Computers Die, Databases Can’t — So Banks, What’s The Plan?

One cannot run a mainframe in the cloud. No one will ever beat the speed of light. Oh, yeah — and no one wants to lose their data to sloppiness, inefficiency and/or theft.

Welcome to three of the most ingrained challenges — “challenges” in a way that mountain ranges can pose problems for pedestrians — of the digital payments and commerce world in 2019. Banks, eCommerce operators and other businesses run dozens or hundreds of databases, which are often maddeningly discrete creatures tied to software, and even hardware, that, in some cases, belong in a museum.

However, the price of cloud computing continues to drop as data analysis technology moves into the realm of machine learning and artificial intelligence — making it ever-more appealing and important to ensure that databases are running at the highest levels possible, and with the best security. With good databases, a company is at risk of running behind competitors when it comes to new products, customer satisfaction and other areas that tend to have major impacts on profit.

Sounds daunting, right? It should.

In a new PYMNTS interview, Karen Webster and Fauna CEO Evan Weaver, along with CMO Dhruv Gupta, dug deep into the challenges and demands associated with better database management, and offered ideas on how to navigate a successful and lucrative path to improvement.

Twitter Inspiration

Weaver’s mission at Fauna is informed by his work at Twitter, where he was the 15th employee, he told Webster, and where he worked as the social media platform’s lead architect, a job that made him responsible for scaling its infrastructure to support the millions of transactions it handles today. Twitter started with a global focus (unlike most banks), but “our biggest problem,” he said, was finding an overall system that would “give us real-time [capabilities] and scale, and flexibility and resilience on the product side.”

The company’s work with banks, he said, has a certain reversed-reflection feel to it.

“As far as payment and financial services providers are concerned, they are experiencing the same problems from the other angle,” he said. “They started with single-system, relational databases that were pretty flexible at the time.” Now, though, the problem for those companies comes down to figuring out how to best scale those databases, while not compromising security or efficiency.

Integration Challenges

Among the biggest problems in the world of payments and financial services providers are how to process different payment types and integrate databases that run on different legacy systems, some of which are living relics of other businesses that have been bought. Those companies must conform to strict regulations that apply to their database operations — among other challenges that make it a daunting project to integrate from legacy systems to cloud-based database operations.

“It does create complexity,” Weaver said, adding that compliance and regulatory issues would serve to prevent many organizations from moving all their database operations to the cloud, even if they wanted to do so. “They are basically stuck,” he explained, indicating those financial and payment services providers. “They are saying that the cloud is super cool, but they still don’t have a system that will let us migrate that core data — the most important products — into the cloud.”

However, even if the idea of a full, relatively clean database integration to the cloud is impossible to achieve for many financial-centric businesses, the importance of finding solid ways to improve database operations can hardly be understated, at least in the view of Weaver. Databases that are not up to the job — tied so closely to legacy systems that are losing steam, and have no more room to grow — can cause system-wide crashes, even if an FI or other business is compliant with regulations about data backup. “If you lose more than one system at a time, there is a knock-off effect,” he said.

Toward Better Databases

So, what’s a forward-minded financial or payment services provider — or eCommerce firm — to do?

Well, it seems obvious, but it’s still important: build a database with the cloud in mind. That means giving up any hope of some massive, on-off database transfer operation. Sure, one will need to move core banking operations to the cloud, but likely in a piecemeal fashion because of the complexities involved.

More specifically, the proper mindset to have about improving database operations and moving them into the cloud — well, at least one part of the pitch that Fauna uses with potential clients — is, in the words of Gupta, to craft a “new system with the same guarantees as the old system. The new system has to offer those guarantees in a way that is operationally similar to the old one. And it must be easy to deploy, easy to scale.”

For Fauna, that means moving businesses to what Weaver calls an “cloud-first operational database,” a purpose-built platform not tied to legacy systems, but is geographically distributed across multiple data centers — which can reduce data latency, a factor that can slow down apps and retail sites, and can keep operating when other parts of the system fail. It means working with distributed ledger technology (DLT), but not blockchain — a middle-ground concept that, according to Weaver, is probably better understood by corporate CIOs than CEOs.

There’s no way around it: Moving databases to the cloud is hard. Eventually, though, there will be little or no choice about doing so. Not only are the mainframe experts getting old, but those machines — no matter how well-maintained and configured — can only do so much.

“You get to the point where the systems … are just fundamentally unsafe to modify, and then your product velocity starts grinding to a halt,” Weaver said. At a certain point, he warned, “you cannot do anything about it. It will be running at its limit, creating a risk that cannot be hedged from the data side.”

Machines die, and so do their caretakers. Yet, every organization must find a way to make sure their databases survive and thrive.