Using AWS Wavelength, Amazon EKS, and Rafay to deliver ultra-low latency applications for 5G devices

This post is contributed by Haseeb Budhani, CEO of Rafay

AWS Wavelength is an infrastructure offering optimized for mobile edge computing applications. With Wavelength, developers can deploy their applications closer to end users, resulting in an improved end-user experience. Wavelength enables the development and adoption of a new class of latency-sensitive use cases that can leverage the availability of powerful compute resources embedded in the telco network.

Enterprises delivering new applications are predominantly choosing to implement applications as containerized microservices. To support customers’ containerized applications, AWS offers a managed Kubernetes service, Amazon EKS, that can be accessed from Regions and Wavelength Zones.

Kubernetes is highly configurable software that requires some expertise of customers who want to implement it. Rafay, an Amazon partner, offers a SaaS-first Kubernetes management platform that enables full lifecycle management of Amazon EKS via the Rafay platform. For AWS customers operating hybrid environments, Rafay also provides turnkey capabilities to provision Kubernetes clusters on-premises. Both cloud-based (EKS) clusters and on-premises (upstream Kubernetes) clusters can be managed through a single pane of glass.

With Rafay’s most recent release, AWS customers can now leverage the Rafay platform to extend EKS clusters to Wavelength Zones. AWS customers can also use Rafay to spin up additional node groups for a pre-existing EKS cluster in a Wavelength Zone. Why is this important? Customers want the same set of management tools to manage their modern infrastructure, be it in the cloud, on-premises, or in the telco network. Customers that plan to leverage Wavelength Zones are likely to deploy applications across multiple locations to ensure that end users get the best experience across geographies. The number of EKS clusters across public cloud zones and Wavelength Zones can easily grow to tens or even hundreds over time. Because the distribution of end-user populations can vary quite a bit over time, the list of target clusters may need to be updated in real time.

To meet these unique needs, the Rafay platform provides complementary capabilities that AWS customers can leverage to operate a fleet of EKS clusters across a variety of infrastructure types and locations. While this blog only depicts UI screenshots, Rafay also provides automation capabilities, so these actions can easily be tied into customers’ preferred automation frameworks:

1. EKS provisioning and node group creation

With Rafay, customers have the option to deploy a new EKS cluster for each Wavelength Zone, or extend an EKS cluster deployed in the AWS Region to the telco network by deploying a node group into a given Wavelength Zone. In both cases, the EKS cluster’s control plane is managed in an AWS Region. Rafay provides UI, CLI, and API based workflows to accomplish these tasks.

2. Blueprint application across clusters

When a new cluster is created, AWS customers must invest time configuring the new cluster before it’s ready for application deployment. The cluster may need an ingress controller to load balance and proxy inbound requests, Prometheus (for metrics collection), Fluentd (for log aggregation), Twistlock (for application security), and so on. Rafay lets you define cluster blueprints that can be applied to a new or pre-existing cluster on demand. Blueprinting is a critical capability that ensures standardization across the cluster fleet, guaranteeing that applications will work consistently and reliably across all clusters.

3. Multi-cluster application deployment

Rafay enables AWS customers to deploy containerized applications across multiple Kubernetes clusters. Customers are free to mix clusters deployed in AWS Regions, on premises, or in Wavelength Zones as part of a deployment policy.

A key innovation that Rafay offers is the ability for customers to describe clusters using labels. Assume a fleet of clusters spread across AWS Regions and Wavelength Zones that are considered production clusters. Each cluster that contains a node group in a Wavelength Zone is also assigned the label Edge:Wavelength. When a new application needs to be deployed to all production clusters, operators can leverage the Rafay platform to deploy the application (e.g. packaged as a Helmchart) to the Edge:Wavelength label. As a result, the application will be deployed to all clusters carrying said label.

Most importantly, if a new cluster is deployed at a future date and is also assigned the label PRODUCTION, the Rafay platform’s application reconciliation logic will automatically deploy the relevant application to this new cluster without any operator intervention. Similarly, if a cluster needs to be retired for any reason, the Edge:Wavelength label can simply be removed from it, which will result in all applications tied to the Edge:Wavelength label to be removed from the cluster being retired.

Conclusion

With Rafay, AWS customers can accelerate their application modernization journey. Rafay’s SaaS-first Kubernetes management platform enables customers to spin up EKS clusters in AWS Regions or in Wavelength Zones. Rafay lets you standardize cluster configurations, while simplifying the ongoing deployment of applications across multiple clusters. For customers building containerized applications for low-latency use cases, EKS, Wavelength, and Rafay are better together.