How Barclays Chose Its Messaging Middleware
It was just about a year ago, on September 15, 2008, that Lehman declared bankruptcy and Barclays agreed to acquire its North American investment banking, and fixed income and equities sales, trading and research operations businesses. On the morning of this almost-anniversary, Stephen Neal, director, middleware at Barclays Capital, told attendees at the High Performance Computing on Wall Street show about the challenge of integrating the former Lehman businesses into the Barclays infrastructure, which spurred his quest for a new middleware solution.
Neal started the selection process in 2006 and gradually whittled his middleware vendor pool from 13, to four and finally to one. In the evaluation process, his most important consideration was the deviation from peak latency to average latency and ouliers. "A small amount of deviation and predictability were key for us," he sayd. His group simulated a trading environment to test the products, and Solace's 3260 Content Routers came out the winner. The firm now uses Solace to integrate messaging applications spanning the front, middle and back office.
Offloading messaging intelligence to the Solace product, Neal says, frees up CPUs for the handling of business applications such as trading algorithms and risk analysis. It's helping to reduce the firm's data center footprint on a 5:1 ratio and reduce the overall number of servers the company uses even as volumes go up.
On Neal's drawing board for the coming months are persistent messaging, improved monitoring and management, custom hardware acceleration, low-latency switching and the AMQP protocol, for which Barclay's has joined a working group, Neal says.
Tweet This