SAS mirroring - best practice

[woo]

Hello, can someone please help to prov8de high level suggestion or find some best practice paper on below topics?
We will be using SAS94 M6 on AIX 7.x. (non-grid). We have two data center and both are prod all in one server + they are going to be replicas. Data will be sync'd everyday to keep it mirrored.
A) what is the best way to get users sync'd across two data center so that we don't have to manage users separately, we will split users in half on those 2 servers.
B) what is the best way to have batch jobs running?
C) what is the best approach for load balancing in this scenario?

Thank you

[AhmedAl_Attar]

Hi,

First, I would like to suggest, you skip SAS 9.4 M6 and move to SAS 9.4 M7 instead. M7 contains the latest fixes and feature enhancements

Second, you make sure you have a plan in place for implementing/applying SAS Hotfixes on a periodic bases.

Now, it would help us better if you could specify/elaborate, Which SAS 9.4 Mx products/solutions you are planning to install/deploy, as they could have their own specific best practices/suggestions

Hope this helps,

Ahmed

[woo]

Thanks. We will be using SAS OA, SAS/ACCESS to DB22, Teradata, SAS SPDS and SAS PC file server on AIX 7.2.

[AhmedAl_Attar]

@woo 

For

• SPDS: I would highly recommend you contact your SAS account representative and ask for Sizing services. SPDS can be really performant if it was properly configured and the underlying file-systems properly provisioned
• OA (Office Analytics): You'll need to carefully read this paper: Getting Started with Designing and Implementing a SAS® 9.4 Metadata and File System Security Design 
• SAS/ACCESS Interface to DB2, Interface to Teradata: Familiarize yourself with the Online Docs for SAS/ACCESS® 9.4 for Relational Databases: Reference 
• AIX 7.2: It's very important to get your Unix server configured and sized properly. Check this paper https://support.sas.com/resources/papers/proceedings16/SAS6761-2016.pdf and all the references mentioned inside it

[Sajid01]

A. Consider metadata based access control for your situation. Multiple users may be needed for each level of access which can be mapped to an authdomain and a library. 

B. Batch jobs use the authdomain option in their connection statement. Production jobs typically run with generic ID.
C. Load balancing best left to DBA's. SAS points to the load balancer.

[Sajid01]

Hello @woo 

C) what is the best approach for load balancing in this scenario?

If SAS programming jobs are being run, SAS grid is the appropriate choice.
You don't have the grid, so the option would be to divide the users in to two groups.
Each group having access to one environment only .

B) what is the best way to have batch jobs running?

Depends upon your environment. Typically at many places SAS jobs are scheduled using external job schedulers such as Autosys/Control-M etc.
If that is feasible the job can be scheduled using any of the schedulers. Again you have the choice of distributing them between servers, with or without reference to SAS Metadata Server.

A) what is the best way to get users sync'd across two data center so that we don't have to manage users separately, we will split users in half on those 2 servers.
Depends upon your environment. If AD is being used, then they can be synced from it. In that case divide users into groups. One AD group for environment and another AD for another environment.
In principle it should be possible to export and import metadata from a SAS dataset. SAS Tech Support may be able to speak on this.