Driven by trends such as algorithmic trading, business models for liquidity providers are expected to undergo dramatic changes in coming years. First, the majority of trading in most instrument classes will be done by computers talking to computers. Second, trading volumes are expected to undergo explosive growth in both pure numbers of trades, and in the complexity of those trades. In fact, it is reasonable to assume that by 2010 large exchanges will have to cope with as many as ten million trades per day, with averages of 200 shares per trade and 100 orders per trade. That translates into a transaction processing load of more than one billion per day, or a peak volume of 175,000 per second.

While algorithmic trading is increasing the overall number of trades, orders per trade are also increasing fast, spurring demand for extreme-performance transaction processing in the financial services sector.

Getting Away From Big Iron

As with most businesses reliant on high volume transaction processing, large liquidity providers have built their operation on “heavy iron” solutions—proprietary mainframes from IBM™ and HP™. These machines, however, are not powerful enough to handle the expected increase in transaction volumes cost effectively. For instance, a liquidity provider today would need 24 fully configured HP Non Stop™ machines to achieve processing speeds of 10,000 messages per second. If peak processing loads scale as expected to 175,000 transactions per second by 2010, that same enterprise will require approximately 350 HP Non-Stop mainframes for order matching alone.

Preparing for Skyrocketing Transaction Volumes

Kabira Trading™ allows liquidity providers and other financial services firms to thrive in a market driven by skyrocketing transaction volumes and increasingly rigorous service level demands. Designed specifically for the emerging transaction-based global economy, the Kabira trading systems are built on our high-performance, transaction technology. With a 64-bit in-memory platform, Kabira is capable of executing more than 50,000 transactions per second running on standard open systems hardware—providing a clear road away from expensive mainframes. Since Kabira software supports real-time data, transactions, applications and connectivity—all in high-speed memory—transactions execute up to 100 times faster than on disk-based systems. With these ultra high-performance transaction processing capabilities, trading entities can handle burgeoning traffic now and scale to meet future growth, while also:

• Reducing costs for infrastructure, operations and maintenance and drive down cost per transaction.
• Integrating and adapting to rapidly changing financial networks.
• Eliminating downtime-induced revenue loss by maintaining service with Kabira's five 9s High Availability and clustering technology.
• Enhancing customer services by readily integrating external products, such as risk management into transaction-delivery workflows.
• Shortening time to market with new service offerings.
• Distributing payment capabilities to utilize more resilient and modularized enterprise architectures.