Ryft made news this week by accelerating petabyte-scale data analytics to 200+ Gigabyte/second, powered by a cluster of high efficiency Ryft appliances built on the company’s hybrid FPGA/x86 architecture.

As I scanned Twitter for the news, I stumbled upon @GigaStacey’s timely repost of an article she penned way back in 2011 calling for new, more efficient architectures to handle the overwhelming growth of data. She boldly stated what so many others had been thinking as they struggled to contend with the never-ending torrent of data volume, velocity and variety:

“As part of that shift we need a computing architecture that will handle the storage of data, and the heavy processing power required to analyze that data, and we need to do it all without requiring a power plant hooked up to every data center.”

Even five years later, this article is worth a read because it so effectively sums up the big problem with big data. Historically, you’ve needed both a big data center and a power plant hooked up to it to extract insights needed to compete in the “Age of Insights.” And, that’s simply not practical or sustainable, especially as we begin to think about taking analytics closer to the source of data to enable the Internet of Things.

This article resonated so strongly with me because it tackles the driving reason behind our introduction of the Ryft converged network attached storage (NAS) and compute appliance—the commercial version of our proven architecture for high performance analytics like fuzzy search, machine learning and image and video analysis. This new, modern architecture was designed from the start to optimize computing, I/O and storage, eliminating all of the bottlenecks that have stifled data analytics appliances and adding the requisite ease of use to encourage adoption in the enterprise. It does so by combining the massively parallel processing capabilities of FPGAs (which happen to be more efficient than any other processor today) with the ease of integration offered by x86 and an open API to make data analytics fast, easy, scalable and sustainable. In fact, a single Ryft appliance requires less power than a hair dryer at peak usage! No power plants needed.

Now, with the addition of easy cluster management, Ryft is enabling Petabyte-scale workloads all while breaking through the bottlenecks that lead to sluggish performance. To summarize, Ryft CLUSTER allows customers to:

  • Scale data analytics performance and storage linearly to petabyte levels.
  • Ensure high availability, reducing the need to scale out inefficient and costly server farms.
  • Speed analytics to 200+ GB/second—including complex workloads like fuzzy search—without introducing any added complexity or time-consuming ETL and indexing for high performance analytics.
  • Streamline data analytics deployments by operating either as a stand-alone cluster of Ryft FPGA/x86 computing appliances or as part of an existing Apache Spark or other big data ecosystem.
  • Manage an entire cluster with the ease of administering a single 1U Ryft appliance.
  • Provide easy and automated multi-tenant analytics services.
  • And, reduce operational costs by 70% or more.

I encourage you to take a look at our approach and share your thoughts. This truly has to be seen to be believed so we are showing off our cluster at the Strata Hadoop World event in San Jose this week in booth 1409. If you’re not attending, take a look at our recorded demo.

Leave a Reply

Your email address will not be published. Required fields are marked *