Lloyd’s of London gears up for regulation

Lloyd’s of London has consolidated its five datacentres into just two

Insurance market Lloyd’s of London will be relying on virtualisation technology and grid computing to deal with new regulatory requirements for the sector and the resulting increase in data processing.

The requirements, dubbed Solvency II, are due imminently and will introduce comprehensive risk management procedures that set capital levels and implement ways of identifying, measuring and managing risk.

This is a major event from an IT standpoint, and it has prompted a new set of technical requirements for insurance firms such as Lloyd’s.

“It is really quite exciting, as the people who do our capital and risk modelling are moving up a notch in terms of complexity and scale – previously, a model might look at 10,000 or 100,000 [data] interactions, whereas the new models will analyse a million-plus interactions,” said Peter Hambling, chief information officer at Lloyd’s of London.

According to Hambling, the company focused on “putting the house in order” during 2009 ahead of the introduction of grid computing, with significant work being carried out around storage and server virtualisation.

Setting up for grid computing
Lloyd’s consolidated five datacentres, which were at the end of their useful life, into two. Forecast to complete by year-end, the sites will cost less to run, be available for more hours and with more capacity than the previous setup.

Hambling sees servers as commoditised products, meaning that critical changes to business processes have not required mass replacement of the kit. Instead, Hambling has opted to replace servers as and when they need updating. Lloyd’s’ main hardware vendor is Dell, while virtualisation software is provided by VMware and Hyper V.

To operate its high-performance platforms and meet the upcoming regulatory requirements, Lloyd’s is now standardising its control practices and procedures.

“We had to create technical space for the computing grids and are now working with customers to understand their requirements before we size and build them,” said Hambling.

The specific modelling component of the grid platform will be developed by an external specialist, while the insurance group is establishing which hardware to run the grid on. According to Hambling, it will almost certainly be run on Blade servers.

“We also had to determine our storage strategy. With this type of [grid] modelling, you have a relatively smaller amount of input and output data, but a very large amount of intermediary data. We are currently in the design stage and looking at how to best service our data needs,” he said.

Lloyd’s is also training staff to support its actuarial modelling process and will need engineers to make sure the grid platform performs optimally.

Tight timeframe
The main challenges of the IT work surrounding Solvency II are posed by the fact that requirements have been released in tranches by the regulator and Lloyd’s has not yet seen the full requirement set.

“We will need to get significant quantities of engineering done in a small amount of time, so there is a lot of pressure,” said Hambling.

“We have put the [infrastructure] foundations in place and are working with the requirements we know about. We were able to anticipate some of the knock-on effects of Solvency II, such as [the need for] resilient technical space and storage,” he said.

Hambling added that one way of dealing with tight timeframes was to build out IT environments before projects started, as well as negotiating with partners to supply extra resources and hardware if needed.

The IT chief said that his biggest leadership challenge over the next year is to continue to improve the service, while meeting ever-increasing expectations from management.

“[Senior management] expect you to reduce your IT fixed cost and reinvest savings into projects. They want more for less and expect the world to continue getting better, slicker and faster and you have to find a way of meeting the requirements,” said Hambling.