The financial services industry has long required compute-intensive calculations for risk management, regulatory compliance, product development, pricing, trading, clearing and surveillance. However, the volume and timeliness of these calculations have increased.
This is where cloud comes in.
Financial institutions (FIs) are now expanding their on-premises grid computing capabilities and taking advantage of the storage capabilities that the cloud offers, in addition to agile processing, lower costs and greater capacity.
Finextra spoke to Mike Kennedy, senior director, engineering and David Welch, engineering manager at TIBCO about its recent report ‘A Cost-Effective Solution to New FRTB Computing Requirements’ and the benefits of running grids at enterprise scale, processing simultaneous multiple workloads and reducing hardware footprints with the cloud.
The possibilities of ‘what if’
FIs are exploring cost-efficient ways to expand their on-premises grid computing capabilities as a number of industry factors are increasing demand for real-time, high-performance computing. As financial calculations for risk analysis, trading simulations and regulatory reporting become more onerous and unpredictable, FIs need a way to right-size resources on demand.
“Imagine being able to build a virtual super-computer in minutes, without an eight-figure investment, and paying only for what you use,” Kennedy says. “As a proof of concept, TIBCO spun up a grid of 1.3 million cores on AWS that provisioned in less than 30 minutes. Financial organisations are looking for this kind of scale and speed to meet day-to-day requirements as well as support unexpected fluctuations.”
Some FIs rely on cloud bursting, a hybrid configuration where they run applications on-premises but can leverage extra capacity when needed, quickly pulling resources from the cloud. Elastic computing eliminates the need to plan and provision for peak usage.
The emergence of artificial intelligence (AI) and machine learning (ML) is also simplifying and automating the operational tasks of running compute. For instance, ML models can help detect usage patterns and even anticipate market swings that may impact demand.
“You can never have a computer that is big enough, fast enough, or smart enough,” Kennedy says. “Financial organisations can always think of more stuff to do to address a range of needs from business requirements to regulatory obligations.”
Addressing current limitations
On-premises grids typically force FIs to make difficult decisions about where to allocate compute resources and slow their ability to move to a more agile development model.
Kennedy explains: “Limited capacity can impact the bottom line - for instance, preventing purchase-sell decisions or even stalling organisational innovation. We tend to see a 70:30 breakdown where the majority of compute is used for day-to-day data crunching, which leaves little left over for strategic initiatives.”
In addition, many FIs are still working in a waterfall development model even as they seek to move toward a more modern, agile development paradigm. Migrating to cloud means enabling development and testing to be done at the same scale as production, eliminating backlogs.
“Organisations want continuous integration,” Kennedy says. “Another advantage of elastic computing is you can run an application at scale without having to wait in line behind someone else who is on-premises.”
But before expanding capacity in the cloud, it is important for FIs to have a clear understanding of their requirements and controls. Welch describes a risk-return relationship that FIs must carefully evaluate.
“What are they required to run? Are they encrypting data on disk? What are the internal processes they have to accommodate? These are just some of the many factors that determine how institutions can deploy out to a cloud service,” Welch emphasises. “It’s important to be thoughtful before starting down the path, so they can be better prepared to deal with any issues that may occur instead of having to re-engineer the entire strategy.”
Preparing for FRTB
Achieving that level of planning becomes more time-sensitive when new or updated regulations arise. For instance, the Fundamental Review of the Trading Book (FRTB) regulation, developed by the Basel Committee on Banking Supervision, is due to be implemented on January 1, 2022 following a revision. FRTB will require significant transformation to the infrastructure that
FIs rely on to access their market risk and the use of more granular risk factors that expand the range of scenarios and historic data that are needed to calculate exposures.
Kennedy highlights that as a result of FRTB, “Some banks are predicting that they might need five to eight times the compute power to run new, more detailed kinds of models. They might not need to run them every day of the month, but due to how mission critical these applications are, they are over-provisioning on-premises grids.”
But there is another way, and FIs are reaching a point where they realise cloud providers can run their data centres better. As the FRTB deadline looms, FIs should consider solutions that automatically burst to cloud resources on demand before investing in time- and cost-intensive overhauls of their existing infrastructure.
“It’s also important to remember that it won’t end with FRTB,” Welch points out. “The farther out in front of change institutions can be, the better off they’ll be. Cloud deployment is a different compute model that can help them adapt more quickly, and they need to understand the model in order to make the best decisions for their organisations.”