What is it about?

Modern applications often use small pieces called microservices running in containers, managed by a system called Kubernetes. Companies frequently deploy these applications across multiple Kubernetes clusters, which can be in different places like various cloud providers or their own data centers. A big challenge is getting the microservices on one cluster to talk to services on another cluster easily and efficiently. Traditionally, this required complicated custom setups. This study compares three specific tools (Skupper, Submariner, and Istio) designed to simplify connecting these clusters. It evaluates their performance – specifically, how fast they are (looking at throughput and latency) and how many computer resources (like CPU power and energy) they use. The research shows that each tool has different strengths and weaknesses or trade-offs in performance and resource use. The best tool to use depends on what is most important for your application, such as needing low communication delays or using fewer computer resources.

Featured Image

Why is it important?

The study is important because deploying applications in multi-cluster Kubernetes environments is becoming increasingly common. Organizations choose this approach for various reasons, including tenant isolation, improved availability, compliance with requirements, greater scalability, and suitability for edge computing and distributed systems. However, managing these environments introduces significant complexity, particularly in enabling communication among microservices spread across different clusters. Traditionally, connecting clusters has required custom configurations like VPNs or firewall rules. What makes this paper unique and important is its comprehensive analysis of network performance characteristics for three specific, popular open-source multi-cluster networking solutions: Skupper, Submariner, and Istio. These solutions are designed to address the challenges of microservice connectivity across clusters.

Read the Original

This page is a summary of: Bridging Clusters: A Comparative Look at Multi-Cluster Networking Performance in Kubernetes, May 2025, ACM (Association for Computing Machinery),
DOI: 10.1145/3676151.3719352.
You can read the full text:

Read

Contributors

The following have contributed to this page