Thread regarding SAS Institute layoffs

“ Viya does not need 100% SASV9 compatibility — only enough compatibility to provide a migration path to sell the product.”

Without copying the rest of your post, it’s pretty clear that Altair only provides a limited set of the total SAS9 stack’s extensive functionality and product set.

Doesn’t Viya provide quite a bit of compatibility with legacy/SAS9 then? Viya has a computer server that will run traditional SAS code apart for CAS or the need for CAS action syntax. It’s like that goes along way toward compatibility. Do we have some actual evidence that this is not enough for customers AND that Altair somehow “saves the day” with features they provide? if this is the case, it seems like the smart thing for SAS to do would be to close that gap pronto. If it’s a matter of providing an analyzer, OMG, the amount of compiler and language translation talent that remains at SAS, or that can be cobbled from open source).if this is the case, it seems like the smart thing for SAS to do would be to close that gap pronto. If it’s a matter of providing an analyzer, OMG, the amount of compiler and languages talent that remains at SAS (or that can be cobbled from open source) should be sufficient to address that concern.

Just not seeing how Altair provides a significantly superior solution such that they can compete on anything but a lower price and a highly targeted time to value proposition. Perhaps SAS is too busy with zone internal politics of market strategizing, Agile dashboards, and social media posts to address this head on. 🤦‍♀️

@1lzj+1uhfEpsR

Viya does not need 100% SASV9 compatibility — only enough compatibility to provide a migration path to sell the product.

Altair has that:

“Altair’s proven migration approach empowers big organizations to complete migrations in just three to six months. Smaller organizations can migrate even quicker.



First, we use Altair’s code analysis tool to scan through the syntax and language used within each of your existing programs and produces a compatibility report. Watch this video to see how the code analysis process works.



We then work with you to test and evaluate the function and performance of your library of existing SAS language programs using Altair software and complete the deployment on your chosen architecture. 



For most organizations, 95% or more of their SAS language programs will require no modifications and can begin running with Altair SLC immediately.“

https://altair.com/sas-language-capabilities

How do you define the scope of “compatibility” with version 9?

Does Altair run the the full range of SAS9 products? Do they have a metadata server, grid computing, an extensive fully built out set of java mid-tier services? Do they have IOM? How about an extensive range of database interfaces/engines? How well do they integrate with Microsoft office? How many different host platforms as Altair ported to and do they provide the level of integration that SAS9 has? Do they provide compatible visual application/analytics products? Can SAS Risk and Fraud run directly on Altair w/o having to purchase anything from SAS Institute? Can any/all of this inter-vendor or cross-release “compatibility” be achieved without considerable manual intervention by consultants, professional services or other SMEs? Or, does Altair mostly just ran “old-school” SAS — Data Step, Procs, Macro, etc.

These questions help illustrate just how extensive the problem set of “compatibility” between anything else (including Viya) and SAS9 is. It takes more than hiring smart people to do this. Most of the engineers who built CAS and other key services collectively had hundreds of years of R&D experience across the entire range of SAS releases going back to pre-MVA. If anyone had the correct skills to make it compatible with V9, they did. It just wasn’t the mandate. And that folks, circles right back to the issues of leadership, effective market research and vision.

Viya was not a bad idea. SAS has always invested in performance — because it’s good business. Historically, Fortune 500 clients and government agencies trusted SAS to perform well on large data sets. This was a point of pride, and a great source of revenue. HPA, LASR, and CAS were all attempts to continue this profitable business strategy.

The Manhattan Project took six years. The Apollo moon landing took eight years. These were hard problems — much harder than Viya. But they were solved — by expert engineers. The slow pace of Viya shows a lack of expertise. People were learning on the job.

This slow pace is not unusual for SAS. R&D had other projects that lacked expertise and took six, seven, or eight years. Some failed. Others succeeded, in the sense that they finally went production. But when you deliver software in six years, while your competitor delivers it in three, is that success?

At the time Viya started, SAS was already threatened by open source — mainly R. This threat increased when Spark was open sourced in 2010. But SAS was slow to recognize the danger.

As a result, SAS now has two product lines, each competing against open source. The market reasonably prefers cheaper alternatives.

