How Sentara moved Epic to the cloud, saved millions and launched a spinoff

Sentara Health, a non-profit health system based in Norfolk, Virginia, that operates multiple hospitals, medical groups and other care facilities, decided to migrate its EHR systems to the cloud – specifically, Epic on Microsoft Azure.

Jeff Thomas is senior vice president and chief technology officer at Sentara Health. He describes himself as not your traditional healthcare IT professional.

He came into healthcare seven years ago and was brought into Sentara Health specifically to help get the organization into the cloud. This was a board-level goal – and definitely the CIO goal. How does Sentara enable the cloud in its environment? The challenges were around capacity limits and days of procurement.

But first, a pandemic

“All of this became very important because of what happened after we started this journey: We were in the middle of the cloud journey when COVID hit, so fortunately we had the ability to scale during COVID,” Thomas recalled. 

“And when I say scale, we were able to scale our front-end and everything else in some interesting ways. If we hadn’t put the cloud in place, we wouldn’t have been able to scale.

“It would have cost us millions of dollars and we would have waited a long time to get things,” he continued. “We were able to do a lot of stuff very quickly because we already had about 60% of our infrastructure in the cloud for computing, for front-end with Windows, virtual desktops, those kinds of things – and the capacity to shift 8,000 people home.”

Further, Sentara’s data was very siloed and locked in a data center.

“So, when we looked at moving toward this data-centric model in healthcare, our data was not close to those people who needed to consume it,” Thomas explained. “It was locked in a data center sitting in a hospital. To be able to do the analytics, you had to go back to a physical data center.

“That’s not even close to where those users are because those users are coming in through the cloud, through the Internet,” he continued. “It was getting that into a platform that’s elastic, that allows us to grow. That was an issue. We were not able to get consistency with that data center.”

What do data centers truly cost?

Sentara did modeling to look at true data center costs: buildings, electricity, refresh of systems and much more. So, for example, if the health system needed to do full disaster recovery, not an emergency blanket style disaster recovery, Thomas started factoring those costs in to figure out the capital consumption.

“What we were finding is that about one-third to two-thirds of what we had purchased was either not used in our data centers or was underutilized or it was just going to other applications,” he said. “And then we hit this tipping point where we have to make these large purchases and that cost should have been put against some of these smaller applications – they never got implemented or never got code against them. So, everybody thought it was cheap in some cases but expensive in others.

“And then that capital refresh limited our ability to innovate because we were then having to either wait for the capital refresh or we were not able to consume some of the new capacities because we had to wait until we got more infrastructure in to handle it,” he added.

Evaluating all these factors, Thomas and his team dug into the concept of elastic scalability.

“How do we automate that and change that demand model?” he explained. “How do we unify in our data platform, giving us real-time models we knew we were going to need for EHR data? How can we get to more of an operational expenditure-driven cost? I can model out and predict my operational expenditure costs because I actually am building that as part of the architecture process.

“And then it allowed us to really redo our security model that was locked into a data center that grew organically,” he continued. “We then built and patented a process for automated deployment into the cloud that gave us a model that gives us HIPAA compliance, zero trust, encryption at rest, and automated deployment that locks down the environment where things get deployed.”

Just what the doctor ordered

All of that added up to what Sentara was looking for – instant scalability, and the ability to centralize and automate data services, reduce capital spend, and strengthen security and compliance.

“The EHR is not just the EHR,” Thomas stated. “It’s all those secondary and tertiary environments that live off of that. It’s the ability to scale those out as necessary and make decisions that allow us to better leverage and harvest the data that’s coming from those systems to make other clinical decisions or to improve the ability of people to make clinical decisions.

“So, we began the move – and a cloud move is not a lift-and-shift, it is a re-engineering,” he continued. “We used a strong process of implementing engineering and architecture. Every single one of our systems went through a transformation process. We re-architected them, we documented how they work, we upgraded where needed.”

Sometimes the team moved systems into what was SaaS or infrastructure-as-cloud. Sometimes it moved systems straight into an SaaS platform and did not ever go to the cloud because the team may have selected a new platform that was more conducive to what staff needed to do. The team took about 1,300 applications and scaled them down to about 600. As part of this transformation, the team also re-architected the way the systems work.

“We had systems that were 90 to 95 servers that would drop in cloud to a 45-server environment,” Thomas explained. “The size of those environments were reduced to meet their demands. We also looked at innovative technologies that allowed us to reduce the cost of operating.

“We were using new technologies and capabilities both from a security and an architecture standpoint that were being developed in the cloud – not being developed in an on-premise model,” he continued. “We were able to architect and drive that through.”

A big change

That changed how IT operated.

