UK-based Aon plc, the ultimate parent company of Aon Benfield Securities, Inc. (ABSI), is a leading global provider of risk management, insurance and reinsurance brokerage. ABSI focuses on securities and other financial products of interest to insurance companies, including catastrophe bonds, contingent capital, sidecars, collateralized reinsurance, industry loss warranties, and derivative products. ABSI offers services such as underwriting and placement of new debt and equity issues, financial and strategic advisory services, and a leading secondary trading desk. In addition, ABSI in conjunction with its affiliates provides distinctive analytics, modeling, rating agency advisory, and other consultative services. Aon plc operates in 120 countries and employs 65,000 people in 500 offices worldwide, with annual revenue of approximately $11.5 billion.

Investing and risk go hand in hand. Many insurance retirement products contain financial guarantees, and these can span 30 years or more before paying off—so understanding whether an investment is a good risk over the coming decades is critical. A client company might run 5,000 different scenarios using a monthly time step for any given investment policy. Multiply that by five million policies, and the potential economic scenarios that might play out over the next 30-50 years, and the need to update risk analysis multiple times a year, and you have a massive computational challenge on your hands. It’s work that requires high-performance, highly scalable computing and hardware that costs millions of dollars.

For a financial services provider like ABSI, it is crucial to provide clients with effective business risk management solutions. ABSI helps insurance companies price their investment products, analyze risk, and address regulatory requirements by using a financial modeling tool called PathWise, that simulates millions of potential economic scenarios and uses stochastic simulations to evaluate potential outcomes.

ABSI customers typically maintain robust grids in their own data centers with as many as 2000 CPU cores, but the company’s intricate, complex growing financial modeling and reporting needs calls for a quantum leap in computing power: GPUs, or Graphical Processing Units. ABSI’s GPUs have tens of thousands of cores to process parallel workloads efficiently. Based on ABSI’s benchmarking, its clients stand to gain large increases in performance per dollar by switching to ABSI’s GPU-powered modeling tool. “Using GPUs is a massive leap forward for them—and that’s without factoring in the huge spikes during quarterly reporting periods,” says Peter Phillips, ABSI managing director. “When you include that in the analysis, using our business solution can be as much as 500 times more efficient in terms of performance per dollar for some clients.”

When ABSI first launched PathWise, it used a broadband HPC processor in a co-located data center, but found that they needed a more scalable service than the co-location facility could provide. “Our challenge was being able to scale up as necessary without the need to build an expensive data center,” says Aamir Mohammad, director at ABSI. “Specifically, we wanted GPUs — the rocket fuel of our industry. They’re also much more commoditized and less expensive than traditional data centers with CPUs.” The company needed a solution that would scale with it during peak periods, so it started to look at cloud computing.

ABSI found that by using Amazon Web Services (AWS) pay-as-you-go pricing, it could spin up large numbers of GPUs quickly and inexpensively, so it decided to move its infrastructure to AWS and deprecate its co-located data center. “We realized that by using AWS, we could have a whole turnkey environment up and running in no time,” says Peter Phillips, ABSI Managing Director. “We didn’t have to spend time or effort building out a new infrastructure, either.”

ABSI uses proprietary algorithms to generate economic simulations that span decades. The company uses the algorithms with Pathwise Modeling Studio to run hedging simulations. Mohammad says, “It’s called a Monte Carlo simulation, and our clients rerun those millions of times with different variables, all in parallel.”

ABSI built a front end on AWS for its processing solution, using Pathwise Modeling Studio to compile a program using its algorithms and automatically running GPU instances on Amazon Elastic Compute Cloud (Amazon EC2) in an Amazon Virtual Private Cloud (Amazon VPC) for security. ABSI uses Amazon Elastic Block Store (Amazon EBS) for persistent storage.

By processing on AWS, recalculating policies takes minutes rather than hours or days. “In regular quarterly financial reporting, it would take you two weeks and a small army of people to complete your regulatory reporting process,” Phillips says. “AWS gives us the computing power to shorten that time to hours and minutes.”

Being able to run these calculations at will means that ABSI’s customers can more effectively judge the risk in their hedging programs for these insurance products , Mohammad says. “When it comes time for our customers to make a trade, they’re going to take the current market information and re-calculate everything at once without any short cuts, and look at the intraday risk information from our system to help them make trading decisions as markets move. With our system, these calculations take minutes, not days or hours. Everything is synchronized with the latest market information, providing customers with situational awareness as market conditions change, which is something legacy solutions cannot provide our clients with today.”

“Using AWS has enabled us to scale our work over hundreds of GPUs very efficiently, and deliver much more granular risk assessments to customers,” Phillips says.

By using AWS, ABSI is able to deliver client solutions more quickly, with richer risk assessments and at a price that enables it to pass on savings to customers. “Being able to use GPUs so seamlessly means we can recalculate as often as we need to. We can run all 5 million policies in minutes, instead of the standard overnight run times,” Phillips says. “We can get a very accurate and unique picture of our customers’ market risk exposure—and there is no other solution that offers better performance at a lower cost for this business.”

Decreasing the amount of time it takes to complete calculations doesn’t just speed things up for ABSI’s customers—it helps transform the way ABSI thinks about those calculations. “Using AWS helps us reduce a 10-day process to 10 minutes. That’s transformative: it broadens our ability to discover,” Phillips says. “We are free to ask a lot more questions of our data now.”

“The beauty is that by using AWS, actuaries don’t need to know how to code. That’s not a productive use of their time,” Phillips says. “They also don’t have to schedule workloads or stand up an infrastructure for something they’ll only use a few times a year. Using AWS is a very efficient way to manage our business and help our clients manage their risk.”

To learn more about how AWS can help you with your performance needs, visit our High Performance Computing details page: http://aws.amazon.com/hpc/.