David Reilly Reveals Bank of America's 3 Major Tech Goals

When Bank of America poached Morgan Stanley's CIO David Reilly in May 2011 to run infrastructure and wealth management technology, the hire made a lot of sense. The bank was looking to streamline its enormous, sprawling IT network and Reilly had proven himself to be a forward-thinking leader in cloud computing and data center efficiency - two areas of critical importance to all banks that rely on in-house technology.

We spent time last week with Reilly, who is now technology infrastructure executive at Bank of America. He reports directly to Catherine Bessant, global technology and operations executive, who reports directly to CEO Brian Moynihan. Reilly shared three major technology initiatives he's overseeing at the megabank.

1. Adopting open, software-defined infrastructure.

Reilly is a big fan of the trend toward software-defined computing, which starts with software-defined networks that can be programmed using ordinary software on commodity servers with standard operating systems. Such networks are more flexible, at least in theory, than traditional networks. Users can make their own updates without waiting for their vendor. They can bring in high-performance technology without having to change infrastructure in which they have already invested.

"Why stop there?" Reilly says. "Why can't the application come along and present almost a manifest to say I need this much compute, this much storage, this much firewall protection, this much geographic diversity, and I want to provision it in real time? That, we think, is the end state, where the application drives the shape of the infrastructure. Configuration management becomes a much more solvable problem because it's all done in software. You can't buy this today."

A few vendors offer software that in their words fits much of the above description, but these are smaller, less influential companies.

"It's coming in networking, and you're seeing it in the software-defined networking space, it's very emerging with compute and storage, but we see it's the future," Reilly says.

In an interesting experiment, Bank of America is creating two software-defined technology "stacks" — one completely open, the other more proprietary. By the end of the year, the bank plans to run 200 workloads on these stacks. To minimize risk, these workloads will be things like application development and testing and artificially generated workloads.

"That's not many, but it's a start," he says. Assuming all goes well, the bank will add 7,000 to 8,000 workloads next year.

"It's still a research project," Reilly says. "But we think as we're constantly making what we've got better, the future will look very different. The infrastructure will become more of a platform that applications can take advantage of and shape. You're no longer dependent so much on individual engineering capabilities to build what you need."

Applications may need to be rewritten to take advantage of such a new world, Reilly notes. "But it's what the applications have always wanted — why can't my app be the thing that governs the shape and performance of the infrastructure? We think it's a three to five year journey. All the research we've done in the community tells us we're right."

Reilly has a word of caution for large technology vendors that have begun working on their own, proprietary versions of software-defined and open networking.

"We believe that the proprietary lock-in is not the way to go," he says. "We should work with you as a partner because you have the best service, the best price, the best capability, the lowest risk, not because it's impossible to get off you once we're on."

There's comfort in going with a large partner. "But if you end up being locked in in a different way, maybe you're no better off," Reilly says.

The definition of "open" is going to be key, in Reilly's view. "A lot of our tech partners are going to struggle with this," he says. "Is the definition of open that it works on everybody's stuff first, or is the definition of open that it works on my stuff first, then your stuff second? As an industry, I think that's the cusp we're on, it's almost a fork in the road."

Software-defined infrastructure will provide elasticity, he believes. "One of the main tenets of Cathy [Bessant]'s philosophy is elasticity in the technology footprint," Reilly says. "We have to be able to grow and shrink, and our partners have to be ready to accommodate. The more we can produce through this software-defined world, the better."

As the bank starts to adopt a software-driven model, it's also made organizational changes. The infrastructure technology executives at Bank of America now have dual roles. "The executive who faces the consumer business and delivers the infrastructure the consumer business needs, happens to run the mainframe and the ops team for the whole company," Reilly says. The executive who supports the payments team also runs storage and data centers for the entire bank. The idea, which was Bessant's, is to have no separation between the enterprise and the individual lines of business. "The closer we can make those lines of communication, the less of that goes on," Reilly says.

"We need very accomplished executives to fill those roles because you've got to be able to live in both those worlds," he says. "We don't see anyone else with that model, and we think particularly at our scale, it's important that we don't lose sight of the lines of business. It would be very easy for us to fall into the trap of being this off-in-the-distance technology provider, that would be terrible for us. Staying close to the business, keeping the enterprise focused on not just technology for the sake of technology, but with an end in mind for the business.

"I often see my peer group struggle with that: I can't get the investment I need, nobody understands me, all of that," he says. "The proximity we've created through Cathy's model has minimized that risk."

2. Streamlining the data centers, driven by industry benchmarks.

One of Reilly's projects at the bank is ongoing modernization of the data centers. "We refresh our data center strategy annually," he notes. "That might sound a bit frequent, but we see a lot of change in that area that's enabling much more elastic compute models."

