Capital One's public cloud strategy at odds with industry

Financial services companies have access to large budgets and tend to be on the leading edge of technology implementation, but innovation is enveloped in a risk-averse attitude.

Capital One is a unique case. Other firms have turned toward private cloud adoption, where assets are protected by internal firewalls, easily earning regulator approval. Capital One is adopting the public cloud through Amazon Web Services.

The company had a private cloud strategy through the first years of its cloud adoption, but it failed to hit the target, according to Lauren Nelson, principal analyst at Forrester.

In 2015, Capital One unveiled its cloud strategy on the mainstage of AWS re:Invent, showing plans for a massive consolidation of data center real estate. Its strategy centered on operating more like a technology company, weaving in next-generation application development with agile and DevOps.

Some companies find it easier to maintain existing technical real estate, but Capital One concluded it needed to adopt cloud technology to successfully operate in the competitive landscape, according to Bernard Golden, VP of cloud strategy, at Capital One.

Now the company is quite far along in its public cloud adoption and in the process of rewriting applications for a modern architecture so they're more resilient, Golden said, in an interview with CIO Dive. The move will also make the technology more scalable with the ability to update applications multiple times a day.

What's unique about Capital One's approach is that in addition to building new apps, the company wants to move as many of its apps as possible to the cloud.

Lauren Nelson

Principal analyst at Forrester

Companies are hesitant to adopt public cloud because of fears about security and discomfort around multitenancy, said Nelson, in an interview with CIO Dive. Capital One shaped the AWS security as a key reason for its public cloud strategy because the cloud company could update its security technology far faster than a financial services company.

The security trust allows Capital One to focus on the rest of the stack instead.

What's unique about Capital One's approach is that in addition to building new apps, the company wants to move as many of its apps as possible to the cloud, according to Nelson. The company is using the increased security to underlying systems to help justify a huge transition.

That was really a "bold claim in the market and what was so different," said Nelson. Not many companies take that type migration approach today and "none of those are financial services companies." Those making drastic transitions are more often found in retail, manufacturing, or oil and gas.

At the heart of the strategy

In financial services, Capital One's public cloud adoption is an outlier. But public cloud has capabilities not readily found in private cloud instances.

One of the biggest things the public cloud offers is scalability with sophisticated data usage, machine learning and data transformation, Golden said. "We need to be using public clouds for that."

Public cloud capabilities make it easier to work with data, which allows Capital One to focus resources on developing its strategy and honing application deployment.

The cloud provides "greater agility" and speed and allows for common formats around data, said Linda Apsley, VP of data engineering at Capital One, in an interview with CIO Dive.

Apsley's mandate is to make data cloud-ready and become a machine learning company. By that directive her team is responsible for "four D's":

Data movement: Data ingestion coming in and out of the company or moving within Capital One.
Data storage: Capital One has to store a large amount of data in secure, resilient platforms so it can perform as required.
Data discovery: The company has to ensure data is accessible.
Data cleansing: To make sure data is ready to use at scale, it requires deduping, cleansing and getting rid of trailing zeros.

Standardizing across languages, Capital One can use common tools with data across a much larger ecosystem. By extension, data management becomes easier.

For example, a vulnerability scan could occur to make sure data does not have problems when it comes in, Apsley said.

Cloud meets consumer demand

Advanced data management gives way to machine learning, capabilities that are in high-demand. Capital One used ML and data to tackle the issue of fraud.

In the realm of online banking, companies have to sift through incoming data points to determine transactions with attributes of fraud. It requires massive amounts of disparate data with threats changing rapidly.

"When you think about fraud, where is the moment you want a customer to know. You want them to know that moment their card is used, because that saves everybody a lot of pain," Apsley said.

Capital One chose a couple of fraud scenarios and figured out how to build them out, then used the prototype to build out the rest of its infrastructure and ecosystem.

"When you think about fraud, where is the moment you want a customer to know. You want them to know that moment their card is used, because that saves everybody a lot of pain."

Linda Apsley

VP of data engineering at Capital One

The company is in the process of determining the most efficient way to use the cloud in coordination with engineering resources. Similar to building a house then decorating the inside, Capital One is working to develop its data ecosystem, Apsley said.

Her group is tuning it, making data more user friendly so people can easily model on it and get users more comfortable with more cloud tools.

In the future, once the cloud foundation is in place, Capital One will be able to create innovative products on much more flexible platforms, Golden said.

The public cloud also demands talent and personnel with the skills to support development, Golden said.

Capital One has worked to recruit people (a plea CIO Rob Alexander made onstage at re:Invent in 2015) and has also focused on education to ensure it has the skills on board to support application rewrites, modernization and the shift toward cloud-native kinds of architecture and practices.

Cloud strategies of the future

As Capital One pivots from on-prem technology, executives are turning attention to the future of what's next in computing.

Golden's charter is to identify what technologies Capital One should integrate; understand the trajectory of AWS' development and ensure the requirements it has work with the strategy; and engage startups with technology on the forefront of the cloud ecosystem.

Does Capital One's strategy place them at the forefront of financial services? That depends.

Capital One wants to come off as "way ahead, and super forward-thinking," but many banks would take the exact opposite approach, according to Nelson.

"I think the barrier is recognizing that tech is now a critical component of the way that companies need to provide products and go to market and making the commitment to treat tech, the IT, as a core competency rather than a cost center."

Bernard Golden

VP of cloud strategy at Capital One

Capital One struggled with the private cloud for many years, creating redundant costs and efforts that were scrapped entirely, according to Nelson. The company is still working toward data center consolidation it said it would reach in 2018. (Golden said the company will close data centers in the relatively near-term.)

Other financial services companies are making bold moves focused on internal data centers. They're taking, normal, on-prem technology, layering atop a development platform and having developers provision the platforms they know, according to Nelson. That limits the investment to costs incurred from adopting a new development tool.

That strategy allows companies to quickly build new capabilities rather than putting old technology in a new, efficient environment.

There are practical obstacles in cloud adoption, Golden said. One school of thought favors "lift and shift" to take advantage of cost savings. But for that to be successful companies have to work around architecting and invest in talent.

That requires a largest investment from the top down, which often acts as the "biggest obstacle" to cloud adoption, he said. "I think the barrier is recognizing that tech is now a critical component of the way that companies need to provide products and go to market and making the commitment to treat tech, the IT, as a core competency rather than a cost center."

The "right" cloud adoption methodology comes down to where companies are in a "moment of change," said Nelson.

Migrations are motivated by the need to fix an aspect of technology, or because costs will increase if an organization does not do something radically different. In other cases, where hardware service contracts are not yet set to expire, radical migration is not the priority.

While estimates have the only 10-20% of workloads are in the public cloud, technology is shifting away from on-premise strategies.

In the future, more and more workloads are going to end up in public cloud, Nelson said. Change won't happen overnight, but companies will slowly move more and more apps to the public cloud.