What can a skateboarder teach you about Hadoop?


Have you ever wondered how extracting value from big data might be like skateboarding?  Really, it never crossed your mind? Actually, to be honest I had never considered it either, not even one little bit.  At least, not until I watched a skateboarding legend explain, “The Art of Good Practice” at a big data conference last month.

Yes, you read that correctly.  Rodney Mullen, considered one of the most influential skateboarders in the history of the sport, presented at the 2014 Strata Santa Clara conference.  Had it not been for his O’Reilly Media t-shirt, I would have guessed he was at the wrong conference.   But as I began to listen to his message, it became clear that he was indeed in the right place, talking to just the right people.

Here are the three things Rodney taught me:

Community of thrashers

Throughout Rodney’s presentation, he mentions the words, “my community.”  He is referring to not just his competitive teammates, but the skateboarding community at large.

He explains that it’s similar to the open source technology community, each of us borrowing ideas from the next, building upon what our colleagues have done.  He also explains that each contributor has a distinct set of skills and attributes.

This collaborative experience is also true for the new generation of people implementing and managing big data projects.  It’s no longer a single team working in isolation, but rather a community of practice involving specialists from virtually all areas of the business.  The application of advanced analytics has never been more collaborative, which of course is facilitated by high-performance analytic tools that span the entire analytic lifecycle: from managing data, to predictive analytics.

The skateboarding executive

Rodney explains the process of making a difficult “trick” automatic.  That is, consistently practicing the same motor functions enough so they become second nature or, as he describes, “executive.”

In big data, this is operationalizing your baseline analytics.  No exceptional skateboarder has ever completed a trick flawlessly the first time.  However, as they mature and practice, they can operate executively with their core skills, allowing them to layer on more complexity over time.

Likewise, to draw value from big data, your community must be able to first execute basic analysis, but develop the skills and adopt the technology to expand on those basic executive tasks into far more complex modeling.  It is only in these advanced use cases, that we find the “tricks” that disrupt our competitors, help us detect risk sooner, predict outcomes more accurately, and win the trophy.

Tuning, timing and food carts

The process of innovation for both skateboarding and big data analytics can be a painful, repetitive process.  Often times it involves re-running the same “trick” over and over for hours, keeping many things constant, while tweaking small aspects ever-so-slightly.  But what if each trick took hours or days to run?  What if you were occasionally knocked off the board by a bully that smelled like cheese?  How much longer would it take you to get it down, to make it executive?

Working with a modern analytics platform is like giving your skateboarder an unlimited skate-park pass, a beer-stocked trailer, and cheap food-carts in the parking lot. For example, SAS In-memory Statistics for Hadoop, available this month, allows the data scientist to iterate through thousands of scenarios, in a massively parallel, in-memory way, tweaking variables with impunity. Essentially, it turns your every-day skate-statistician, into a gnarly shred-scientist!

I'll leave you with these questions:

  • Does your big data community of shredders have seamlessly integrated tools that can carry you from data to decision, then all the way to execution?
  • Does your community develop a core set of “executive” analytics and processes upon which to innovate and grind the rail?
  • Can you iterate through your high-value “tricks” faster than Rodney can pull off a “Caballerial?”

With SAS In-Memory Statistics for Hadoop, the resounding answer is, yes.  At the end of his presentation, Rodney said with a giggle to the audience of big data practitioners, “Hopefully, somehow this translates into what you do.”  Well Rodney, my answer to this is also yes.  Yes it does.


About Author

Keith Renison

Principal Solutions Architect

Keith Renison is a Principal Solutions Architect with SAS’ Global Technology practice. His background traverses a broad range of industries with a current focus on program management, big data, Hadoop, and visual analytics. Prior concentrations include sustainability and performance management, cost and profitability modeling, enterprise data integration, and business intelligence. Previous roles in SAS include technical support and delivery consulting. Prior to joining SAS, Keith was a systems and database administrator in the telecommunications industry, and holds a Bachelor’s Degree in Business Management and Organizational Leadership from George Fox University.

Leave A Reply

Back to Top