Abacus Insights might count the ancient counting instrument as its namesake, but that’s about as old-school as it gets for the healthcare data startup. Recipient of a $12.7 million Series A funding announced last summer, Abacus endeavors to provide healthcare insurers with easy-to-access, digestible info by harnessing the power of the cloud, or, as CFO Larry Begley previously told Built In, by “unlocking the value of payers’ vast amounts of insightful data” that’s bound up in old systems.
In this case, product echoes practice: just as the Boston company erases industry-standard silos in the name of transparency and accessibility, the DevOps team at Abacus strives for a similarly frictionless framework. The goal is to keep deployments and workflows humming. Recently, the team illustrated that ethos when they implemented a decoupling process to align their platform’s core deployment model with a defined stratification between infrastructure and application code, Cloud Operations Support Engineer Seth Allen said. The result? Improved efficiency of abstractions, enhanced predictability of software deployments and other benefits. The cumulative gain was clear.
“Through these efforts, an appreciably complex application deployment has been abstracted to a handful of simple CLI commands,” Allen said.
Below, Allen elaborated on the core tenets that buoy the team’s integrative, “all-hands-on-deck” attitude toward DevOps.
What DevOps best practices have been most impactful for your team? How have they evolved over time?
An initiative was recently finalized to align our platform’s core deployment model — comprising nearly 20,000 lines of Terraform and over a dozen Helm charts — with a more rigidly defined stratification between infrastructure and application code through a decoupling process. This has provided numerous benefits to many teams.
We’ve been able to leverage more efficient abstractions and design patterns.”
We’ve been able to leverage more efficient abstractions and design patterns and have centralized and increased transparency of deployment configurations. We’ve also gotten more granular control over development and testing workflows, all in the name of increasing predictability. Through the application team's efforts to provide its component consumers with even more effective ways of managing their resources, much of the Kubernetes deployment automation has evolved to feature custom deployment configuration, secrets management, command-line arguments and additional time-saving features. Through these efforts, an appreciably complex application deployment has been abstracted to a handful of simple CLI commands.
How does your team balance a need to incorporate best practices into their work with the desire to try and test new methodologies, tools and strategies?
There are strong factors at Abacus in general that facilitate this decision-making process in addition to intra-team management. Principally, the existence of a well-architected, cloud-native platform virtually eliminates the inertia for novel, experimental feature work to take flight by reducing the staging time and overhead, the region in which poor design decisions often take root, due to lack of development time or lack of established scaffolding to build from.
The opposite pole of this balancing act is taking a consistent and thoughtful “all-hands-on-deck” approach to feature grooming and inter-team communication to establish a broad, yet sane, range of implementation parameters and acceptance criteria. Together, these two factors create a tinderbox of development support, needing only a good idea and a purposeful strike to light the spark.
