topic Re: SAS 9.4 M6: Migration from Unix to Linux in Administration and Deployment

SAS 9.4 M6: Migration from Unix to Linux

MariaD — Wed, 22 Apr 2020 14:13:21 GMT

Hi folks,

We are changing our SAS installation from Unix (HP-UX server) to Linux (RedHat) server. One quickly question: is it necessary to make any transformation on our .sas7bdat files to works in Linux server?

Regards,

Re: SAS 9.4 M6: Migration from Unix to Linux

MargaretC — Wed, 22 Apr 2020 14:19:36 GMT

You are not required to migrate your data from HP-UX format to Linux format, it can be access via the SAS Cross Environment Data Access method. This feature is documented here: https://go.documentation.sas.com/?docsetId=movefile&docsetTarget=p0mvwvg25ceml5n1c2digfpk6vkh.htm&docsetVersion=9.4&locale=en
As you can see from the documentation, the SAS data sets can be accessed via CEDA in a read only method.
Catalogs and indices cannot be accessed via CEDA. So if you randomly access a SAS data set via an index, you will not be able to do that so you will be sequentially accessing the data in file, this can cause SAS jobs to run much longer. Also, user defined formats, which are stored in SAS catalogs, are not available to you until you convert the SAS data file to Linux format and recreate the index.
Hope this helps.

Re: SAS 9.4 M6: Migration from Unix to Linux

MariaD — Wed, 22 Apr 2020 14:34:56 GMT

Thanks!

Re: SAS 9.4 M6: Migration from Unix to Linux

MariaD — Tue, 05 May 2020 18:20:49 GMT

Hi @MargaretC ,

So, as per your information, we'll need to recreate all indexes for our SAS tables. If we use a REPAIR sentence for PROC DATASETS on a Linux server, the index will be recreated? If not, there is any way to read (before migrate the data) the index definition to create a process to recreate them on Linux server?

I know the PROC CONTENTS show the index detail but in a report. There is any way to send this information to a table to create an automatic process?

Regards,

Re: SAS 9.4 M6: Migration from Unix to Linux

SASKiwi — Tue, 05 May 2020 20:01:25 GMT

@MariaD - If you use SAS/CONNECT and PROC UPLOAD to migrate your data you can recreate your indexes automatically: https://documentation.sas.com/?docsetId=connref&docsetTarget=n12q8v8qgfbtw8n1kqlzm78mvaz8.htm&docsetVersion=9.4&locale=en#p1lfr770s8voxbn1kvx2mnr8m6jq

Re: SAS 9.4 M6: Migration from Unix to Linux

MariaD — Tue, 05 May 2020 20:08:54 GMT

Thanks @SASKiwi. No, we'll change the storage from one server to other.

Regards,

Re: SAS 9.4 M6: Migration from Unix to Linux

SASKiwi — Tue, 05 May 2020 20:13:30 GMT

@MariaD - That is exactly what SAS/CONNECT can do, and it handles the automatic conversion of SAS files between operating systems, including SAS datasets and catalogs.

Re: SAS 9.4 M6: Migration from Unix to Linux

maggiem_sas — Tue, 05 May 2020 23:19:22 GMT

PROC MIGRATE can be used with SAS/CONNECT from one file system to another file system to avoid CEDA limitations. Indexes and integrity constraints are rebuilt on the member in the target library.

Here is the PROC MIGRATE documentation but a better example that uses a CONNECT spawner is below (the doc will be updated at next release). Run this in the target environment, using your values for the italic values in the example. Note that PROC MIGRATE can only migrate one library (directory) at a time. The CONNECT server must match the data representation of the source library, which in your case would be HP-UX.

options comamid=tcp;
%let myserver=host.name.com;
signon myserver.__1234 user=userid password='mypw'; 

libname source 'source-library-pathname' server=myserver.__1234;
libname target 'target-library-pathname';

proc migrate in=source out=target <options>;
run;
signoff myserver.__1234;