The London Stock Exchange (LSE) has finished developing a new core trading system and has begun a year of testing. The project, based on commodity PC server hardware and a Microsoft .net infrastructure, is the culmination of a four-year, £35m technology refresh programme and is due to go live in early 2007.
The system is designed to cope with the boom in trading volumes due to the growing use of automated systems such as algorithmic trading.
Robin Paine, chief technology officer at the LSE, said, "Algorithmic trading and automated trading account for 40% of volume in our markets. These systems benefit from improvements in system performance and reduced latency [response time]."
Developed over the past 18 months by Accenture in India, the system has been delivered to London, where a team of 80 people will be testing it for 12 months.
The new trading platform uses the same core infrastructure as the LSE's Infolect system, the "tickertape" information delivery platform, including the metropolitan area network, servers and high-speed middleware from IBus. This has allowed the LSE to test how the infrastructure copes in a live trading environment.
Infolect is based on a Cisco network, the Microsoft .net framework and bespoke middleware written in C#, and can process real-time information 15 times faster than the system it replaced.
It has been rolled out in three datacentres. Two are active and run in parallel, while the third acts as an arbitrator if there is a discrepancy between the other two feeds.
A total of 120 HP Proliant servers have been deployed and the Infolect system processes 12 million real-time messages a day and is accessed from 100,000 terminals in over 100 countries.
The LSE has experienced 44% compound annual growth on SETS (stock exchange trading system) over the last five years, which has led the IT team to focus on processing delays in the system as well as delivering the required capacity.
To ensure it has enough processing capacity, Payne said, "We have bi-weekly meetings to discuss macro-economic trends and collate this with volume trading to make decisions on upgrades."