Upgrading the exchange

Betfair’s online betting exchange is among the UK’s most successful technological innovations of the recent era. Allowing gamblers to agree odds with one another, even during the course of a sporting event, it is today the largest betting exchange in the world, supporting more trades each day than all of Europe’s stock exchanges put together.

In October last year, the company was launched on the London Stock Exchange, raising more than £200 million (Betfair has since issued a share buy-back scheme to appease disappointed investors).

Before the initial public offering took place, Betfair had laid out a three-year plan to upgrade its technical architecture. According to software development director Hugh Fahy, the injection of funds has not led to an IT spending spree – the extra scrutiny of the public market has instead solidified its focus on that plan. “That’s been useful, because it has reduced the chance of us overextending ourselves as we redevelop our core platform,” he says.

One of the key components of the platform upgrade is to extend the use of service-oriented architecture principles.

As Betfair’s business began to take off around 2005, Fahy explains, its priority was to make sure that the betting exchange worked as reliably as possible. It therefore built supplementary functionality, such as the customer’s online wallet, to be tightly integrated with the exchange.

In recent years, however, the company has branched out into services including online poker and arcade games. Although customers can use their winnings from the exchange on these services, the fact that the wallet is tightly coupled to the exchange means that their ability to play might be constrained by their betting position – not ideal when Betfair wants to encourage customers to use as many services as possible.

This is one reason why the company is redesigning its core platform according to SOA principles. When the wallet functionality is supported by separate services from the exchange, gamblers will have greater freedom to spend their money.

Global Growth

Another reason is international expansion. Although betting exchanges are banned in most US states, New Jersey and California have recently passed laws permitting some forms of exchange wagering.

To tap these markets, Betfair has to introduce new, separate exchanges tailored to each state’s regulatory restrictions. However, it does not want to build separate wallet functionality for each exchange, for example, and a service-based approach allows it to reuse its existing assets.

The company has therefore implemented three new SOA components: a service container (or repository), a service look-up (or registry) and a ‘publish/subscribe’ messaging framework. The latter is based on Progress Software’s Sonic MQ messaging system, and Fahy describes it as the ‘event handler’. “Our system generates a whole bunch of events – matching a bet, a customer changing their loss limits, or an account being locked or suspended,” he says.

Part of the reason for selecting Sonic MQ, Fahy says, is that it can compensate for degraded network quality. Betfair has nine data centres around the world, and many of them are in locations where the wide area network (WAN) is not always reliable.

The first use of the messaging framework is to satisfy the Italian gambling regulator, which is introducing regulation of betting exchanges this year. Unlike countries such as the UK, where the Gambling Commission audits betting companies periodically, Italy’s regulator creates a unique ticket for every bet that is placed in the country, in real time. So each time a bet is matched on Betfair’s exchange in Italy, the company will soon have to inform the regulator that very minute.

The business that would have been lost in Italy without it was enough to justify the entire investment, says Fahy.

Pete Swabey

Pete Swabey

Pete was Editor of Information Age and head of technology research for Vitesse Media plc from 2005 to 2013, before moving on to be Senior Editor and then Editorial Director at The Economist Intelligence...

Related Topics