Thoughts? Was highlighted more deeply at Innovate today.
Combines SAS9 and Viya procs into a single, non-CAS runtime.
Basically what we’ve all been asking for after all these years.
17 replies (most recent on top)
as others mentioned, it only manages the decline - but that is still a necessary thing to do. the users of 9 are retiring, and meanwhile the decision makers are phasing out older generations of on-premise software. newer generations of analysts and users are using open source and cloud native solutions. is it possible to bridge those gaps? very difficult challenges to be sure.
I don't think Viya / SAS9 compatibility is the panacea to all of SAS's problems. It may convince some SAS customers to shell out a few more $ to run their existing code faster, but it does nothing to address the bigger problem of competing against open source, and nothing to address the usability issues that plague all the SAS tools going back decades.
Let’s consider strategy. I believe 100% compatibility can’t be done, because so many of the older folks who understand the problem have left.
But surely 80% compatibility is achievable. So I’d aim for that, and treat the remaining 20% as bug fixes and consulting opportunities.
Will that work? or is it too late?
...and I also did not mean to imply that all managers are part of the problem. I know several still at SAS who are bright, capable and considerate human beings I'd be happy to work for.
They aren't high up enough to make the big decisions, though.
@1viq+1s6O841W probably over-generalized like I did, when I said “the skills and culture needed to start new revenue streams aren’t there.”
I did not mean to imply that the rank and file were the problem. I know many engineers and sales and other folks who remain at SAS. They have enough skills to build and sell anything.
But management has demonstrated that they lack the skills to start new revenue streams. They also failed to maintain compatibility.
They’ve identified compatibility as a goal now -- an achievable goal, if not too late.
“ Tolerance for not following marching orders(Viya compatibility with V9) is THE BIGGEST REASON why SAS is where it is.”
On whose part? I hope you mean management because they were the ones that could never decide whether it was a true priority. Start and stop, high priority and no priority is what I saw. Actually, I recall no specific point in time when “Viya compatibility with V9” was a firm priority and it certainly wasn’t the lower-echelons calling the shots.
"The skills and culture needed to start new revenue streams aren’t there. But that doesn’t mean imminent doom, rather a managed decline."
Nailed it.
Tolerance for not following marching orders(Viya compatibility with V9) is THE BIGGEST REASON why SAS is where it is.
Selling a declining company is always harder than selling a prospering company. It su-ks that it has played out that way.
Where would SAS be today if the Big German had followed orders? Specifically, the order to make Viya compatible with V9.
It is VERY doubtful that SAS would be any worse off now had that compatibility "thing" happened as requested.
SAS was always reactive. It never offered superior usability. And it was never cheap.
However, SAS historically offered robust analytics with fast performance. That secured its market niche.
That’s still a valuable market niche. It will decline, in competition with open source. But making Viya compatible with V9 can mitigate that decline.
I think that’s the ballgame now. The skills and culture needed to start new revenue streams aren’t there. But that doesn’t mean imminent doom, rather a managed decline.
Should they rename the conference "Reactive-ate"? Not as catchy a title as "Innovate", but much more accurate. "Ge---r-ate" is probably most accurate, but obviously won't fly.
Unfortunately I tend to agree, we are behind the curve. Pricing is actually very good, cheaper than Analytics Pro alone, but requires an AWS instance for the customer. This loses out on the on-premise/desktop users, which are still a very large user base.
One thing I found interesting was the included Python package sasviya.ml. It is syntactically the same as sklearn, but leverages SAS procs on the backend. Essentially an open-source user could take their Python code and run it, without modification, and leverage SAS compute. In the example shown on stage it was many times faster on the same table. I wonder is this just an edge case that’s good for demoing? Or, are most ML algorithms faster in SAS than sklearn? I haven’t been able to find much info on the package, but hope to learn more.
Who is going to leave free or better performing products to pay for workbench? That's the question. As usual it's a come lately product created as a reaction against other products already long in market that work better and/or cost less.
Reactive reactive reactive. SAS specializes in reactive products. Being reactive does not equal being innovative. Case in point - whatever SAS generative AI product they're cooking up now because JHG said us too! Us too! We are behind again! Make a generative AI product we can sell so I can have more billions!
Maybe too late. Surely too little.
SAS loses market share on performance, usability, and cost.
If Viya can be made compatible with V9, that addresses only performance. Customers will still move to Tableau and PowerBI for usability, and to R and Python for lower cost.
This may slow the bleeding but does not attempt to stop it.
Too late. Nobody cares except for a handful of old time SAS users who are about to retire anyway. Sad to say, but SAS is irrelevant now.
Too late
Viya Workbench is quite interesting. Will be interesting to see how the monetisation efforts go.
Careful. You are about to get accused of spouting the party line.
Be more negavite. Make sh-t up.