PHLX rolls out Equity Options AutoQuote System to traders as the first application to leverage its enhanced data architecture.

With the growing popularity of options and the resulting need for faster, more reliable fail-over capabilities, the Philadelphia Stock Exchange has made some big changes to its data architecture. The exchange decided to re-architect its in-house built system and is relying on TimesTen real-time event-processing software as its new backbone.

The first application to utilize the new technology is the PHLX's Equity Options AutoQuote System, an options-price-quote calculator and quote generator for specialists.

"We needed to be able to get market data and configuration data from an application when we had a server failure," says Doug Schafer, first vice president, options trading system development, PHLX. "We use a thin-client architecture, so it runs completely on a server and if we lose a server, all of those people have to fail over."

More specifically, Schafer says the exchange was looking to accommodate up to 100 traders to fail over and be up and running again within 30 seconds, rather than minutes.

In addition to simply failing over to another server, those traders also had to be able to load a considerable amount of market data and configuration data in order to trade again. Such configuration data includes interest rates, volatilities and dividends as well as custom information traders use to control the execution of orders within the trading system, such as how far they are willing to go up to match a best bid or offer price.

The homegrown model previously in place - called Electronic Data Access or EDA - required enhancements to handle the number of queries and the amount of data necessary. Rather than building an enhanced infrastructure internally, Schafer says the decision to buy was a better use of resources for the exchange. After looking at a number of database products in the disk space like Sybase and Oracle, and some Storage Area Network (SAN) technology databases, Schafer says TimesTen was selected because of its ability to handle the speed of updates and the number of queries at the exchange.

"We built a new application called Data Broker with TimesTen," says Schafer. "And it creates a common interface to any other application and you can query via direct SQL or precanned queries."

While the AutoQuote system is the first application using the technology to be rolled out to end users, Schafer says more will be added in the future.

Schafer adds that with the Data Broker technology, the exchange has standardized the way any trading application gets data. Rather than applications connecting to various databases for various data sets, "Applications just have to be knowledgeable to connect to Data Broker for data storage," he says. In other words, the connectivity is simplified as applications have to connect on the Data Broker level. The data is also partitioned into different Data Broker areas depending on how and where it needs to be accessed.

The PHLX implemented four TimesTen databases, running in two areas. "Two of the databases run at the data-server level where they're collecting market data and order and trade information and they are in hot backup mode, meaning each one backs up the other," explains Schafer. "Then we're running the same thing at the AutoQuote level."

Schafer says that the PHLX performed about 8 months worth of analysis and testing with the TimesTen database technology before deploying in an insert-only mode. That phase ran for another 5 months, with market data and information running through into the databases, but not out of the database and into applications. Then the AutoQuote application was rolled out to about 200 traders, with about 100 on each server. Moving forward, Schafer sees the TimesTen technology as being scalable up to 300 users if necessary.