IBM to Provide NYSE with Extreme Availability Processing

IBM and the New York Stock Exchange announce that they are collaborating on a new order management and messaging system in support of the 1.6 billion shares traded daily.

IBM and the New York Stock Exchange announced this week that they are collaborating on a new order management and messaging system in support of the 1.6 billion shares traded daily.

The order management system includes 3,000 handheld devices, designed by IBM Engineering & Technology Services, that will be used by brokers on the trading floor to place orders.

The IBM-NYSE work on the new TradeWorks infrastructure includes what is believed to be the most extensive testing ever done for a Java-based application environment. IBMs WebSphere infrastructure software, DB2 data management software and Tivoli management software will act as the back-end engine room for TradeWorks, processing the increasingly heavy flow of transactions. The DB2 back end will run on an IBM mainframe running zOS.

In addition to the handhelds, IBM is supplying custom-built Linux-based workstations with high-resolution graphics, low power and redundant connectivity. Traders, brokers and clerks will use these workstations to relay real-time market intelligence from the Exchange floor to upstairs trading desks.

/zimages/3/28571.gifThe NYSE clarifies its automation campaign. Click here to read more.

IBM officials described the system as one of the most tested systems the company has ever developed, not unlike the testing that was used for the space program. TradeWorks is based on Java, J2EE and WebSphere, with DB2 on the back end storing the trade records and customer information. The wireless devices have to be seamlessly integrated, and the end-to-end testing reportedly lasted over a year.

One of characteristics of the tests involved mimicking the spiky nature of placing trades. The system needs to respond in subseconds as well as be able to handle failure. How does the system recover from a software or hardware break? Much of the testing was done to flush out those issues before it went into live production.

There was an IT team internal to the NYSE that looked for "the most devilish ways to break things," according to Roger Burkhardt, chief technology officer of the NYSE. "Our team tried to conceive of the worst thing that could possibly go wrong, and then see how the system could recover. Were very happy to be out of the business of writing middleware and hand that over to IBM."

Next Page: Needed: Uninterrupted availability.