When I talked to SAS CTO Keith Collins recently, he said SAS is coming out with new innovations in the high-performance analytics space at the rate of every six months. And he’s not just talking about small product upgrades twice a year. In many cases, he’s referring to a whole new paradigm that changes the way products are developed and revolutionizes the way organizations work with their data.
Of course, SAS is not expecting customers to change their world every six months or to learn new coding techniques twice a year. Instead, we’re making every effort to build high-performance analytics into your existing environments so that the changes are invisible to analysts and end users. You still create models and find answers the same way you always have. Now, you can just do it faster.
Next up, the SAS LASR Analytic Server will offer another new way to perform in-memory analytics on data stored in Hadoop. It is the first system on the market specifically engineered to address advanced analytic use cases in diverse environments. It can handle the computational complexity of large-scale exploratory data analysis and visualization as well as predictive analytics and data mining on big data.
Two ways to run SAS® LASR Analytic Server
SAS LASR Analytic Server is thin-layer technology that enables SAS to run within distributed computing environments such as Hadoop, or alongside distributed relational databases such as Teradata and Greenplum. It provides applications with the responsiveness and high throughput required by large analytic workloads and analytic-intensive applications. Applications access the SAS LASR Analytic Server using direct SAS connections and standard interfaces. Read on to learn a little more about the two SAS architecture options.
- SAS® LASR Analytic Server running within Hadoop
In the past, it would have cost millions of dollars to store even a few terabytes of data. Hadoop has changed that game. Hadoop can aid in the storage of the data as well as in exploring and analyzing it by allowing businesses to deploy distributed applications running on thousands of nodes and sifting through petabytes of data. For SAS solutions, Hadoop provides an open, simple and robust architecture that addresses the needs for fault tolerance, redundancy and scalability. How does SAS LASR Analytic Server work with Hadoop? Within each Hadoop node, a thin-layer SAS process takes incoming SAS commands and returns results. It’s that simple. You don’t need to install SAS on every node because the Hadoop system takes care of distributing the processing, managing memory, controlling the job and managing the workload.
- SAS® LASR Analytic Server running alongside Greenplum or Teradata
Data warehousing, first popularized in the mid-1990s, is now a mainstream technology that has moved from back-office, query-and-reporting style analysis to become an aid in operational decision making. The hardware architecture and enterprise-class features of Teradata and Greenplum data warehouse appliances make them excellent vehicles to deliver SAS in-memory technology. SAS has worked closely with RDBMS partners such as Teradata and Greenplum so that SAS can now run alongside their distributed computing architecture, bringing SAS processing to the data rather than bringing data to the SAS processor. Both are good choices for SAS in-memory technology. They provide the data distribution, process management and memory management needed to run analytics alongside relational processing.
Faster, more precise insights
SAS LASR Analytic Server supports the architecture you already have in place – or it allows you to build one quickly with Hadoop. As a result, the move to analyzing more data, running more complex calculations or more iterations of existing models is much easier than ever before. Bottomline: This new technology from SAS makes big data accessible to a much wider user base.
In-memory analytics for Teradata and Greenplum are available now, and The LASR Analytic Server will be available for Hadoop later this year. You can read all about it in the white paper, "In-Memory Analytics for Big Data." To learn more about SAS capabilities for accessing Hadoop that are available today, read Mark Troester's blog post about SAS and Hadoop.