The crash of the TradElect trading system on the London, and Johannesburg, Stock Exchange for seven hours was a disaster that probably led to the loss of around £3bn in trading.
The system was based on SQL Server 2000.
High volumes of trading on global exchanges were expected that day, following the news that Fannie Mae and Freddie Mac were being taken over. At 9:15 AM GMT, the Exchange's software failed due to "connectivity issues." It took Six-hours and 45-minutes for the London Exchange, along with the Johannesburg Stock Exchange, once more to start trading.
This incident has put a new question-mark on the LSE’s viability. In terms of cost, it is one of the most spectacular in recent history
The system was built, evidently, built using Windows Server 2003, SQL Server 2000, and a custom .NET application in C# by Microsoft and Accenture. Microsoft described the application as being ‘One hundred per cent reliable on high-volume trading days’ http://www.microsoft.com/uk/getthefacts/lse.mspx. The TradElec runs on more than a 100 HP ProLiant servers in several locations in London.
The system promised to give sub-ten millisecond response times, rather than 7 hour response times.
The LSE describe TradElect in their website as ‘…the Exchange’s new trading system. It brings unprecedented levels of performance, enhanced functionality and new services to our markets whilst maintaining our exemplary record for reliability.’
A spokesman told Reuters on Tuesday. "It was software-related, a coincidence, due to two processes we couldn't have foreseen, We've introduced a fix and we're confident it will not happen again.". She said the fault was not due to high trading volume. "That's a red herring. We were able to eliminate that pretty quickly."
So what was it? Another statement attributed the failure to network software issues between the LSE's Extranex private network (which links the exchange and clients) and the trading gateway to TradElect.. As yet, there is no confirmation of this by Cisco.
The whole incident has had the Linux world rubbing their hands with glee and maintaining that Microsoft technology just wasn’t up to it. Some commentators have even tried to make out that the LSE would have done better with a LAMP-based system with MsSQL.
The whole incident has been a major embarrassment for Microsoft since they made so much of the reliability of the system. Once the amusement of searching Microsoft sites for references to TradElect has waned, though, there comes the thought that, despite the crowing of the Linux gurus, it is quite likely that it had nothing whatsoever to do with the Microsoft technology, and it really was a problem with the software within the network. If this was the case, why hasn’t a statement been made? The delay is bad news because, in the meantime, what damages the reputation of TradElect , also damages confidence in the Windows Server, SQL Server , and NET platform. It was, after all, going to be the ‘benchmark’ application for the platform.