SAS 9.4 architecture – building an installation from the ground up


Often, the SAS 9.4 administration environment architecture can seem confusing to new administrators. You may be faced with questions like: What is a tier? Why are there so many servers? What is the difference between distributed and non-distributed installations?

Understanding SAS 9.4 architecture is key to tackling the tasks and responsibilities that come with SAS administration and will help you know where to look to make changes or troubleshoot problems. One of the ways I have come to think about SAS 9.4 architecture is to think of it like building a house.

So, what is the first thing you need to build a house? Besides money and a Home Depot rewards credit card, land is the first thing you need to put the house on. For SAS administration the land is your infrastructure and hardware, and the house you want to build on that land is your SAS software. You, the admin, are the architect. Sometimes building a house can be simple, so only one architect is needed. Other times, for more complex buildings, an entire team of architects is needed to keep things running smoothly.

Once the architect decides how the house should look and function, and the plans are signed off, the foundation is laid. In our analogy, this foundation is the SAS metadata server – the rest of the installation sits on top of it.

Next come the walls and ceilings for either a single-story ranch house (a distributed SAS environment) or a multi-story house (a non-distributed SAS environment). Once the walls are painted, the plumbing installed, and the carpets laid, you have a house made up of different rooms. Each room has a task: a kitchen to make food, a child’s bedroom to sleep in, and a living room to relax and be with family. Each floor and each room serve the same purpose as a SAS server – each server is dedicated to a specific task and has a specific purpose.
Finally, all of the items in each room, such as the bed, toys, and kitchen utensils can be equated to a data source: like a SAS data set, data pulled in from Hadoop or an Excel spreadsheet. Knowing what is in each room helps you find objects by knowing where they should belong.

Once you move into a house, though, the work doesn’t stop there, and the same is true for a SAS installation. Just like the upkeep on a house (painting the exterior, fixing appliances when they break, etc.), SAS administration requires maintenance to keep everything running smoothly.

How this relates to SAS

To pull this analogy back to SAS, let us start with the different install flavors (single house versus townhouse, single story versus multiple stories). SAS can be installed either as a SAS Foundation install or as a metadata-managed install. A SAS Foundation install is the most basic (think Base SAS). A metadata-managed install is the SAS 9 Intelligence Platform, with many more features and functionality than Base SAS. With SAS Foundation, your users work on their personal machines or use Remote Desktop or Citrix. A SAS Foundation install does not involve a centrally metadata managed system, however in a metadata managed install, your users work on the dedicated SAS server. These two different SAS deployments can be installed on physical or virtual machines, and all SAS solution administration is based off of SAS 9.4 platform administration.

We hope you find this overview of SAS platform administration helpful. For more information check out this list of links to additional admin resources from my new book, SAS® Administration from the Ground Up: Running the SAS®9 Platform in a Metadata Server Environment.


About Author

Anja Fischer

Solutions Architect

Anja currently works as a Solutions Architect for Alteryx. Anja is a native of Heidelberg, Germany, where she started working for SAS in 1999. In her long tenure at SAS, Anja worked in technical support, education, pre-sales, and post-sales and has written presentations and customer documentation on what she is the most passionate about: SAS administration. She developed customized post-sales strategies for a successful software adaption for SAS platform and SAS® Viya® administration, SAS® Grid administration, and SAS® Visual Analytics administration. Anja has a bachelor of computer science and a bachelor of science for nursing. When she is not happily knee-deep in administering, she loves to go on adventures with her dogs, Benji, Gwendolyn and Carlie.


  1. Hi,
    I just ventured into the SAS subject by accident and I would like to know some implementation/deployment details regarding SAS9.4. As I have to write an architecture document to implement a SAS environment....
    I understood that SAS has for major components, ie. Mid-Server, Meta data server and Compute (grid) server.
    * Is it possible to install ALL of this server-roles onto a SINGLE Linux Virtual Machine?
    * Is so, what are the resource pitfalls I have to think about, in terms of CPU, MEM, HD...?
    * Can this setup be loadbalanced (let say Level4, Round-Robin) to be more High-Available?



  2. Hello Anja,

    This is a great book for novice SAS admins like me. The only pain point is, I brought paper copy, and having hard time finding that pdf doc on your author page, that has the hyper-links you referenced in your book. Also, most of the 'SAS Documentation' links are NOT working. Would it be possible to get a latest copy of the 'Hyper-Link' pdf document?

    Thank you

  3. Dear Anja,

    I didnt get the below sentence ,

    "Next come the walls and ceilings for either a single-story ranch house (a distributed SAS environment) or a multi-story house (a non-distributed SAS environment)." ,

    Can you please clarify,


    • Chris Hemedinger
      Chris Hemedinger on

      Hi Subin, I think Anja was just using a metaphor for how you might deploy SAS. Are you creating a "personal" environment for each user, with separate data and structure, or are you creating a multiuser environment for collaboration where all users must agree on a set of rules for working?

Leave A Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Back to Top