The bank shut down nine data centers last year and it's closing three more this year, the third this month. "Nine in one year is a lot, and they were all on time," Reilly says.

Such IT "rationalization" is driven by benchmarks.

Bank of America benchmarks its entire infrastructure "ruthlessly" every six months, Reilly says, to keep costs down.

The benchmarking is mainly about price - how the bank's cost for network ports, servers, storage, desktops, etc. compares to its competitors'. Gartner provides tables of efficiency leaders that show each bank where it is in the lead table, but not specifically who is above and below it.

By refreshing every six months, the bank can see how it's trending in efficiency. "The goal would be, for every product we've got, to be best in class," Reilly says.

Along with price, the bank also closely watches quality, service levels, risk and competitive differentiation.

This year for the first time, the bank's storage expense was down, although storage needs didn't shrink. "We were able to deliver that at a lower price point, and we were quite proud of that," Reilly says. The bank also "rationalized" its carrier footprint this year and changed its network sourcing model, hiring 282 network professionals to bring that expertise back into the bank. Some of these were veterans.

"Engineering and architecture expertise we have to have close," Reilly explains. "You can't drive that next generation of networking in a nonproprietary, open fashion without expertise, and we needed to truly do that. The size of network we run is this close to carrier size so we've got to have that capability."

Within the data centers, the bank seeks to simplify, removing what Reilly calls "valueless complexity."

"We're big believers in the modular data center, where the outside looks like a shipping container, and inside is all the mechanical and electrical, fully integrated with the technology," he says. The bank sees using such containers, which are sometimes called data center pods or "data center in a box," in its own real estate, in on-prem colocation agreements and in off-prem colocation.

Bank of America is working with a couple of small organizations in this space. "We think we shouldn't be in the engineering business if someone has a commodity platform" that will work well and scale up and down, Reilly says.

Three of these containers are already installed in one of the bank's Midwest data centers to house IT equipment.

"It's our first foray into it; we deployed the three modules in the first quarter of this year and we're learning," Reilly says. "It's a change for us, it's important we don't make the new thing look like the old thing, make this container look like a traditional data center. We avoid that by working in close partnerships with smaller organizations that help us realize the full potential of this modular design. You never go inside it. You never need to. It's completely unmanned, you plug in electricity and network, and then it's yours to shape."

The modular data center will turn the idea of a data center idea on its head, Reilly says. "You can deploy much faster, in much smaller units," and energy costs can be saved because the bank is only cooling or heating smaller contained areas rather than an entire cavernous building.

Power unit efficiency for containers tends to be about 1.1, whereas the typical data center runs at an energy hogging 2 or 2.2.

"We have a major role to play in lessening the bank's environmental impact," Reilly says. "Data centers are a major consumer of electricity." Modular data centers are also cheaper, much more flexible and elastic, he says.

3. Driving innovation.

Part of Reilly's job is to drive innovation in the bank. "In these jobs it's far too easy to fall back on the large, very well established tech partners, who we love and rely on heavily to say, tell me what's going on your labs?" he says. "That's got its place."

One way innovation is fostered at Bank of America is its annual technology innovation summit, which was recently held for the fifth time in Silicon Valley. Reilly and 120 bank executives attended, both to share with the emerging technology, investor and venture capital community the problems the bank will face in the next couple of years and to see new solution providers demonstrate their wares.

"Sometimes the solutions are very immature, sometimes they're just slides, sometimes they're working prototypes, sometimes they're full-fledged products," he says.

This year, 641 people attended, up 18% from last year, and 45 start-up technology companies presented to bank executives. Over the first four years of the summit, 15% of presenting companies have gone on to be vendors of the bank.

"It's a hit rate we're proud of," Reilly says. "That simply would not happen if we didn't get off our backside, it forces us to think about this a different way, and be out there with the vendor and venture capital community. The venture community wants to know, what's an enterprise like Bank of America faced with? Where should we be investing? What are the kinds of companies that if we could find, would be of interest to you?"

A small tech partnership development group at the bank is led from the West Coast and works with fintech startups all year round on their businesses and on working with Bank of America. "As you can imagine, it's pretty daunting for a 10-20 person company to get into the bank," Reilly says.

Bank of America is also involved in Accenture's FinTech Innovation Lab, where every year senior bankers at large New York banks coach promising startups. The bank hopes to host the demo event next year. 

Two companies - Dashlane and Inktank, caught Reilly's eye at this year's event. "We think both of those companies will go on to be vendors of some kind," he says.