I have question - i am not sure even this is feasible,
if purpose of server is to execute just batch jobs using IBM Tivoli, do we require metadata server install/configured with sasapp server OR we can be running without SAS Metadata Server?
You can use /sas/SASFoundation/9.4/sas to run batch jobs, but you must provide the whole environment (settings, libraries) through config and autoexec files.
AFAIK, if libraries are registered in metadata, but not metadata bound, they can still be used with conventional libname statements.
May I ask why?
Do you plan to have a SAS installation without a metadata server?
And what would you gain?
We are using Tivoli at our site. We're deploying SAS job flows in SMC, and let Tivoli call the as Tivoli Jobs.
Works fine, except that SMC create new script names with each deployment (in Linus/UNIX only, not in Windows).
Connected to the discussion on libraries, you may also have tied resources to metadata authorizations, like access to a database account.
Thank you so much all for input.
As I said earlier, just thinking if there is any way to use just required SAS components for batch jobs without metadata server, have less number of machine running. Just trying to understand if this is something feasible.
First of all, the metadata server does not have to be on a separate server. Our data warehouse runs on one server instance, Compute, Midtier and Metadata all in one.
You can run SAS jobs without access to the metadata server, but as soon as you need to use an object bound to a metadata entry, your job will fail.
And you will be forced to keep the non-metadata setup synchronized with your metadata-driven setup (e.g. if the physical directory name of a library changes in metadata, you need to repeat the same in your batch autoexec.sas).
This is all avoided by using the BatchServer/sasbatch.sh within your server context.
The SAS Users Group for Administrators (SUGA) is open to all SAS administrators and architects who install, update, manage or maintain a SAS deployment.