September 2018 Summaries

Multi-cluster is a strategy for deploying an application on or across multiple Kubernetes clusters to improve availability, isolation, and scalability. Organizations are increasingly adopting this approach to ensure compliance with different regulations, increase the speed and safety of software delivery, and eliminate vendor lock-in. A multi-cluster architecture can be replicated or split by service, providing stronger isolation between parts of the application at the expense of greater complexity. Configuring multi-cluster Kubernetes involves choosing a strategy that works best for an organization's needs, such as a network-centric approach that creates connectivity between clusters or a Kubernetes-centric approach that supports and extends core Kubernetes primitives. Most organizations adopting multi-cluster are evaluating network-centric approaches due to their simplicity and ease of adoption.

The key points of the text are that API Gateways and service meshes will both be needed at the edge due to their different use cases, but they will share some commonalities such as using Envoy Proxy for L7 routing. Most organizations will start with an API Gateway over a service mesh because everyone needs an ingress solution, while not everyone needs a service mesh. API Gateways can provide much of the functionality that a service mesh provides, such as observability and dynamic routing. The edge use case is distinct from internal routing due to the lack of control over the client, requiring handling of various permutations and edge-related issues. There is ongoing investment in integrating API Gateways with popular service meshes like Istio, and it's unclear if convergence between the two will happen soon.

Onefootball, a media company with over 10 million monthly active users, needed a reliable and efficient Kubernetes-based API Gateway and Ingress solution for its workload of 6,000 requests per second. The company chose the open source Edge Stack Kubernetes API gateway due to its cost reduction, observability, and maintainability benefits. By using Edge Stack, Onefootball reduced the number of cloud-based load balancers from approximately 100 to four, leveraged Prometheus' capabilities for better observability, and decoupled cluster settings and applications' delivery process for increased velocity in shipping new features.