepage.htm
Documentation on the SAS ODBC Driver:
https://go.documentation.sas.com/doc/en/odbcdref/9.4/n0maisizj6qtffn1ps3ez8g09ge7.htm

[SASKiwi]

This is a very uncommon way of connecting to a SAS Grid and I would be surprised if you have SAS/Share available to you. Another option would be SAS/Connect which needs to be installed and licensed on both PC SAS and SAS Grid. Check with your SAS administrator to see if this is an option.

[MTSlaughter]

Thanks, Kiwi. The motivation here is to use the SAS ODBC Driver to serve SAS data to an external application via ODBC. I don't think SAS/Connect is intended to allow connections from external applications treating SAS as a DBMS; am I wrong?

[SASKiwi]

You are correct. What's the external program? I've always found it easier to load SAS data into a relational database for accessing with external programs. That works well as long as you don't need real time data access.

[MTSlaughter]

R, for instance. There are R packages for reading SAS data, of course, but they generally require the entire dataset to be read into memory. Using the ODBC driver makes it easy to pass a complex query to SAS and bring back just the subset we want.

You're correct that loading the data to a different DBMS would obviate the issue and we may go that route. For now, I'm trying to build off of the existing processes without requiring asking anyone to change their architecture.

[SASKiwi]

Going the SAS/SHARE route looks like the only option if you don't want to change your architecture. That could involve additional licensing costs and installation and configuration requirements.