At Imagine Learning, we strive to empower educators and inspire breakthrough moments for over 18 million students across the United States. As a digital-first education solutions provider, our mission is to deliver robust, reliable, and secure experiences to support the boundless potential of K-12 learners. Achieving this at scale—spanning hundreds of thousands of daily users—requires a technical foundation as innovative as our products.
To meet these demands, our engineering team has embraced modern cloud-native technologies. The cornerstone of our infrastructure is Linkerd, supported by Buoyant, running on Amazon’s Elastic Kubernetes Service (EKS). Linkerd gives us the critical capabilities that we need to scale effortlessly while maintaining the performance and reliability that our users depend on.
Delivering Reliability and Security at Scale
Our users—students and educators—depend on fast, reliable, and secure access to our digital tools. Dropped traffic, slow load times, or security issues in the face of ever-evolving threats would disturb the user experience, potentially leading to customer churn. For that reason, reliability and security are a top priority for Imagine Learning. As our platform grew to include hundreds of microservices deployed across several Kubernetes clusters on AWS EKS, managing communications at scale to avoid these problems became extremely challenging. We needed a solution that would:
- Encrypt traffic seamlessly.
- Simplify service-to-service communication.
- Provide deep observability into application-layer (L7) networking.
In short, we needed a service mesh.
Linkerd: Opting for Simplicity, Performance, and Security
After evaluating multiple service mesh options, Linkerd stood out due to its simplicity, performance, and security. Unlike other solutions, Linkerd uses a tiny Rust-based sidecar microproxy, offering a small compute footprint, significantly reduced CVEs, and enhanced overall security. Linkerd’s ease of configuration allowed us to deploy it reliably and scale it within hours, enabling our team to focus on delivering value to our customers.
Additionally, Buoyant Cloud has amplified Linkerd’s capabilities by providing real-time metrics and intuitive alerting. This ensures that our operations team can swiftly diagnose and resolve issues, maintaining an exceptional user experience.
Building a Scalable and Observable Platform
Our engineering team integrated Linkerd with several other CNCF tools to create a robust platform:
- Argo CD: Enables GitOps workflows, ensuring consistent deployment of Linkerd across all clusters.
- Argo Rollouts: Enable us to roll out to customers progressively using Linkerd’s Gateway API Integration informed by metrics.
We started the implementation by adopting the GitOps mindset: our Kubernetes manifests are stored in Git, and we use Argo CD to deploy resources into our Kubernetes clusters. Argo CD manages Linkerd, Argo Rollouts, the microservices our application teams build, and the other components needed for our platform. Additionally, Buoyant Enterprise for Linkerd provides a simple Helm chart to deploy its lifecycle Operator, which makes it easy to manage our Linkerd deployments with a single custom resource. Once deployed, all meshed pods were automatically mTLSed!
Next, we started leveraging the Gateway API in our environment to use Argo Rollouts with their Gateway API plugin. We now use Argo Rollouts for canary deployments, enabling controlled rollout of new application versions to our customers: Linkerd seamlessly changes the amount of traffic going to the new version based on Argo Rollouts’ instructions, and Argo Rollouts uses Linkerd’s HTTP metrics to understand whether a rollout is going successfully. This allows us to minimize the impact customers could experience due to a release deficiency.
Together, these tools in this architecture allow us to maintain a highly observable and manageable Kubernetes environment. With Linkerd, service-to-service communication is secure and reliable, while Buoyant Cloud’s metrics enable proactive performance management.
Transformational Gains
The adoption of Linkerd has yielded significant technological and business benefits:
- Reduced Operational Overhead by 20%: By streamlining service mesh management, we’ve cut the time spent on operational tasks by a substantial margin.
- Enhanced Efficiency: Linkerd’s lightweight design reduced the compute requirements of our service mesh by over 80% compared to our previous service mesh implementation.
- Reduce Costs: Buoyant Enterprise for Linkerd’s High Availability Zone Load Balancing with Linkerd is on track to reduce our regional data transfer networking costs by at least 40%.
- Improved Reliability: Real-time observability ensures rapid issue resolution, translating to better user experiences and fewer disruptions.
- Improved Security: Linkerd has reduced the amount of service mesh-related CVEs we have been exposed to by 97% in 2024.
On a personal level, these improvements mean fewer fire drills for our engineers, allowing them to focus on innovation.
Looking Ahead
As we continue to scale, Imagine Learning remains committed to leveraging cloud-native technologies like Linkerd to deliver exceptional educational experiences. With a solid foundation on AWS EKS, powered by the reliability and support of Buoyant and Linkerd, we are poised to inspire breakthrough moments for millions of students every day.