Bank of Montreal runs risk calculations twice as fast and at 30% less cost with Azure

The Bank of Montreal’s Capital Markets Trading Desk and Market Risk teams closely monitor risk, performing value at risk (VAR) analyses and other calculations after the close of business each day. Doing so ensures that the bank can anticipate risk and maintain stability when faced with any adverse or volatile event. As the clock ticks toward 6:30 AM Eastern Time in Canada, the race is on to get results to regulators before the market opens. Bank of Montreal (BMO) eased the pressure of that daily deadline significantly when it migrated to Microsoft Azure, gaining both efficiency and flexibility.

The next generation of market risk management on Azure

Headquartered in Montreal, BMO provides personal and commercial banking, global markets, and investment banking services to more than 12 million customers across Canada, the United States, and international markets. The 200-year-old financial institution has evolved into a modern, digital-first bank with cloud-native technologies supporting critical processes.

As part of that digitally driven model, BMO, a longtime Microsoft customer, uses a multitude of Microsoft services to help serve its customers and employees. After adopting Microsoft Teams to maintain communication and collaboration early in the COVID-19 pandemic, it moved its 45,000 global employees to Microsoft 365 in 2021. It’s also a large Microsoft Power Platform customer, using Power BI for risk data visualization and reporting.

BMO frequently seeks to apply Microsoft tools to solve problems or create opportunities, which the bank did to supercharge its risk management platform. Over the years, many banks have had to scale their risk platforms to handle more frequent stress tests and run increasingly complex risk models reflective of the changing times. Faced with another increase in demand, the BMO Market Risk team sought a Microsoft solution to address its capacity challenges. “Within Market Risk, we were bringing onboard new businesses and capital markets, like mortgage-backed securities and asset-backed securities, and our ability to generate calculated risk for those products was lagging to where it would take four to six hours nightly,” explains Carl Gomes, Chief Information Officer, Market Risk and Treasury Technology at Bank of Montreal. “We had to think about how we could expand our compute capabilities by adding more compute power and machines to our grid so that we could get to those end results faster.”

The team identified the Microsoft Cloud as its missing piece, adding computing and performance optimizations like Azure Spot Virtual Machines for cost-effective scalability and the use of Bicep to deploy Azure resources. Within a year, BMO migrated and expanded its Market Risk Next Generation (MRNG) platform on Azure. While Azure was initially brought on as a backup cloud service, the bank quickly realized significant performance improvements that moved Azure up to its cloud platform of choice. BMO has also used Microsoft grid computing capabilities to improve market risk analysis and calculation performance and speed, with remarkable cost savings boosted using Azure Hybrid Benefit.

“Extending our risk engine to Azure had a lot of immediate benefits for us,” recalls Gomes. “We gained a lot of flexibility to run our end-of-day batches faster and get our numbers out in a timely manner. Plus, we can stress test additional types of scenarios around what’s happening in the world, such as political conflicts, to help us better manage our business and our capital markets risk.”

Scaling up to 80,000 cores a night and maximizing compute power

Besides being extremely effective for the business, BMO’s MRNG platform can connect back and forth between multiple cloud providers as needed. “When regulators did an assessment on all of the major Canadian banks, they made the uniform recommendation that we can’t have all our eggs in one basket, or in this case, use one cloud service provider,” explains Gomes. “I think this speaks volumes to how Canadian banks operate because we have a very strong focus on how we can always effectively manage risk, whether it’s risk for our external clients or even risk internally. We have to be able to manage our own internal risk because that ultimately benefits our clients as well.” So, BMO uses Azure for the MRNG’s production environment, but it maintains the ability to run it on a second, alternate cloud platform.

Before moving to Azure, the MRNG platform ran up to 35,000 cores nightly as it captured and calculated the bank’s commercial risk based on daily activities, including lending and investing, and prepared a report for the following day. With Azure, BMO will soon be able to scale up to 80,000 cores should job requirements necessitate it, taking advantage of parallel processing with Spot Virtual Machines to efficiently calculate the widest possible range of risk profiles. “Having those extra 45,000 cores means more timely risk metrics on a daily basis and cost savings from decommissioning on-premises infrastructure and licenses, giving us both business and economic value,” says Gomes. With this new strategy, BMO moves leaps and bounds ahead of its early on-premises environment. Limited to only 5,000 cores, it put the bank at higher risk of incomplete or failed jobs, and consequently, higher risk of regulatory fines. Through its Azure environment, BMO has better scalability, a fuller picture of risk assessment, and thus improved risk management.

BMO also moved its compute grid infrastructure, including its Windows Server 2019 operating system and a series of .NET libraries, from on-premises to Azure. The bank uses TIBCO DataSynapse GridServer, available through Azure Marketplace, to manage the distribution of compute tasks across the grid together with VMware GemFire servers on Red Hat Enterprise Linux, which are used in Azure and support memory cache. The team determined the appropriate virtual machine (VM) types based on the types of workloads that BMO would run and then worked with the Azure team on scripting infrastructure as a service components and VMs. Together, the teams enhanced operations in Azure with automation and optimized flexibility with Spot Virtual Machines. “Using Bicep with Azure, we spin up and spin down nodes and switch between VM types, especially spot VM instances, to ensure we have the right amount of compute power needed to run any job,” says Gomes.

Next, the BMO team optimized compute and storage resources in a shift from migration to modernization. “As we started to run, we took an incremental approach where once we got connectivity, we’d rerun one job and look at the performance, then we’d run a few more positions, and then we’d run an entire product or asset class, and then we’d run the entire book,” explains Gomes. “Along the way, we looked at the results and worked with the Azure team to rightsize the infrastructure footprint and make sure we have the right VM types and the right distribution of CPU and storage.”

A billion calculations a night and counting

BMO’s strong risk management culture and practice can continue to flourish now that the bank has gained the flexibility and scalability to meet the demands of a growing business and changing marketplace. As Gomes points out, “If we want to add 10,000 structured deals to our trading or banking books today, we have that ability relatively seamlessly with Azure.”

With its MRNG platform more agile than ever, BMO can foster future business transformation and ongoing customer trust, aligned with the bank’s core commitment to help people at every stage of their financial lives. “I look at the technologies now used by BMO employees, and whether they’re a teller at one of our branches, a market risk analyst looking at VAR and stress bar reports, or a member of our technology support team, they’re doing their jobs effectively because the right Microsoft services are there and available,” says Gomes. “And with those services, we can focus on addressing customer queries and protecting their critical assets.”

Gomes considers the decision to make Microsoft BMO’s primary cloud provider and run the MRNG platform on Azure a wise use of employees’ time and bank resources. “BMO’s use of Azure has ultimately led to a sixfold reduction in overall analysis time, cut overall costs by 30%, and helped us run jobs twice as fast, and we can do more than a billion calculations a night and counting to value all of our capital market positions,” he says. “All things considered, we’re in a much better spot now than a year ago.”

BMO has already developed an interest in using Azure Machine Learning, AI, and high-performance computing to further improve the performance, speed, and accuracy of its models. The bank plans to make its MRNG platform fully cloud native and plans to work with Microsoft to design and build architecture that will support migrating additional capabilities to Azure. “Relationships are only as successful as the level of transparency and partnership that each group brings to the table, and I think we were successful with Microsoft because we had great people on both sides,” he says. “That’s how our two organizations will continue to be successful as we look at all of the Azure tools available to help us reach our goals.”

Find out more about Bank of Montreal on X, Facebook, LinkedIn, and YouTube.