Thread regarding SAS Institute layoffs

“…the verticals that worked, Risk and Fraud.

Why did those two verticals work, when every other one failed?”

I don’t know, but I wouldn’t be surprised to discover that Risk and Fraud vertical solutions carried greater sales incentives than other solutions. SAS salespeople have a long and distinguished history of going where the incentives tell them to go. And I can’t blame them for doing so.

@3zpm+1rgLAksJ

In V7, building verticals was a good plan. It should have worked.

Now @2qge+1rgLAksJ suggests modernizing the architecture and focusing on the verticals that worked, Risk and Fraud.

Why did those two verticals work, when every other one failed?

@2wwr+1rgLAksJ
"In the long term, all the problems must be solved. Pivoting from tools to vertical solutions vendor could be a good long-term plan".

I think it was back in the v7 timeframe when there was a lot of talk about SAS transforming itself from being a "tools provider" to providing "vertical solutions". It was at this time when the collection of (SAS/AF-based?) applications were referred to as "vertical solutions".

Long term plan, indeed ;-)

@2blj+1rgLAksJ

Not yes men — just a few who view SingleStore as a solution to a problem in SAS’ product line. It’s not the only possible solution, but as described previously, one cheap enough to acquire.

As you say, Viya upgrade from 3.5 is another large problem unsolved. Even worse is Viya compatibility with V9. Those problems cannot be solved by purchase, but only by hard work internally.

In the short term, reselling other people’s software solves some problems. In the long term, all the problems must be solved. Pivoting from tools to vertical solutions vendor could be a good long-term plan — perhaps funded by axing less profitable products.

But is there a long-term plan at SAS? They seem focused on preparing to IPO.

… analytics platform leader

@2blj+1rgLAksJ

Just spent 15 minutes crafting a detailed response, only to realize it’s “kryptonite” and could potentially out good people who had/have nothing but excellent intentions for both companies. In summary, the reason this topic is important on a form about layoffs is because it addresses how a company, SAS that so many of us love and invested decades of our life in might have an ongoing pathway as an analytics platform later. From where I sit that’s pretty damn important in minimizing layoffs.

Lots of yes men SingleStore fans on a blog about layoffs at SAS. They are related. BH spent 2 years and $1B of JG's money writing software for other companies like SingleStore and no time making an upgrade path for SAS's own Viya 3.5. How much money did SAS make as a reseller of SingleStore last year? AEs can't resell SingleStore at a SAS price. No new SAS software, no new money, no jobs. But BH can lecture fired testers on how they have only themselves to blame.

@2fkz+1rgLAksJ

Respectfully, you’re missing the point regarding SAS+SingleStore (S2). I’m pretty certain S2 alone does not compete directly with either Databricks/Spark or Snowflake. In fact, S2 interfaces very well with these products and provides a considerably higher scalability and performance for storage+retrieval. SAS would not be deepening their partnership nor acquiring S2 Simply to try to create head-to-head competition with Spark or Snowflake. That SingleStore currently has, by comparison, a relatively low per-IPO valuation would be a significant advantage in attempting to acquire them.

Instead SAS would acquire SingleStore to gain an ultra-modern high performance distributed, cloud native database (something SAS has thus far not been able to build, having failed at lesser attempts “more than twice” in the last 40 years) with a multiplicity of innovative platform and data management features. This would enable SAS to move away from being an analytical tools vendor, (a space that nearly everyone here agrees SAS V9 cannot dominate going forward and will eventually cease to play in at all) to providing highly specialized (e.g Risk, Fraud, new Verticals, etc.) products at enterprise scale atop a streamlined, elegant data platform instead of the current Viya hodgepodge. Providing a data platform that can be a true system of record makes your products much stickier.

Over time, SAS would retain its identity as a data driven analytics company (tho diminishing as a tools vendor) and could leverage innovation based in a refactored/ evolved Viya/CAS. Is this an easy thing to do, from a culture and engineering perspective? NO, Not at all … But it sure seems like a better long-term bet they’re becoming a reseller/systems integrator and continuing to slowly lose identity (as OP suggests) as a provider of innovative homegrown products.

What have I missed?

All this talk of Singlestore and that partnering or acquiring them might help to safe SAS from it's tragic demise, strikes me as a little ridiculous.

Singlestore's pre-IPO valuation is only just over $1B. Compare that with the pre-IPO valuation of Databricks at $43 billion or the current market cap of Snowflake at $77 billion. I don't think Singlestore has a hope in he-l of catching up at this point.

Surely JG wouldn't be foolish enough to blow over $1B on acquiring a product so far behind it's obvious competitors?

