Real-time Kubernetes: How Entain Australia 10x'd throughput with Linkerd

Deploying a service mesh to production can provide immediate operational and cost economy benefits for a large number of services using long-lived RPC protocols in Kubernetes.

• Transitioning to Kubernetes logical services improved service discovery and connectivity between services.
• Using gRPC for backbone protocols resulted in long-lived connections and inefficient load balancing.
• Deploying a service mesh to production provided immediate operational and cost economy benefits, including reduced costs for moving data between zones and mutual TLS support for all communications within the pod.
• Future plans include deploying Chaos Mesh more fully and switching to isolated individual clusters per geographic area.

The speaker's team processed around 10 billion external updates a month and needed to do it reliably and as fast as possible. They had zero downtime and had to have high performance updating of the prices sub-second every second 24/7. As they grew their microservices stack, they faced challenges in service discovery, governance, and observation. Deploying a service mesh to production to hundreds of services with zero prior experience in the space of a few hours provided immediate operational and cost economy benefits.

Entain Australia operates a large scale Kubernetes deployment consisting of 300 microservices, 3000 pods, and multiple Kubernetes clusters across multiple regions. Due to the nature of Entain's business, this system must be as close to real-time as possible. In this talk, the Entain team describes how they moved from connection-level load balancing to request-level load balancing and increased overall application throughput by an order of magnitude by installing Linkerd. They discuss the impact of Linkerd's load balancing on gRPC calls, the decision-making process for choosing a service mesh, and lessons learned rolling out a service mesh to production.