High-performance analytics: it may be a show, but it's not an act

4

Last Friday morning I took a meeting with SAS CEO Jim Goodnight. Despite the hassle, I was happy to rearrange my busy schedule to squeeze him in.

Ha! Just a little joke there about my inflated sense of importance! Actually, the meeting was part of a regular series called "Conversations over Coffee", when Dr. Goodnight makes himself available for a small group of employees to talk about...well...anything that they want to talk about. Any employee can sign up to participate; with a cap of about 10 or so employees, it turns out to be a rather intimate crowd. This access to leadership is one of the things that makes SAS a great place to work.

I kicked off the discussion with a question about the biggest buzz here at SAS: SAS high-performance analytics. I knew that Dr. Goodnight and a technical-savvy entourage had just completed a "roadshow" across Europe to demonstrate these latest advancements from SAS, and I joked that I hoped he wasn't tired of talking about it. My question was: with all of the focus on these high-end technologies to solve specific (albeit huge) business problems, how will these advancements improve the lives of our foundation SAS users? You know: the folks who sling DATA step, SQL, and PROC code for a living?

Jim responded by first explaining the idea behind high-performance analytics and how SAS brought it to life. He explained the idea of rethinking huge problems that involve big data, and how SAS applies state-of-the-art computing technology to solve those problems in a small fraction of the time. Using hand gestures, he showed how with today's hardware, data can be loaded into memory and operated on across thousands of processor cores and...wait, instead of reading my report on this, you can watch Jim explain it himself. I'll wait.

So, do you get the picture? Dr. Goodnight is enthused about this stuff, and justifiably so. But back to my question: how will this focus on high performance help today's SAS programmer and BI user?

The answer, without getting too specific, is that we can expect the lessons learned from implementing high performance at the "appliance level" to work their way into the SAS language that we know and love. Already, we can see "HP" versions of the most computationally intense analytic procedures. It might take a little while, but as workstation-class and server-class hardware become more capable, the SAS language should evolve to take advantage of it.

I view it as similar to the space program in the US. For much of the 1960s, the nation was focused on a single goal: get a person to the moon. We reached that objective long ago, but we still benefit from the technology that was created as a result.

The analogy may be more appropriate than you think. During our Conversation over Coffee, one new employee asked Dr. Goodnight about his early career. Jim talked about several of his pre-SAS jobs, including a stint as a programmer on the Apollo space program. He knows first-hand the tremendous results that such an intense focus can bring.

Share

About Author

Chris Hemedinger

Senior Manager, SAS Online Communities

+Chris Hemedinger is the manager of SAS Online Communities. Since 1993, Chris has worked for SAS as an author, a software developer, an R&D manager and a consultant. Inexplicably, Chris is still coasting on the limited fame he earned as an author of SAS For Dummies.  He also hosts the SAS Tech Talk webcasts each year from SAS Global Forum, connecting viewers with smart people from SAS R&D and the impressive work that they do.

4 Comments

  1. I am sure it's not a surprise I am commenting on this post.

    I have never had an issue with SAS creating innovative products, but I have outlined my concerns (WRS Forum) that Visual Analytics has diverted development away from the bread and butter SAS EBI platform, particularly WRS. There are many fundamental problems with the EBI platform that I would have hoped SAS focused their attention on. It is hard for me to get excited with Visual Analytics given our organization needs a robust medium sized data solution rather than a high performance and expensive big data solution. It would be hard to justify us ever putting the resources needed not only to install the hardware for Visual Analytics, but also to maintain it. Furthermore, most of our analytics tasks can be performed in an adequate amount of time without needing in memory processing on multiple cores. I would rather have that slick Visual Analytics web interface in Web Report Studio now, rather than having to wait months for it to be retrofitted (if that is even going to occur).

    I think your analogy of the moon landing project is a good one (and forgive me if I push this too far). Billions of dollars were invested in getting man to the moon in what everyone thought at the time was a worthy mission, and would lead to bigger and better things for manned space exploration. However, in the end, the USA spent a lot of money, and invested a lot of time running a race that in hindsight did not need to be won, and which was basically for "show". Whilst there was some flow on effects, those big, overpowered rockets were decommissioned; the number of missions cut short, and investment was made in more sustainable technology. Unfortunately, even the Space Shuttle was overly complicated, and full of bugs that not only caused constant delays, but failure to fix has lead to disasters. In the end, it is those reliable rockets that do the day in day out work of putting satellites into space that have proved the most useful. But they don’t usually make the headlines.

    Perhaps in 5 years all of this will pay off, and even I will be thankful SAS invested in it. But right now, investment in the current EBI platform is what I think is needed.

    Nick

    • Chris Hemedinger
      Chris Hemedinger on

      Nick, thanks for the thoughtful comments.

      I do agree we've taken the "space program" analogy a bit far -- while I like to draw the similarities using a reference that mostly my American audience will relate to, we all know that government-sponsored endeavors often have aspects of waste and misguided motivations. Clearly, with SAS as a private company and its long history of commercial success, we are less vulnerable to such things.

      I appreciate that you don't feel that you need high-performance analytics right now. In fact, Jim Goodnight said as much during our meeting. "Most SAS users don't need high-performance analytics", he said. But some customers have an urgent need to solve huge problems against big data, and those customers have a high-impact on us: they may be finance services, health care, or even governments. Thus, it's encouraging for all of us to see SAS lead in this area.

  2. Pingback: Behind the scenes with SAS Visual Analytics - The SAS Dummy

  3. Pingback: Behind the scenes with SAS Visual Analytics - The SAS Dummy

Leave A Reply

Back to Top