USING IDLE EMPLOYEE AND CUSTOMER COMPUTERS TO ASSIST IN ACTUARIAL MODELING
Publication Date: 2008-Nov-11
The IP.com Prior Art Database
AbstractThe computational requirement for developing insurance and financial risk models is ever increasing. More and more information is being used to determine how to insure, offer loans, or provide financial products to consumers. By using idle computers, insurance and financial organizations can leverage the computational power of many computers. This can reduce the cost of upgrading servers and utilize resources already available within the company.
Using Idle Employee and Customer Computers to Assist in Actuarial Modeling
Insurance and financial companies across the world base their rates and customer risk probabilities on various models. Building, generating, and updating these models are very computationally intensive. Companies spend countless amounts of time and money supporting such models; and the systems necessary to generate and update these models have constantly been upgraded to use faster and more expensive hardware. As a result, these upgrades have driven up the cost of working with these models.
Description of Innovation
Insurance and financial organizations use many computers within the organization; however, these available resources many times are underutilized during non-peak work hours. Typically these computers are left on during the evenings even though they are idle. The reason for this practice is to allow tasks such as updates and security patches to run on those computers. However, since more and more information is being used to determine how to insure, offer loans, or provide financial products to consumers, insurance and financial organizations, by using idle computers, can leverage the computational power of many computers.
This solution proposes using these idle computers in a virtual grid configuration to perform the calculations necessary for model generation and updating. The system will consist of one or multiple server applications plus a desktop application that would
be deployed to computers within the organization. Figure 1 depicts a simple
configuration of the solution.
Figure 1 - Basic Configuration
There may be one or more servers (and many computers in the environment. If there is more than one server in the environment, then one server would act as the master. This master server would delegate portions of the model calculations to the additional or secondary servers. These secondary servers would then parse out calculations to a subset of the computers in the environment (see Figure 2 below). This process would allow for the servers to act in a load balancing fashion, while also allowing the system the capability to handle issues such as network outages or server outages.
Figure 2 - Multiple Server Configuration
The method for processing information is shown in Figure 3. The server that is handling the calculations is called the Model Server (MS). The Model Network (MN) consists of at least one MS and any number of computers in the environment. In this process, the MS determines what calculations are necessary for the model. On each computer there will be installed a piece of software that activates when the computer is idle-- similar to the way a screensaver works. When this software activates the computer, the computer s...