Ms Fang Xu1, Mr Andrew Brock4, Dr Mark Endrei5, Mr Alastair Ferguson2, Mr Dean Taylor6, Mr Craig Windell3, Dr Ryan Sullivan1
1The University of Sydney, , , 2Macquarie University, , , 3Queensland University of Technology, , , 4SAHMRI, , , 5University of Queensland, , , 6University of Western Australia, ,
The Australian Imaging Service (AIS) aims to address the challenges shared by multiple universities and clinical sites that manage large volumes of imaging data, balancing patient privacy, and the value earned through secondary-use by trusted research communities. In order to achieve this, we’ve built standardised deployments of open source imaging informatic platform XNAT and image processing application RSNS MIRC Clinical Trial Processor (CTP) across multiple institutes.
Why do we choose Kubernetes?
We chose Kubernetes to eliminate the infrastructure dependencies, so that AIS members can use the same sets of deployment templates to deploy XNAT, CTP, analysis pipelines, or ML on their own infrastructure. We speed up the process of building, testing, and releasing our applications. By sharing the same tooling among institutes and clinical sites, we ensure the imaging data flow smoothly and securely.
How do we work with Kubernetes?
Kubernetes is used to ease the burden of configuring, deploying, managing, and monitoring large-scale containerised application. We start by building the application container images using Packer. We make Helm charts to deploy common resources while infrastructure specific resources are managed by Kustomize. We leverage Helm to manage updates or rollback. Also, we use common toolsets for logging and monitoring.
Challenges and lessons learnt
Adopting a new technology is not always straightforward, especially as this project requires high levels of collaboration among different universities. We’ll present on lessons learnt on collaboration approaches and technology. This not only completes the project, but also sets up the foundation for AIS to go forward.
Biography:
Fang Xu is the senior DevOps engineer in CloudOps team at the University of Sydney. He currently leads the AIS deployment team and manages the University of Sydney nodes of AIS.