Hiscox replatforms to Microsoft Azure

Insurance firm choose Azure due to relationship with Microsoft, and starts working in a DevOps way

Insurance firm Hiscox has replatformed its core systems onto Microsoft Azure. The organisation's group CIO, Ian Penny, explained that the strategy is to increase agility.

"We run an agile DevOps approach," said Penny. "We do continuous integration and deployment, with [software] releases every two weeks. We can only achieve that if we use cloud computing."

He explained that at any one time his organisation runs between ten and thirty different copies of its environment for developers, staging and user acceptance testing, amongst other needs.

It's a truly agile approach, Penny said, as his teams create and tear down these environments "at will".

"We use open source tools like Bamboo, Puppet and Terraform, often from quite small, innovative startups, and it all deploys into our Azure platform," he said, adding that Azure was chosen because of the organisation's relationship with the Redmond-based firm.

"We use Azure as have strong relationship with Microsoft across the board. They help us with their deep technical expertise, which frees us up to focus on the business elements.

"Microsoft was an obvious first choice as we have enterprise terms and conditions in place with them already, and they understand how to sell to a regulated industry like ours. It's a skill that the other players are still learning."

This move into the cloud has dramatically reduced the time for certain price modelling activities, Penny said.

"In the US if you buy home owner insurance, there's no flood insurance included, you have to purchase it separately. We examined whether we can provide cost effective flood insurance there, and realised we can as long as we calculate the flood risk well."

However this process is more complex than it sounds, with Hiscox holding over a billion geolocations in its flood datasets, which means that interrogating that database and calculating risk accurately was a lengthy operation.

"We have more than a billion geolocations in our flood data for the US. Calculating all flood risk using on premise infrastructure would have taken us around eight months. Through the use of Azure batch processing and scaling, spinning up the resources as we need them, we got it down to 12 hours.

"So the conversation now isn't can we calculate it, but what's the business benefit of calculating it every night, or every time we get big policy coming in."

He argued that this is part of a strategy to move IT away from being seen as a cost centre, to a business partner.

"We're moving the conversation away from cost and towards value. What additional premiums can we generate and benefit to customer base?"

Penny will be speaking at Computing's new event ‘Artificial Intelligence and Machine Learning Live!' which will be held in Central London on 19th November.

Read Computing's report into Cloud Access Security Broker technology, a way of helping enterprises use the cloud more securely.