Fine-tuning an SOA
- Reduce text size Decrease text size
- Increase text size Increase text size
- Print article Print
- Jump to comments Comment
- Share this article Share
- Email article to a friend Email
How Hiscox resolved the performance blind spots in its service-oriented architecture
In the past five years, Hiscox has radically revised its approach to selling insurance. Having operated mainly as an underwriter at the Lloyd’s of London insurance and reinsurance market, it has since made inroads into the retail market.
“Go back five or six years and most of our revenue was derived from Lloyd’s,” recalls the company’s CIO, Andrew Turner. “But we’ve now got a thriving retail business with individual customers such as homeowners and small businesses.”
What this means operationally is that more sales come in through Hiscox’s website and through an ecosystem of insurance brokers operating via their own channels. Scaling its legacy applications across these new channels was a technical challenge that drove Hiscox to adopt a service-oriented architecture (SOA).
Today it uses Software AG’s webMethods SOA suite to integrate and reuse application components, such as its ratings engine and payments authentication service. “It means that we don’t have to throw away whole applications – which take years and millions of pounds to implement – every three to five years,” says Turner.
This approach did not come without its technical challenges, however. With so many points of integration between the various services, when an error or performance issue arises it is not always easy to locate the source of the problem, Hiscox found.
“If you get issues during development, it can take a lot of time to work out exactly where the bottlenecks and pinch points are,” says Turner.
In 2010, the company sought a solution to this particular blind spot, and its search led to application performance management provider dynaTrace. Hiscox decided to deploy the Austrian company’s tools following a successful proof-of-concept project.
The most immediate benefit has been to eliminate the need to monitor application performance manually, and for any subsequent investigation when issues do arise. “It runs across applications to look for ways we can improve the integration,” he comments. “It spots problems [automatically], which really cuts down the time in terms of investigating issues.”
It may have required some supplementary tools to fine-tune, but Turner nevertheless endorses the SOA model wholeheartedly, as its distributed approach to application functionality reduces the risk of failure.
“In the past, insurance companies have looked for this nirvana of one integrated application that can do everything from policy administration through to claims, as well as putting up a front-end website,” he says. “The trouble is, when one of those things breaks, you end up just throwing the whole thing away.”
