Servers and storage are a primary focus for one hospital’s support upgrades.
What will next-generation data centers look like? As healthcare organizations amass unprecedented volumes of data that need to be stored, processed, managed, protected and analyzed, IT leaders must confront the challenge of transforming — and optimizing — data-center infrastructure. Hospitals seeking to meet population health goals, generate data-driven insights and improve patient outcomes require robust, scalable and agile infrastructure that provides essential data security, availability and reliability.
What are the best strategies for ensuring your data center supports your organizational goals? What is the impact of the healthcare Internet of Things? We talked to Scott Dresen, senior vice president for information services at Spectrum Health, who weighed in about how his organization tackles these challenges and shared the measurable benefits gained from building a new modular data center.
Dresen and his 225-person team are responsible for IT infrastructure and operations at the Grand Rapids, Mich.-based healthcare system, which serves millions of individuals across the state through its hospitals, medical group, health plan and post-acute care services.
DRESEN: First of all, Spectrum Health looks at technology as an enabler to achieve our mission and vision, so we’ve chosen to make investments that support our strategic goals and drive business value for our organization and our region. We need a flexible, agile infrastructure that we can manage to meet changing and emerging needs, such as population health, Big Data and analytics. We feel like we have to anticipate business needs so we can be out in front, ready to move at the pace Spectrum Health requires.
Second, consumer expectations are pushing us. Even though healthcare is a consumer-oriented business, from a consumer-experience perspective it’s behind the curve compared to retail, finance and other sectors.
Third, Spectrum Health’s dependence on technology has certainly influenced our design strategies and data-center decisions. Since we’re the largest healthcare provider in western Michigan, the impact to our service area would be significant if we had a major technology disruption.
DRESEN: Starting around 2008, when virtualization became more widely adopted, the technology density in our environment increased substantially. The physical footprint of the data center hasn’t changed much, but the compute and storage capacity we have now is dramatically different from 10 years ago. Technology has gotten smaller and more powerful. But that means devices are drawing more power and creating more heat, which can create cooling inefficiencies in older data centers because of the way they’re air conditioned.
DRESEN: We have two data centers about 12 miles apart. We built the newer of the two facilities because of our desire, from a strategic perspective, to have redundant data centers for our key technologies and systems. Part of our decision to build new was to consolidate two smaller data centers into one and drive efficiency there, as well as being able to support our growth.
DRESEN: I think it’s easiest to compare and contrast. Our older data center, built in 2005, has a more traditional data-center design with raised floors, racks of equipment, vented floor tiles and cooling units inside the data hall. Our new data center takes an innovative modular design approach. We eliminated the raised floor and use hot aisle containment to keep the ambient air temperature a consistent 72 to 76 degrees. Plexiglas housing around the racks creates a convection stream that pushes the hot air up into the dropped ceiling, where it gets pulled out into externally attached air conditioning units, is cooled, and then is brought back into the building.
DRESEN: We have more even temperature distribution across the data center, rather than having the equipment create heat pockets in certain areas. Also, because the cooling units are no longer inside the data center, we have more floor space available for equipment. But even more importantly, with the externally attached cooling units, we can leverage external ambient air temperature to cool the hot air when it’s colder outside — and you know we have cold winters in Michigan.
As a result, we don’t have to cool the hot air as frequently using the mechanical air conditioning system that requires additional power. In 2015, we had only 10 to 15 days where we had to use the mechanical air conditioning system to cool the hot air. The rest of the year we leveraged the ambient air temperature and evaporative cooling.
DRESEN: The gains in power and cooling efficiencies have reduced our operating costs by 25 percent and reduced our power consumption by 27 percent. We estimate savings of $10 million over 10 years.
DRESEN: Thanks to advances in convergence, we put networking and storage on the same fiber optic cable. This allowed us to locate electronics in the actual row itself as opposed to having a giant patch field with cables running everywhere to provide device connectivity. Yet we have more bandwidth and higher-speed connectivity and throughput than ever before, so we’ve reduced the cost of electronics but increased the capacity of our computing and storage environment.
DRESEN: I would say we’re leveraging the cloud opportunistically. We use Software as a Service for several applications, but security remains a high concern. We’re not doing as much Infrastructure as a Service because of data and security control concerns with cloud providers. We’re also finding that how we choose to adopt cloud services depends on “data gravity,” meaning that your technology components typically will gravitate toward where the data you need to access is actually located. Because so many of our key systems depend on our clinical data, the majority of which is still localized onsite, most of the technologies we leverage are physically closer to these data sources to reduce latency.
DRESEN: There’s a huge amount of buzz in the industry about IoT, but it’s not really new to us in healthcare. The biomedical technology security challenge we have right now is a harbinger of what’s to come with IoT. The devices have old operating systems that make it very difficult for us to apply security controls appropriately. These devices tend to have long lifecycles, plus it takes time for the manufacturing firms to re-engineer the products to incorporate better security controls.
This means we’re years away from resolving this problem, which puts us in a difficult position. Some of these technologies are used for key clinical processes — yet they present significant vulnerabilities. They not only pose a threat to patient safety, but also to the rest of the enterprise infrastructure, because if one device is compromised, it becomes an entry point into the organization. So when I think about the IoT, I just see the problem ballooning.
We’re also going to really struggle to apply security measures to consumer devices such as Fitbits, for example. Security is not usually a top-of-mind concern in IoT new product development. It should be.
DRESEN: My advice is strikingly fundamental.