Case Study: Using K8s to Navigate Medical Shortages

Kubernetes (K8s) continues to grab the attention of engineers within many sectors as the industry-standard method to deploy and orchestrate large fleets of containers. Cloud-native strategies involving K8s are having a modernizing effect, helping teams make good on the promise of rapid application development.

Simultaneously, new cloud-native tools have been paramount in health care to address ongoing hospital staffing shortages. As COVID-19 continues to overwhelm jurisdictions with unanticipated peaks, health care institutions must juggle staff and equipment availability to smartly respond. Often, optimizing the communication here hinges on innovative digital technologies.

I recently met with Reed Glauser, director, platform engineering, CHG Healthcare Services. CHG recently built a container-as-a-service infrastructure platform to standardize around a common deployment model for their developer teams. We’ll examine how the company used Kubernetes and stateful tools to create this model and consider what sort of benefits it has brought them.

The Scenario: Medical Staffing Applications

CHG facilitates staffing for 30% of the temporary health care workers in the United States. As Glauser describes, CHG’s software accelerates the ability to connect doctors with hospitals. These tools help match qualified workers with the right positions, helping to coordinate rotating doctors or fly-in specialists.

As one can imagine, managing staffing for such a large number of workers involves many individual software services working in concert, not to mention storing a lot of data on specific credentials, documents and assignments. A big part of the software requirements consists of a portal to aggregate relevant staffing data.

The Goal: Increased Standardization of Deployment

Nowadays, collaboration around rapid delivery is crucial for digital tools to evolve. Yet, CHG originally had traditionally divided developers and operations departments. De-siloing these units was necessary to enact more rapid delivery, says Glauser. “It’s always difficult for any company to see that visionary aspect.” But, the Agile creed was a huge component of getting developers on board, he describes.

Standardizing the company’s software deployment mechanism was also a driving force behind the re-platforming. “My team’s consumers are internal developers,” said Glauser. “How do we accelerate their delivery with the right toolsets and standardization?”

The term “guardrails” sometimes has a negative connotation—engineers might assume that strict rules will inhibit their flow. Glauser, on the contrary, sees guardrails as opening more doors than they close. Standardizing common logic around things like container initiation, monitoring or secrets management brings many net benefits for developers, “giving them the toolsets to enable them to be the creativity engine that they are,” says Glauser.

The Solution: A Common Stateful K8s Deployment

To increase the portability of their offerings, CHG opted for a cloud-native approach using a microservices architecture, explains Glauser. They first tried EC2 instances but found they didn’t scale well. Eventually, the team landed on EKS for native Kubernetes on AWS, utilizing Portworx for storage. Purportedly, the business has also cut costs considerably by migrating their Kafka event streaming to persistent storage.

With every iteration, IT continues to onboard more and more developers onto a common deployment framework, says Glauser. Whereas the first iteration has about 25% adoption, the second iteration has reached around 40% to 50%. Over time, he hopes to achieve 100% adoption from the developer side.

Standardizing the deployment model has also brought security benefits, notes Glauser, since there is now a minimum bar of entry for every app deployed. This includes image scanning and vulnerability detection to spot potential threats early on. Cybersecurity is a major priority in health care settings that must meet patient data-handling compliance requirements.

Future Improvements

Of course, Glauser notes that their DevOps setup is still a work in progress. They’ll be devoting continued efforts to increasing developer experience with more abstraction. Placing more power from the command line, for example, could reduce the time it takes to stand up infrastructure. “We’re trying to shrink that window as much as possible,” he says.

Other long-term goals include encouraging broader DevOps thinking and promoting domain-driven design.

End Benefit: Accelerated Development

A standard deployment model brings the ability to move quickly. For example, Glauser estimates that it once took three weeks to get services up, but this timeframe has decreased to a day or less.

Guardrails can also streamline the integration side of building core infrastructure, as developers can now pick from a suite of common tools. Together, these factors edge software teams closer to accelerating as fast as the business would like.

“The opportunity to standardize enables the opportunity to move quickly,” said Glauser. “It’s changed the game, for sure.”

Bill Doerrfeld

Bill Doerrfeld is a tech journalist and analyst. His beat is cloud technologies, specifically the web API economy. He began researching APIs as an Associate Editor at ProgrammableWeb, and since 2015 has been the Editor at Nordic APIs, a high-impact blog on API strategy for providers. He loves discovering new trends, interviewing key contributors, and researching new technology. He also gets out into the world to speak occasionally.

Bill Doerrfeld has 105 posts and counting. See all posts by Bill Doerrfeld