The New York Stock Exchange and IBM announced in a conference call todaythat TradeWorks, a new processing system complete with custom hand-helddevices — all designed and built by IBM — was up and running on the tradefloor.
“The key point is that the NYSE represents the largest application in theworld,” Mike Gilpin, a Forrester Research analyst, told TechNewsWorld. “It’svery high volume, very high value — it’s the kind of software that reallyrelies on critical infrastructure, so it says a lot about IBM’s ability.It’s a very significant endorsement of IBM’s WebSphere platform.”
First Outside Company
The NYSE previously had written its own middleware through its SecuritiesIndustry Automation Corporation (SIAC) because it had never been able tofind commercial products up to the job, Gilpin said.
The new order management and messaging system supports some of the 1.6billion shares traded daily. Floor brokers, who make transactions for largeinvestors such as mutual funds, now have wireless devices that allow them totrack orders, manage paperwork, and buy and sell stocks.
IBM designed deviceswith screens bigger than those of an average hand-held to make use easier. The NYSEbought 3,000 of the hand-helds, designed by IBM Engineering & TechnologyServices. It also bought 3,000 custom Linux workstations.
Brokers send and receive an average of 75,000 messages a day — up 200percent in the last two years. The 1.6 billion shares traded isup from 1 billion in 2002, the companies said.
IBM labels the system “extreme availability.” It comprises WebSpheremiddleware, DB2 database and Tivoli management software on the back end forthe TradeWorks application, which is Java-based. IBM’s mainframe zOS sitsunderneath DB2.
No Room for Error
“The New York Stock Exchange is the financial center of the world. We cannotafford to have anything that cannot stay up,” Willy Chiu, IBM vice presidentof High Performance on Demand Solutions, said in the call. Testing, in whichany conceivable problem was considered, took over ayear.
“They built the technology for failover in a way to handle all the scenariosthey had tested for,” Gilpin said. “They found stuff that nobody ever foundbefore that needed fixing. Eventually it will get plugged back intoWebSphere.”
While most companies do not need the kind of extreme availability andscalability that IBM engineered for the NYSE, other firms will still see somebenefit from it.
“It’s like a science experiment that proves something,” Gilpin said. “Itdoesn’t mean it will happen in your environment, but it shows how far it canbe taken, and the benefits eventually trickle down to others.”
Today orders and reporting are still running on the NYSE’s old system, butNYSE CTO Roger Burkhardt said in the call that IT personnel are working ontransferring that capability to the new system.