Fully transitioning to the cloud is never easy. Just ask Capital One’s Anne Johnston, VP of cloud costs and engagement. As one of the first banks to transition to a cloud platform from an on-premises data center, Capital One had some real challenges. Johnston told IT Brew that like any big corporate project, it’s important to keep morale up during the migration process.
“Celebrate along the way,” said Johnston. “It’ll help folks really see the outcomes that they’re achieving, and also help you with the momentum along the way.”
Johnston talked to IT Brew about Capital One’s cloud transition, cutting costs, and lessons learned.
This conversation has been edited for length and clarity.
Tell us why Capital One decided to move to the cloud.
What inspired our move to the cloud was really thinking about what our customers expected of us. And what they expected was tech forward, real-time experiences. And in order to do that, we had to rethink how we thought about our technology. And the cloud became very clear to us. It enables us to scale our data environment in order to run machine learning in a way that our data centers just wouldn’t allow. It allows us to have faster, more reliable deployments to produce features for our customers so much faster than we could in a data center.
What’s been the most surprising thing about your transition to the cloud?
I don’t think I quite thought about how fast we could iterate and how fast we could get features out. The speed at which we can now deliver software, that’s been something that’s amazing. And our customers get the benefit of that and our engineers also love to move fast. You know, when you think about a bank, financial institutions, a lot of us think about legacy. But when our engineers come to Capital One, they often find that we move quite a bit faster than they expected. And it’s because we’ve built an environment that allows them to do that, that allows them the freedom to develop and deploy in a safe way all of the great features that they hoped they’d be able to deliver.
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
The cloud is expensive. What are some of the ways you cut cloud costs?
We actually have a centralized FinOps team. And that team does a few different things. It focuses on architecture and standards. It focuses on providing tooling for engineers, so they actually know how much their applications cost. It focuses on analytics and modeling, so it does some machine learning to detect anomalies before it’s too late. And they also focus on our tech culture in terms of being aware of spend. [It’s just] part of our DNA, knowing what you’re spending and optimizing that spend. So, that team is completely focused on our cloud costs and understanding them, making sense of them to all of our engineers, so that there’s general awareness as to how much their infrastructure costs and what they can do about it.
What lessons would you give someone who’s embarking on a transition to the cloud?
When you’re working through milestones for the transitions to the cloud, it’s important to think about those big milestones—closing the data centers, migrating the applications. But think about milestones in terms much more granular. And think about milestones for an application, a business unit, and so on. Migration to the cloud, especially for someone at an enterprise scale, takes time, and with time, you need to be able to sustain momentum. So, from a lessons learned standpoint, thinking about what the right granularity of milestones are so you can sustain that momentum and keep the foot on the gas to do your full migration.
The other lesson learned, I think, is to think a little bit about strategy in terms of lift and shift versus modernization. When we were moving to the cloud, it was super important to us that we got out of our data centers. If you have the time, think about modernizing your infrastructure, like moving to serverless. Whether that be serverless compute [or] serverless databases, it’s important to think about that because it is never a good time to modernize your architecture until it’s too late.