Sans Repetitive Data, Networks Run BetterSans Repetitive Data, Networks Run Better

Network congestion and slow response time threatened IDEXX Laboratories Inc. with having to write a bigger check to AT&T each month. The maker of animal diagnostic products planned to add four T1 lines to its frame relay network, for about $3,000 a month.

That is, until senior network analyst Rob Edwards tested a Sequence Reducer network appliance from startup Peribit Networks Inc., which adopted pattern-matching algorithms used in DNA mapping to find repetitive data on a network. The box increased available bandwidth by 50%, prompting IDEXX to sign a global contract with Peribit. It was an easy decision, Edwards says. "Spend $3,000 a month on a new link, or spend $9,000 one time for a device."

IDEXX needed the extra bandwidth for a strained warehouse-management application. Representatives at headquarters in Westbrook, Maine, would send product and shipping information to the Memphis, Tenn., warehouse over a frame relay network. But congested bandwidth between the sites resulted in a 20-second delay for each bar-coded shipping label that needed to be printed. After installing the Peribit boxes in September, the delay dropped to 9 seconds. IDEXX, which plans to put boxes in five of its 32 sites, expects a return on investment in less than nine months.

Peribit finds that on average 78% of traffic traveling across a network is repetitive. When someone sends a file, a Peribit box detects and strips repetitive data, replaces it with a code, and sends it across the network, where the codes are replaced with the actual data at the Peribit box on the other end.

A potential problem: As users gobble up the extra bandwidth, significant congestion is likely if the Peribit box goes down.