Emma
Zheng
AnvilOps: Accelerating the Kubernetes deployment cycle
Abstract profile. Full document pending author claim.
Authors:
Emma Zheng
Date Created:
Not specified
Course Title:
Professor:
Not specified
About Paper:
The container orchestration system Kubernetes enables researchers to run and manage container-based tools and applications, providing a valuable complement to traditional HPC workflows. However, deploying applications to Kubernetes requires familiarity with container image building and resource configuration, posing a significant challenge to users. To add further friction, many deployment steps must be repeated with each application update. To ease adoption of Kubernetes, we developed AnvilOps, a platform-as-a-service solution that automates the complex task of application deployment. AnvilOps is comprised of a Node.js web application and several Kubernetes components, including a Buildkit daemon that manages container image builds. The application leverages the Rancher API to provision resources inside a user's projects and maintain access control. Through the web interface, a user can deploy an application from a GitHub repository with little to no configuration needed. AnvilOps automatically builds a container image, creates the necessary Kubernetes resources, and makes the application available on a custom subdomain. In particular, AnvilOps can build applications either from user-provided Dockerfiles or by using the Railpack frontend to Buildkit to generate build plans, eliminating manual build configuration. Furthermore, AnvilOps enables continuous deployment through GitHub webhooks; when updates are pushed or triggered by workflow events, the application is automatically rebuilt and redeployed. The platform streamlines a lengthy deployment procedure into an automated loop that requires minimal user intervention. AnvilOps bridges the gap of support for users unfamiliar with Kubernetes infrastructure, augmenting the accessibility of persistent services in HPC. Keywords: Kubernetes; Docker; Automation; Continuous Deployment; Platform as a Service
Source:
Purdue University / 2025
Topics:
No topics listed
Co-authors:
Emma Zheng