SA logo

Even though organizations are replacing more and more of their proprietary infrastructure with commodity gear, not every workload is better served by x86-based servers. According to the Maryland-based Ryft Systems Inc., data search is one such use case. The vendor claims that its namesake appliance can help improve query performance by up to 600 times thanks to a combination of specialized FPGAs and a new fuzzy string matching tool that it’s rolling out today.

Much of the credit for the software’s performance goes to the fact that Ryft’s engineers have removed the need to turn unstructured datasets into a more organized form before executing searches. In fact, the company says that the engine doesn’t even need to prepare an index of the information that it’s used to access the way more conventional alternatives do. Its algorithms avoid the hassle of the preprocessing stage entirely and instead run queries directly against the raw records, which can save a great deal of overhead in a large environment.

From there, Ryft’s fuzzy search software doesn’t look for exact matches to the string entered by the user but rather scoops up all the text within a certain margin of error from the input. The functionality can come tremendously handy when trying to navigate through natural-language content like social media posts, which often contain spelling and grammatical errors that trip up traditional search engines. It’s also useful for sifting through an organization’s internal business documents, customer records and practically any other kind of textual information produced by humans.

Ryft says that customers can in this way scan more than 200 gigabytes of unstructured data per second if they combine several of its 1U appliances in a cluster. The company provides the choice of deploying the hardware behind the firewall or having the systems hosted in a remote data center.