Insurance Technology: Time to Get Your Head in the Clouds

As the spring 2011 conference season winds down, I attended the ACORD/LOMA Insurance Systems Forum in San Diego, CA. ACORD (Association for Cooperative Operations Research and Development) is a global, nonprofit standards development organization serving the insurance industry. LOMA (Life Office Management Association) provides training and education for insurance professionals worldwide. The event is billed as the premier business and technology event for insurance professionals.

One of my goals at ACORD/LOMA was to better understand cloud computing in the insurance industry. There were several sessions that touched on cloud and Software-as-a-Service (SaaS). One of the most interesting was “Cloud Computing for Insurers: Time to Get Your Head in the Clouds” by Bob Hirsch, Director Technology Strategy and Architecture, Deloitte Consulting LLP. Bob provided some interesting thoughts on why cloud isn’t more prevalent in insurance. One reason is that cloud vendors have been slow to meet the regulatory demands of insurance. Another is that vendors are not in the “core” space–most cloud implementations are at the “edge for specific workloads.”

Insurance firms also have concerns about data loss, security and privacy, audit and assurance, backup and disaster recovery, vendor “lock in”, and IT organizational readiness. Bob described vendor “lock in” as the inability to easily migrate your company’s information from the cloud provider’s data center to your own if you decide to bring processing back in-house.

Bob suggested that with quality datasets, computing advances and maturing tools, analytics could become a strateStatistics Image gic cornerstone of the enterprise.  As an example, he talked about the cost savings from moving volatile computing needs to the cloud. Bob explained that insurance companies need to run stochastic models each quarter to estimate risk. Large insurers are running grids of 2500 nodes and growing for this type of computing. Running the models can take 24 to 48 hours, but the rest of the time the servers are idle.  Bob stated that current grid systems can be modified to be cloud aware and “burst” capacity to clouds as needed by storing the grid image in the cloud and deploying it across servers as needed for periods of peak demand.

Bob also walked through a cost/benefit analysis for Monte Carlo simulations for hedge funds which have limited in-house IT resources. The analysis showed in-house monthly costs of $14,280 vs. $6,930 for cloud, a 51% savings.

For the moment, Bob said that smaller insurance firms are ahead of larger ones with using cloud-based applications.  This is because insurance systems are very fragmented within larger organizations and they are slow to consolidate systems across the enterprise.