SAS Viya migration: SAS 9.4 has an end date, so where do your programs go?
By Xavier Franco, CEO de Elternativa · Grupo Zemsania
· 6 min read
Written with the help of AI and reviewed by Xavier Franco

In this article
Few tools have as much history inside a company as SAS. Programs written fifteen or twenty years ago that still calculate provisions, risks, tariffs or the report that reaches management every month. They work, and that is precisely why nobody wants to touch them. But SAS 9.4 has dates, and it is worth knowing them before they decide for you.
The dates, as SAS publishes them
I will start with what the vendor says on its support page:
- SAS 9.4 M8 was released in January 2023 and its standard support lasts until 1 February 2028.
- SAS 9.4 M9 was released in June 2025 and its standard support lasts until 1 July 2030.
- M7 moved to limited support on 1 September 2025, and M6 and earlier on 1 February 2025.
In other words: if you are on M7 or earlier, you are already in limited support. If you are on M8, you have until 2028. And if you upgrade to M9, you gain time until 2030.
In practice: ask your team which version of SAS 9.4 you have installed. It is the first date in your plan.
The vendor’s proposal: SAS Viya
Naturally, SAS proposes its current platform, SAS Viya. It is a legitimate option, and in some cases it may be the best one. But it is not the only one, and it is worth comparing it calmly before you sign.
My view is this: most of the SAS programs in a company are, at heart, reading data, transformations, SQL and calculation. A modern data platform does that very well today, without a specific licence for each program or each user. SAS licences are priced per processor or per user, and they weigh on you every year.
In practice: ask for the annual cost of your SAS licences and put it next to the number of programs that really run. That ratio already says a lot.
What is really in your SAS
When you take stock of a SAS estate that has years behind it, the pattern repeats:
- programs that have not run for months or years, but that nobody dares to delete;
- duplicated logic: the same calculation copied into several programs, with small differences that nobody remembers;
- knowledge held by few people, the ones who know which macro does what;
- results that feed reports and models that someone looks at every month.
The first thing is to separate what is used from what is not. What does not run is not migrated: it is switched off. It is the quickest saving and the one with the least risk.
Three destinations for your programs
For what is used, there are different paths depending on what each program does:
- SQL on your data platform. Databricks, Snowflake or Microsoft Fabric. Most SAS data steps and procedures translate into SQL and native processes. The logic lives where the data already is.
- Your own Python. For statistical calculation and models, Python is today a standard with a huge ecosystem, and no licence.
- Reports to your BI tool or to a custom dashboard. Reports and Visual Analytics move to Power BI, Tableau or Qlik Sense, or to your own code with no per-user licence.
What stays is your data and the results of your processes and models. They are the reference against which everything new is compared.
Migrating without losing a figure
Here is the real fear, and I understand it. A SAS program that calculates a provision or a tariff cannot give one cent of difference after migrating.
That is why the method matters more than the destination:
- Read-only inventory of programs, macros, libraries and schedules: what runs and what each one feeds.
- Documentation extracted from the code, so that the rule is written down before it is rewritten.
- Reconstruction with orchestrated AI, validated by your team.
- Parallel run and number-by-number comparison with today’s results. Each program is cut over when it reconciles.
In practice: pick the program that worries your team most, not the easiest one. If that one comes out right compared number by number, the rest is method.
The cost of not deciding
There is an understandable temptation: upgrade to M9 and forget about it until 2030. It is a legitimate option and it buys you time. But it is worth seeing it for what it is: a postponement, not a decision.
Upgrading SAS is also a project, with its tests and its risk. And meanwhile, everything stays the same: licences per processor or per user, duplicated programs, knowledge held by few people and results that can only be reproduced inside SAS. In 2030 the question will be the same, with four more years of programs on top.
That is why my advice is to separate the two things. If you need time, upgrade. But start the inventory now and switch off what is not used: that reduces the size of the problem whatever you decide afterwards. And migrate a first block compared number by number to have real data on cost and timing.
That way, when the time to renew comes, you do not negotiate blind: you negotiate knowing how many programs you really have, how much it costs to move them and what saving there is on the other side.
In practice: before approving a SAS upgrade, ask in parallel for the inventory of programs in use. With that figure, the renewal conversation changes.
Where to start
- Confirm your version of SAS 9.4 and its real support date.
- Take inventory of programs and real runs.
- Switch off what is not used.
- Decide the destination for each block: SQL, Python or BI.
- Migrate a first block with measured parity and compare cost and timing before going on.
If you are on M8, you have until February 2028. It seems like a lot, but a serious inventory and a first compared migration take time. Better to arrive with the work done than with a renewal on the table.
How we help you
In Dashboard and BI migration we move SAS programs and processes, Viya models and flows, and their reports to SQL and processes on your data platform, to your own Python or to the BI tool of your choice, with no SAS licence. The same figures, compared number by number.
We do it with the Datagentis migration module, which connects read-only, takes inventory of what you have, flags what is not used and generates the documentation and the plan before anything is moved.
How many of your SAS programs have run this year?
In a demo we show you the program inventory, what is really used and how the results are compared number by number after migrating them.


