Case study: How two financial titans are modernizing data center infrastructure

Like all other modern companies, financial institutions are data-driven. 

But because of their unique risks and compliance requirements, they handle and store data differently. 

So what does a modern data center look like in the financial industry? And how are these organizations modernizing their data collection, storage and analysis?

Data and analytics execs at two major multinational financial companies sat down with VentureBeat to discuss their ongoing digital transformations. 

Mastercard: A push to master data

Data strategy must always go hand in hand with business strategy, said JoAnn Stonier, chief data officer at Mastercard — there’s really no point in having a data strategy if it’s not achieving business goals. 

For a multinational company operating in more than 210 markets and processing more than 125 billion annual transactions, this is especially critical. 

“We are a network of networks now,” said Stonier, pointing out that Mastercard is known for its flagship credit, debit and prepaid products, but has greatly expanded further into debit, ACH and loyalty markets.

Given the sensitivity of the regulated information it deals with on a moment-to-moment basis, Mastercard must continually invest in its own secure data center, Stonier explained. At the same time, the company’s tech stack modernization must drive towards modularity and scalability and be able leverage more commodity infrastructure and technologies. 

This translates to a continued migration to multicloud infrastructure — where appropriate — balanced with maintaining physical warehouses. With security paramount, Stonier’s team analyzes which applications should stay in-house and which the company can afford to put out on the private cloud, she said.

“Because we have to be more nimble, we have to be able to process more data in a global way that is also more effective and more efficient for us,” said Stonier.

Speed, consistency, security

As she pointed out, customers expect “instantaneous transaction processing.” Catering to this, Mastercard must structure its stack so that it can process data more quickly and securely. 

Also, data quality is so much more important than volume, she emphasized; when the company purchases data sources from third parties, it is very careful about “the veracity, the accuracy, the completeness, the consistency” of that data.

“It is also important to determine what information is appropriate for the task,” she said. “What’s the use case? What’s the problem you’re trying to solve?”

The company intends to use more AI, machine learning (ML) and data analytics tools to drive insights that will help craft its next generation of products and solutions, said Stonier. 

All this while security risks continue to go up. “We have to really look at the bad actors and keep the barbarians at the gate, if you will,” she said.

In doing so, Mastercard operates a security operations center (SOC), which monitors its services 24-7.

“That affords us the opportunity to really put security first,” said Stonier.

People are now the data ‘center’

She pointed out that those of a “certain vintage” remember the data center as “racks and racks of servers that took up a significant amount of real estate,” with language around them including “castles, farms, moats.”

Clearly, today, organizations do not need those types of server farms to enable technology. Now instead, it’s people who have access to data who are that “center.”

“Those people now are the backbone of our innovation,” said Stonier. 

Mastercard’s is much more of a command center operation, with people with access to faster technology and data innovating in ways that simply weren’t possible 15 or 20 years ago. 

Also, in ensuring consistency in its strategy across the business, Mastercard has a team of data strategy leaders and subject matter experts who are “federated” throughout the firm and work alongside business teams.

The company continues to evolve its “network of the future” based on all the different types of data that will need to be processed, as well as the need for data provenance and data lineage, Stonier said.

In the end, Mastercard has several goals in mind. “We need to get closer to our customers. We need to make sure that we’re providing the right services. We need to be innovating faster. We need to be taking advantage of the tools that are out there.”

Citi: Driving at more data insights

Data sprawl is a problem for everyone — and Citi is no exception. The company tackles the issue on a day-to-day basis, says Promiti Dutta, head of Citi’s U.S. personal bank analytics team. 

The financial institution has been dedicated to wrangling a massive sprawl of data across multiple legacy systems into a centralized data hub, she explained. The company is focused on making sure that its tech stack remains as lean as possible — so that its data footprint is as lean as possible, too.

First and foremost, Dutta pointed out, Citi has been able to shrink its footprint by understanding the extent of its data copying. While the company’s data largely resides on-premises, it does use analytics SaaS in the cloud, particularly as more external data becomes available from key providers such as credit bureaus Experian, FICO and TransUnion. 

Going forward, on-premises tools will continue to be at Citi’s core, said Dutta. The company doesn’t see itself copying everything to the cloud. Its focus will be a hybrid strategy, and they will copy what they need and find different tools that enable them to safely do that.

A holistic view

As data will always evolve, data-driven organizations must feed forward, said Dutta. 

The company has centralized all of its customer data into a dataset providing a holistic customer view across all products, channels and interactions.

Citi’s next big challenge, she said, is using unstructured data to gain even more customer insights. The company also plans to employ advanced statistical and AI methods. A key priority is having data to drive decisions and outcomes, whether internal or those impacting customers.

Driving efficiencies and a data-first culture

In seeking further efficiencies and reduced storage and licensing costs, Citi is taking advantage of open-source tools. The company uses PySpark and Python, among others.

Also, Citi is doubling down on a data-first culture. For example, the company has built its own in-house analytics tool. This “correlation hub” includes a no-code/low-code self-service search engine for all of its data, and is accessible to all employees, said Dutta.

Users can ask questions about any cataloged data that Citi has, she explained. This allows the company to get a response in the hands of its business partners so they can start using data themselves.

Ultimately, it’s a lifecycle of continuous improvement that is always evolving, she said.