Virgin Australia automates lifecycle management using Amazon EKS Auto Mode

Founded in 2000, Virgin Australia operates a fleet of more than 90 aircraft and serves over 40 domestic and international destinations on over 80 routes. The company relies on secure, reliable digital systems to support everything from booking and ticketing to flight operations.

“These systems are running critical services that affect airline operations, on-time performance, and our ability to service customers through our website, retail, ticketing, reservations, flight searching, and partner functions,” says Shaw Marsh, cloud and infrastructure architect at Virgin Australia. “This is absolutely mission-critical infrastructure, and we want to maintain it with as little risk as possible.”

As Virgin Australia grew, it also had to manage its increasingly complex digital infrastructure footprint. Self-managing Kubernetes at scale was becoming increasingly difficult. Virgin Australia’s engineering team managed complex lifecycle management tasks, such as node provisioning, patching, scaling, and coordinating upgrades. Kubernetes requires control plane upgrades 3–4 times per year, with each upgrade requiring careful testing. Engineers were spending more energy maintaining infrastructure than improving customer-facing services.

In early 2024, Virgin Australia sought to modernize its container infrastructure with a fully managed, flexible, and scalable solution. The announcement of Amazon Elastic Kubernetes Service (Amazon EKS) Auto Mode—a service that fully automates Kubernetes cluster management for compute, storage,
and networking on AWS with a single click—provided the ideal solution. By adopting this technology, Virgin Australia could streamline its operations, automate cluster provisioning, rightsize compute, apply automatic patches and upgrades, and adopt preconfigured add-ons for networking and storage.

Virgin Australia’s engineering team adopted a methodical approach to modernizing its Kubernetes operations through a 6-month blue/green migration. For an airline where minor disruptions can affect thousands of passengers, this careful testing of workloads before shifting production traffic was crucial for risk management.

While Amazon EKS Auto Mode handles operationally intensive elements—compute management, ingress control, and core add-ons—the airline maintains control over customized aspects like observability tools and governance systems.

The impact was immediate after migration. The team reduced time-consuming discussions about instance types and scaling concerns during peak periods. Amazon EKS Auto Mode automatically manages rightsizing and scaling, removing the traditional tension between overprovisioning and service reliability.

One of the most significant improvements came in system maintenance. Previously, updates required extensive planning and often led to delayed upgrade schedules due to downtime concerns. Amazon EKS Auto Mode keeps instances up to date with the latest Bottlerocket-based AMIs and security patching. Virgin Australia now maintains a consistent upgrade cadence without consuming engineering resources.

The preconfigured networking and storage add-ons freed Virgin Australia’s engineers to focus on their priorities: observability and governance. “Amazon EKS is flexible, and Amazon EKS Auto Mode represents a forward-thinking approach to Kubernetes management,” says Marsh. “This service delivers automation without compromising flexibility. It’s exactly what we needed to advance our containerization strategy.”

The migration delivered results at enterprise scale. More than 200 APIs were successfully migrated onto Amazon EKS Auto Mode, confirming that it could support critical production workloads. Virgin Australia now completes cluster upgrades in a predictable cadence through in-place upgrades—a process previously considered too risky with its old tooling. Each cluster upgrade requires approximately 2 weeks of effort per team member, compared to at least 1 month per cluster that blue/green migrations would have required. In 6 months, the airline completed two full in-place upgrades across 8 clusters. Virgin Australia has also stated that its operations have run smoothly, with no unplanned downtime and zero security incidents since the migration.

Looking ahead, Virgin Australia plans to expand its use of Amazon EKS Auto Mode beyond APIs to include external-facing web workloads. It has already built four production clusters and initiated a project to host web workloads, demonstrating the scalability and flexibility of its modernized infrastructure.

This migration has given these engineers the freedom to focus on strategic innovation rather than routine maintenance tasks. For example, the team is now developing progressive delivery capabilities and self-service options that empower developers to deploy workloads to Amazon EKS more efficiently. They also built a comprehensive observability solution with dashboards and multi-channel alerts. “Using Amazon EKS Auto Mode, our team can focus on more interesting work,” says Marsh. “It gives us the ability to service our organization’s system requirements and extend those systems to be more agile and flexible than we could otherwise.”