Again: given the directive to become “IPO-ready” in the next year, what choice do they have?

Nowadays, you can’t IPO without AI, and SAS can’t build that quickly. Even Micro$oft can’t — that’s why they paid $13B for OpenAI.

Maybe SAS could build the functionality of SingleStore — but not quickly. SAS is better off partnering with SingleStore, or acquiring them.

As for R and Python, those ships sailed long ago. All SAS can do now is integrate with them.

Finally, how much talent remains to build such things? During the pandemic stimulus, other companies paid well for talent, and remote work made changing jobs easy. SAS dropped the ball, failed to pay market rates, and hundreds of SAS engineers left for greener pastures.

There’s still talent remaining at SAS. Properly led, they can address some of the problems. But there are multiple large technical holes in SAS’ product line, and not enough people to fix them all.

When you’ve let other companies get ahead of you, and you’re not willing to pay for top talent to catch up, and you still want to IPO — there’s only one choice. When you can’t build the software, you must buy it.

SAS customers who spent $300M on Viya 3.5 will see only Viya 4. They will find out there is NO migration path to upgrade. BH could have made a swarm of all of the testers and DevOPS he fired and made them work on that but under BH DevOPS means Develop Other People's Software.

AEs with an upgrade target will quit because they cannot sell an Openai chatbot in an overpriced Azure cloud. The AEs will take the customer connections they make in Las Vegas with them when they leave to work for partners. Even the developers know they can't sell a python editor just because it has a SAS logo. They are quitting right after bonus.

@tie+1rgLAksJ

SingleStore (S2) has deep integration with CAS in Viya. CAS actions can run directly against S2 database tables without having to convert to HDAT … the native CAS in-memory and persistence format. Thus data does not have to be transformed/replicated angain and CAS can interoperate concurrently while S2 does a multiplicity of DBMS tasks. This combination can be a game changer if architected and implemented correctly for particular SAS solutions and customer engagements.

https://www.sas.com/en_us/software/viya/singlestore.html
https://www.singlestore.com/partners/sas-viya/

WRT SingleStore, the following article might help address concerns about the future of data management related R&D at SAS. Clearly S2 provides many of the connectors that SAS traditionally did. An architecture that uses CAS with SingleStore as the underlying data management and storage platform would not likely need SAS data connectors because they wouldn’t be ingesting those data sources directly into nor storing data in SAS.

https://blocksandfiles.com/2022/08/22/sas-viya-singlestore/

However, as previously mentioned on another thread, for the past 10+ years SAS has only minimally/sporadically invested in the kind of industrial strength “storage layer up” data management infrastructure that is native to advanced modern DBMS platforms like SingleStore. SAS’ R&D data management division seems to be focused more on UI/UX based tools to integrate data from a variety of sources. Not sure how much of that would be needed if S2 is chosen as the underlined data platform.

S2 also does many of the things that the SAS ESP engine does. JG should’ve considered either making a bid to purchase S2, or heavily investing in its VC. Done right it could fix a lot of what ails SAS as V9 sunsets and Viya attempts to compete in a progressively more uncertain future.

Thoughts and prayers!

From leader in analytics to reseller of other people's software? Oh come on...how has the mighty fallen so low. Let's pray for SAS.

@ztl+1rgLAksJ

SAS has been interested in Singlestore for many years. Have they done anything of substance to integrate? If so, wonder if there will be more paring down of the SAS Data Management team.

I’m no fan of BH, but in this matter I don’t see that he had much choice.

He was directed to make the company “IPO-ready” -- originally, in this calendar year.

Three years was not enough time to solve the technical problems in SAS’ product line. Some of these problems are huge, due to architectural decisions not easily reversed.

When you’re directed to make a quick fix, and you don’t have time to build it, then you must buy it.

HEAD ON APPLY DIRECTLY TO THE FOREHEAD!!

Ummm SAS's reputation is already ruined.

@lqu+1rgLAksJ

Probably because it’s SAS last hope to recapture any semblance platform relevance going forward. SingleStore addresses most of the major data management deficiencies in Viya and provides an excellent foundation for moving, stressful verticals, like risk and fraud into the future. SingleStore employees the kind of engineers that SAS should’ve been hiring over the last 15 years.

You’ve confused Snowflake with SingleStore. SingleStore is the favorite of BH. No idea why.

Snowflake? Is that one of BH's direct reports? (In my experience it was.)

Man dat cold … y’all need this medicine:

https://youtu.be/i9a5wfB4WoU?si=Nitbpb90zjCx1dsK