Google Leverages Kubernetes to Help Banks with Risk Models

Google says it is using Kubernetes and other open-source technologies with a number of financial institutions to help them accelerate their risk calculations.

Adrian Poole, head of financial services for Google Cloud, says investment banks are concerned about how to scale up their computing resources for risk management as deadlines are approaching for the Fundamental Review of the Trading Book (FRTB). 

FRTB, an initiative by the Basel Committee on Banking Supervision (BCBS), introduces a set of standards for new market risk capital requirements for banks. The date for the implementation of the regulation is 2022, with reporting expected to begin sooner as lawmakers work on implementing the standard as regulation in their respective jurisdictions.

“We are working with a number of clients where we are using things like Kubernetes and some of our open-source technologies to really accelerate how they approach some of their risk and pricing grids,” Poole says.

One method of tackling the large computing capacity required for risk analytics is to use grid computing. In such a distributed architecture, each computer’s resources are shared with other computers in the system in order to solve complex problems.

FRTB requires computationally-intensive calculations, such as expected shortfall, capital charges for each trading desk, and various risk factors. Banking clients are looking to utilize the Google Cloud Platform (GCP) to run these grids, using a combination of the platform’s managed offering for Kubernetes called Google Kubernetes Engine (GKE) and DataFlow, a managed service for overseeing and transforming data in real-time and batch modes. 

Kubernetes is an open-source platform for running containerized applications that Google engineers developed and released in 2014. In 2015, the firm donated Kubernetes to the Cloud Native Computing Foundation, a part of the Linux Foundation, that runs large open-source developer conferences. Poole says the benefit of being part of the open-source community is that development is sped up and the market is able to develop the technology themselves.  

Kubernetes can be deployed on-premise or with one of the major cloud providers, such as Amazon Web Services (AWS), Microsoft Azure and Google Cloud. Poole says that Google tries to make offerings that don’t lock in users to one service provider so that they can choose alternatives. This works well for financial services, where firms have to have flexibility around meeting regulatory requirements, he adds.

A major announcement for Google this year was the development of its Anthos platform, which allows users to build hybrid applications on-premise or in the public cloud.  

Poole says work on interoperability with these other providers is ongoing, and the company is currently working on making their service available on AWS.