Follow Datanami:
September 13, 2017

AWS Partner Ryft Leverages Cloud FPGAs

As more data moves to the cloud, analytics vendors are also embracing new cloud technologies designed to boost performance for emerging workloads such as business intelligence.

With that in mind, analytics specialist Ryft Systems Inc. said this week it would accelerate its business analytics offerings via Amazon Web Service’s (NASDAQ: AMZN) F1 FPGA-based instance announced last fall along with its “elastic GPU” service.

The accelerated business analytics service, now available on the AWS Marketplace, is designed to boost data search and analysis capabilities, including broader “fuzzy” search criteria. Specifically, the FPGA upgrade would allow users to perform advanced searches called Perl Compatible Regular Expressions, or PCRE2, on files before and after they are indexed. The FPGA-accelerated architecture would also allow fuzzy searches and matching on either indexed or un-indexed data, the company said.

Ryft, Rockville, Md., said its technology running on the Amazon F1 instance would help it leverage FPGA-accelerated computing hosted in the AWS cloud. The company released a pair of Amazon machine images on Wednesday (Sept. 13) used to deploy the new FPGA services on the AWS cloud.

The first is a toolkit designed to help users integrate new search and analysis capabilities into existing analytics interfaces or applications. The toolkit also is billed as “turning big data into small data” through analysis and “thinning” without data indexing.

The second image, called “Elasticsearch,” boosts the performance of the ubiquitous Lucene-based search engine by adding PCRE2 regular expression and fuzzy search capabilities. It also accelerates search and analysis across unstructured data and a range of files.

The analytics vendor asserts that data indexing requirements stifle the ability to mine data. The combination of Elasticsearch running on the AWS F1 instance reduced processing and searching a 1-Tb log file from more than 62 hours on a commodity AWS Elastic Compute Cloud server to 0.69 hours on the combined Ryft-F1 instance, the company claimed after benchmark testing. That, Ryft further claimed, equals a 91-fold performance increase.

Other cloud analytics features besides the AWS FPGA instance include a frontend running on the Ubuntu distribution of Linux along with Open API. The Ryft cloud also integrates with Apache Spark and database connectors.

The company added that it is readying software and algorithms to transform the leading public cloud into a “data analytics machine.” Ryft’s middleware layer provides the connectors and algorithms to help speed cloud analytics since, it asserts, a “cloud platforms alone simply doesn’t have what it takes.”

AWS said last November is new F1 instance addresses the continuing data explosion generated by the Internet of Things, video streaming and other demanding workloads. Specifications for the F1 instances pair Intel’s Broadwell E5 2686 v4 processors with up to 976 Gbytes of memory and up to 4 Tb of storage with one to eight FPGAs.

Ryft CEO Des Wilson predicted at the time of the rollout that the “AWS marketplace will embrace heterogeneous computing principles with the advent of its new F1 instance, opening up a whole new set of data analytics capabilities.” Ryft specializes in hybrid FPGA/x86 configurations.

Recent items:

FPGA System Smokes Spark on Streaming Analytics

Neo4j Touts 10x Performance Boost of Graphs on IBM Power FPGAs