liveBook · Manning

MultiClusterIngress

As explained in Google Anthos in Action

Multicluster Ingress is a Google-hosted controller that deploys shared HTTP(S) load-balancing resources across multiple GKE clusters and regions.

MultiClusterIngress is a Kubernetes custom resource for exposing an application across multiple Anthos GKE clusters. It works with a Google-hosted controller that programs an external HTTP(S) load balancer. A single public address can serve clusters in different regions, allowing traffic to reach an available deployment without changing the application’s IP address or domain name. section 7.2.1 section 10.2.1

The deployment separates configuration from traffic routing:

  • Designate one fleet member as the config cluster. Store the MultiClusterIngress and MultiClusterService objects there for the Google-managed controller to read. section 7.2.1
  • Use MultiClusterService to identify the service and participating clusters; use MultiClusterIngress to define the external routing rules and their service backends. section 10.2.1
  • The controller programs Google’s external HTTP(S) load balancer using network endpoint groups (NEGs). Routing accounts for service or Pod availability across fleet members. section 7.2.1 section 10.2.1

Expose a service in two regions

A MultiClusterService example selects an app on clusters in europe-west1-c and asia-northeast1-a. In the same namespace, a MultiClusterIngress rule maps foo.example.com to that service on port 80. The illustrated manifest is a configuration example, not a ready-to-apply file: its displayed MultiClusterIngress apiVersion contains an apparent typo, and the excerpt is incomplete. section 10.2.1

Choose this pattern when one service needs regional proximity, multicluster availability, or a stable public address during cluster migration. Health checking and traffic failover help keep the address useful when a deployment becomes unavailable. Keep the clusters and their service configurations aligned with the intended fleet-wide routing rules. section 7.2.1 section 10.2.1

Figure 10.8 Multicluster Ingress to multiple GKE clusters in GCP
10-08

View the multicluster diagram to locate the shared Ingress path relative to the GKE clusters. section 10.2.1

How Google Anthos in Action covers it

How deep it goes
Explains controller, config-cluster, and NEG responsibilities with a resource example, but defers deeper Ingress networking detail.
Examples
Shows a MultiClusterService selecting European and Asian clusters and a MultiClusterIngress mapping a hostname to its service backend.
What it assumes
Assumes familiarity with GKE clusters, Kubernetes custom resources, services, and HTTP(S) load balancing rather than teaching their foundations.
What you can do after
Operate/configure it by placing routing and service resources on a config cluster for a shared multicluster application address.
How it connects
Connects MultiClusterService, Anthos fleets, network endpoint groups, external HTTP(S) load balancing, health checks, and traffic failover.
Cover of Google Anthos in Action
liveBookmanning.com
Read Google Anthos in Action on liveBook Get the book at manning.com

This explainer was generated by Manning's knowledge graph from the text of Google Anthos in Action. The section links point to the parts of the book it draws on.