Services
Services work similar to Deployments in that they use selectors to match pods, but instead of managing workloads, they create virtual IPs that can be used to reach the pods.
When a service is created, it is assigned a cluster IP that is stable and unique within the cluster. This IP is used to reach the service, and it will load-balance the traffic to the pods that back it. This means that we can reach the service from any node in the cluster, and the traffic will be routed to the correct pod(s).
While doing so, the service will forward traffic only to the pods that match its selector and are responsive. This means that if a pod is not ready, it will not receive any traffic from the service.
Kubernetes provides three kinds of services:
- ClusterIP - the default type, creates a virtual IP inside the cluster that can be used to reach the service
- NodePort - creates a port on every node in the cluster, and forwards traffic to the service
- LoadBalancer - creates a load balancer in the cloud (via the Cloud-Controller-Manager or similar means), and forwards traffic to the service
Internal networking and implications of connectivity differ between each variant, with one service type building on top of another, so we will look at all of them separately!
Created: September 15, 2023