“Our operating model now is focused on engineering and architecture and cost optimization and re-architecting,” Thomas explained. “Every system we have comes through that process. So that doesn’t just reduce my spend, it actually allows us to budget and tell you what the spend is at the front of the project because we engineer and architect it. And I can give you a monthly cost and a yearly cost to run a system.

“And now you’re going to make a different business decision because now that system that looks cheap may not be as cheap because what you were previously doing was you used to be able to piggyback off other purchases you made and say, ‘Oh, this application’s cheap,'” he continued. “Because I spent a whole lot of money on an EHR system or a PAC system or some other system. And I’m basically cannibalizing some of that capacity to put this system in to make it cheap or more cost-effective.”

In reality, that’s not making it more cost-effective – it’s actually reducing what was originally made, he added.

“And it blurs the lines of the cost of the decisions business makes,” he said. “So now we look at those costs in that realm. We’re also able to optimize what’s there. We optimize constantly. We look at new opportunities of how we back up, how we maintain data, how we engineer and support data in that environment.

“It also forces us to be a lot more mindful with our demands because we are now thinking about it like the business does,” he continued. “So, in the business world, in a hospital, if you have more patients, you’re going to use more products, so you have to understand what that cost looks like and then you’re going to start to make business decisions differently.”

The same needs to be done in the IT world, he added.

Looking at IT through a business lens

“If they want more demand or we need to do more compute, we can show them what that demand looks like in a business sense,” Thomas explained. “As opposed to just saying, oh, I need more storage, I can say, here’s the dollar figure it costs for you to do this.

“So now I’m going to start talking in business terms, in terms of cost,” he continued. “I’m talking in business terms and how this works. So that changes how I look at my cost model. It also changes how I make architectural decisions. I start looking at things differently.”

This enables the team to move into the right space, he added. From how one operates now from a business standpoint, it gives the team a lot of elasticity on data, he said.

“I’m able to really spin up data and to really allow those data teams and the data analytics guys to really start using their data in a more consistent way,” he explained. “Because there’s a lot of new tool sets being built in the cloud that enable us to do things in a more dynamic way and start going into real-time analysis and consumption. It also puts our data in a better place to show what happens or to be able to access that data for our consumers.

“When I start building applications, I’m not having to replicate that data in the same way,” he continued. “Because I’m able to have these data lakes, I’m able to use some of the tools out there that allow me to do better analysis and also to quickly scale up new capabilities with that data.”

A powerful new tool for the tool belt

Moving forward, the ability to deploy systems in an automated fashion and drive economies of scale through automation, through scalability, then gives the IT team a big tool in its tool belt to address business issues, he said.

“And we saw this in spades when we spun up 8,000 knowledge workers to their home machines off of Windows virtual desktops where I piped cloud on premise Avaya call center to people for COVID,” Thomas noted. “I piped all their stuff to them at home until we could get the laptops. And I was able to do that at a fraction of the cost without any demand. And I didn’t in any way impact my clinical operations doing that.

“And I could scale it out,” he continued. “In the cloud it changes the way you function in terms of architecting. So, we architect everything. Nothing gets deployed. You don’t have a person who can just go out and put a server in. We’re going to just throw a server in here. It actually goes through an architecting process. It gets automated deployment, all the security gets applied to it and so it’s very quick.”

Further, the team can architect in real time because it can deploy and pull back and re-engineer in real time, he added.

Millions in savings

“So, I now bring my DevOps, my SecOps, I bring my architects all together in a room and we discuss these things and we in real time build and that changes a lot on how we do things,” Thomas said. “As part of the original business case, based off that baseline, with the systems we had at the time when we started doing the transition and at the end of that, we have verifiable about $2.5 million in annual cost savings through rights that allowed us a three-year period. So, it’s about $7-8 million savings total over a three-year period.

“This year we’re going to get close to 2.5 million right-sizing opportunities, new capabilities, reserved instance commitments,” he continued. “How do we auto-scale, auto-shutdown environments, how do we do these types of things? As we’re growing out new capabilities, I’m reducing my cost. So, in a sense I actually have a flat cost as I’m innovating and getting new capabilities.”

Out of all of this work, Sentara took all of its learnings and spun them out into a company called OptaFi. OptaFi offers best practices and technical capabilities for moving Epic EHR workloads from on-premise to the cloud.

“OptaFi is a joint venture between Infinite Computers, which has a history of managed services and large-scale support, and Sentara’s experience with movement of healthcare to the cloud,” Thomas said. “Not just the infrastructure, but the organizational transition and the upskilling – OptaFi is built by healthcare for healthcare.

“This is a company going out and helping CIOs and CTOs go through this transition process,” he continued. “The learnings we have that make not just getting to the cloud but running effectively in the cloud work over time.”

Follow Bill’s HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.

WATCH NOW: Why you can’t just dump AI on a CIO or CTO to make a Chief AI Officer