Market research could have identified open source as a threat. But that wasn’t necessary: everyone in R&D already knew it was a threat. Maybe Spark was not an obvious threat in 2010, but by 2015 it should have been.

Market research could have also shown that SAS9 users want Viya to run their SAS9 jobs. But again, people in R&D already knew that. A few dared to speak up at the time, but were ignored. This was arrogance: people in high positions wanted to believe that Viya was so superior, the market would accept its lack of compatibility.

Some have also argued that SAS9 compatibility is a hard problem. Again, it’s not — not if you hire people with the right expertise. Altair has made their software SAS9-compatible. SAS Viya has not.



Over the long term, for-profit software can’t compete against open source. The only solution is to go where they aren’t, to produce value where they don’t. Tableau, QlikTech, and others made profits in this way. SAS has not.



That is the true price tag of Viya development: not the dollars expended on salaries, offices, and equipment — but the opportunity cost of what we could have built instead.

But was it true that Viya initially just a framework that customers had to upgrade to (and pay more for) in order to get new functionality in the future? That it offered nothing new initially?

We all know customers are very cautious in upgrading their systems and rightfully so. So was SAS struggling to convince them to upgrade especially since it gave nothing new?

The development for CAS commenced on the heels of VA/LASR (part of the V9 era) being the fastest product to generate $100M in new revenue. Before that, OS worked on HPA and was basically JGs “right hand man” for the modern incubation of highly parallelized analytics at SAS. OS was the chief architect of LASR and then on to evolution to CAS — the foundational computer server for Viya (remember, Viya is a branding term conceived just a few weeks prior to launch that SGF 2016). This history was roughly between 2006 and 2012 — Literally at the zenith of SAS’ success both in terms of advanced analytics market share. At point SAS apparently had 37% of that market — more than its next seven competitors combined.

For two years in a row (believe it was 2010 and 2011) SAS received the #1 spot in Fortune’s 100 top companies to work for in the USA. SAS was riding very tall! JG Was likely richer than ever. Hubris is one of the great curses of the human condition and honestly it was very easy for SAS to be pretty blind from the top down — blind to a brewing storm, the double headed dragon of Open Source Analytics (shortly thereafter also OSS Data Management) and “utility computing” as elevated and delivered by the major cloud vendors.

From 2012 to 2018, a swarm of optimism engulfed much of SAS as a company because, after all, the Shangri-La die had been cast long and our culture had many admirers, including Google, who had come to study it years back. JG was still building massively beautiful and luxurious new corporate buildings. SAS and its success was the talk of MBA programs at major business schools! Yet, back home the consequences of not developing ongoing deep cloud platform expertise across SAS R&D (including up-to-date understanding of not only cloud computing as it evolved, but even basic distributed Linux and the fast growth of open source tool sets and languages for overall development activities) would come home to roost, accelerating by the year. SAS was Firmly stuck in “Day 2” and almost no one apparently realize it.

As already explained in great detail on threads going back a year or more, this all set the stage for what ultimately became Viya. COVID-19 landed just as Viya making some headway at major customer sites followed by a mass exodus of 200+ of SAS’ best engineers and customer facing technologists, including OS, did not help matters.

Anyone with a deep understanding of SAS history Knows that there have been significant compatibility issues with moving to major new architectural releases in the past. Some of these took years to resolve — for example Version 5 to MVA/Version 6.

Does this mean that Viya Could be fabulously successful today had most of those employees remain, Covid not occurred, etc.?

Well, we will never know, but one thing is for certain, taken in total what is described above helps explain the perfect storm that resulted in the current state of Viya and really SAS as a company.

In a word: arrogance. The CEO had been so successful for so many years, in his mind, how could he be wrong about anything? Besides, he surrounded himself with sycophants who always told him how great he was, so again, how could he ever be wrong? Then he put all his faith in a forester who was both immature and arrogant and could not tolerate even the most minor perceived slight. Together, they made something the market did not want, incorrectly believing if they built it, the market would come. Meanwhile, they did their best to ki-l a set of products bringing in 3B a year. As others have said before me, business schools will study the rise and fall of SAS and its primary founder.