Fleet networking communications
1. Fleet networking communications
To begin, what is fleet networking in GKE and how does it work? In GKE, clusters communicate directionally-- north-south and east-west. North-south traffic includes ingress and egress traffic and communicates with clusters via load balancer. The load balancer receives traffic from the internet or from internal applications. Next, let's explore east-west routing, which manages communication between clusters. East-west traffic is essential if you have dependencies between services in different clusters. It is also used to implement fallback strategies and perform blue-green deployments Cloud Service Mesh is a key component for managing east-west traffic, providing service discovery, traffic management, and security features for services that communicate across clusters. Let's examine why companies often choose to host clusters in different locations. One reason is proximity to customers. Companies close to customers can serve traffic with minimum latency. Another benefit is increased availability. Redundancy provides resilience and recovery if a region becomes unavailable. Compliance is also important. Companies often choose to locate clusters in the same jurisdiction as their customers to meet data residency requirements. And many customers also benefit from data locality. Keeping compute resources close to data reduces cost and increases performance. Multiple clusters can also achieve isolation. For example, you can use separate clusters for different environments, like development and production workloads, workload categorization, splitting workloads into different categories. Isolation can also help limit the impact of failure and provide upgrade flexibility without reducing availability. And with multiple clusters, you can separate sensitive data from applications, ensuring compliance and security. In GKE, several technologies work together to enable multicluster networking. Anycast IP is an IP address that can be deployed to multiple regions, routing traffic to the nearest cluster. Global HTTPS load balancers use Anycast IPs and Network Endpoint Groups, or NEGs, to distribute traffic efficiently. GKE uses Google's premium network for optimized routing. Traffic from outside Google's network is quickly handed off to the nearest Google Point Of Presence, or POP. Once inside Google's network, traffic is efficiently routed within the network, minimizing hops and latency between your clusters. To route the traffic, the global HTTPS load balancer uses NEGs to send traffic to dynamically configured endpoints, including serverless offerings, like Cloud Functions or Cloud Run, primary IPs like Compute Engine VMs, and secondary IPs, like pods on Kubernetes for container-native load balancing.2. Let's practice!
Create Your Free Account
or
By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.