V/Line is Australia’s largest regional public transport operator. More than 1,300 V/Line branded coach services connect with the rail network and serve regional Victorian communities where trains do not operate. In addition to being a passenger service operator, V/Line also provides access to, and maintains, 3,520 kilometres of rail track used by passengers and freight rail services.
V/Line’s TrackR application provides real-time displays of train locations, train performance and expected arrival and departure times. V/Line had identified a need to migrate their TrackR application to Elastic Kubernetes Service (EKS) and wanted to undertake a proof of concept before implementing the solution in production.
Specifically, evaluating EKS for the TrackR application was to ascertain its suitability to overcome the following issues:
- The platform TrackR currently runs on is not highly available and is not designed to be scalable
- The infrastructure components on which the application runs is not fault-tolerant, leading to potential unplanned outages.
- Application maintenance is time consuming and require outages.
V/Line engaged Cevo for a Proof of Concept to understand the suitability of EKS platform for their application.
In collaboration with the delivery and testing teams of V/Line, Cevo undertook the following steps:
- Created a Kubernetes cluster using Amazon’s Kubernetes as a service (EKS)
- Implemented a DevOps build pipeline to build, test and deploy a secure and scalable AWS EKS cluster in a reliable and repeatable manner
- Used AWS ECR as the image repository for the TrackR microservices
- Used AWS CloudFormation to consistently build the required AWS Infrastructure components of the TrackR solution (AWS RDS, EC2 Instances used for the Bastion Hosts, SSL Certificates, Route53 hosted zones and IAM roles and policies)
- Application logs and cluster metrics were shipped to AWS CloudWatch and sample dashboards that can be used for monitoring were also built.
Along with implementing the solution, Cevo also provided the daily/monthly cost of running the application on EKS Infrastructure.
As a result of the Proof of Concept, V/Line were able to obtain the answers around:
Suitability – Proof that the TrackR application is able to be deployed on the Kubernetes platform
Scalability – Ability to scale the infrastructure resources of the application with no interruption to the application
High Availability – Ability to tolerate multiple simultaneous failures of Kubernetes worker nodes.
After considering the V/Line application team’s knowledge in AWS, Kubernetes and the size of the TrackR application, Cevo also suggested alternatives to EKS, such as ECS and Fargate. This gives V/Line a broader perspective, and be better placed to consider all variables when making a strategic decision.