Credit value adjustment risk calculation is a common example of a massive, parallelized computation system in the financial services industry. Computing costs are high but necessary, considering the importance of the calculations from both a financial risk and regulatory perspective. However, distributed, cloud-based systems can change that by reducing overall costs with a flexible, pay-per-use model that avoids up-front spending on resources that are not used 24/7.
This reference architecture whitepaper discusses why a cost-effective, “straight-through-processing” credit value adjustment risk calculation system is a worthy pursuit for financial services companies. A combination of in-memory storage, stream processing, and distributed computing serves as a model for a large-scale calculation that involves tens of thousands of CPU cores.
While the use case in this paper focuses on risk calculation, this architecture and the associated technologies apply to other massive deployments in a variety of industries.