Modern analytics are fast, no matter the size of the data you’re crunching, so analysts can run several iterations of their models and reach insights more quickly than using systems of the past.
In the past three years, Jim Goodnight, the CEO and co-founder of SAS, has been pushing his company to modernize the software algorithms it began developing 35 years ago. The goal: to create a massively parallel processing platform to run analytical jobs faster than ever before.
The result is SAS’s High Performance Analytic server, which is capable of running complex models on billions of rows of data in seconds. The company announced at its Analytics 2012 conference in Las Vegas on Oct. 10 that it would roll out optimized versions of its most popular software, ready to run on an analytical appliance built on Greenplum, Teradata or Hadoop, every six months.
In this interview with Data Informed Staff Writer Ian B. Murphy, Goodnight discusses the complexities of rewriting statistical algorithms for massively parallel processing, the boost in productivity that fast analytics provides, and the benefits of running models on a whole data set instead of a data sample. Goodnight also discusses the thinking behind SAS’s new Visual Analytics and how it ties in to the in-memory appliance. (Podcast running time: 17:26.)