By 2021, DPHS had reached an impasse: Growth was driving the need to deliver new features faster without compromising stability, and their reliance on a partially manual VM-based pipeline just no longer cut it. With Architect, they’ve been able to automate deployments, synchronize across functions, and cut deployment times by 80% or more.
“We went from thirty-minute deployments at fastest to about five minutes with Architect.”
– Errol Markland, Product Development Team Lead
Founded in 2013 and headquartered in Boston, MA, DecisionPoint Health Solutions (DPHS) offers a smarter approach to patient and provider engagement. For years, health plans have relied on descriptive data and reactive engagement. DecisionPoint empowers clients to understand and predict the whole member journey, enabling sustained improvements in member health outcomes and plan performance. DecisionPoint combines the latest, most practical technologies and a deep understanding of healthcare, bringing innovative, pragmatic solutions to an industry that touches us all.
The cost of manual effort
By 2021, DPHS had reached an impasse: Growth was driving the need to deliver new features faster without compromising stability, and their reliance on a VM-based pipeline just no longer cut it. In the existing setup, Jenkins would monitor for code changes and take care of the deployment. But the process wasn’t fully automated, meaning manual steps still had to be coordinated among engineering, data, and product teams. And this had led to variations in the process per application team, as well as differing VM specs per application. Consequently, releases that should have taken minutes took hours.
One command to rule them all
The first challenge the team chose to tackle was to address the issue of VM scalability by moving to the cloud, thus eliminating some manual steps while addressing some of the other cost constraints. This improved time and ease to deploy, though some manual steps remained. The team needed help getting the rest of the way to a fully automated pipeline. Enter Architect.io.
With Architect, the team could issue a single command to orchestrate provisioning of new cloud environments and deployment of code into them. Most critically, Architect’s dependency management meant that any developer working on a component could simply declare what services and interfaces they needed, and those components would be pulled into the deployment automatically. The result of this level of automation was an 83% reduction in time to deploy.
Such dramatic improvements led the team to ask what impact they could have on pre-production if the same infrastructure, processes, and technologies supported every stage of the development pipeline. Today, Decision Point uses Architect to automate deployments at every stage of the development lifecycle, reducing cycle times, and increasing overall MTTD. Because the process is the same from end to end, deploying new code is faster, easier, and more reliable. And because the entire process is automated end to end, teams can deploy with confidence.
DecisionPoint utilizes two key features of Architect to increase site reliability and observability. Using the architect service configuration (
architect.yaml) to reference the relevant health check endpoints, Architect can detect when an application is sick. Since Architect also maintains replica deployments of the service, Decision Point can automate pointing to the replica while Architect redeploys the sick service.
For data privacy reasons, DecisionPoint Health maintains customer data in single-tenant services. Architect makes it easy to create a new single-tenant API instance and alias it, making onboarding a new client much lower cost.
Architect supports path-based routing, which makes it easier to define routing to services within a single domain from within the
Zero trust security and privacy
As an innovator in the Healthcare space, DecisionPoint doesn’t compromise on security and data privacy. To ensure data privacy, sensitive services get deployed single tenant, in their own segregated infrastructure. Architect makes it easy to manage these data silos, as well as keep them highly available. DecisionPoint uses Architect’s health check feature to notice a service isn’t healthy, failover to a hot spare, and redeploy the unhealthy service, all without interruption and without needing human intervention.
DecisionPoint + Architect: A better future together
In the future, DecisionPoint is planning an initiative to implement one-click client provisioning all the way through the pipeline, so that new client services can be previewed, tested, and deployed to production in a fully automated fashion. And given the successes they’ve seen in their patient and provider-facing application, they are considering whether the same benefits could apply to their internal-facing applications as well.