Ingress
Applications in Kubernetes can be exposed outside of the cluster in several ways, such
as via NodePort
or LoadBalancer
type services, or via ingress controllers or
gateways.
NodePort
services expose the service directly via a port on each of the nodes, which
can then be manually managed by firewall rules or external load balancers.
LoadBalancer
services require something that can provide indvidual load balancers for
each service, which could be native load balancers provided by the underlying cloud
(see the relevant Cloud Integration section of the docs) or something like MetalLB.
Ingress controllers provide a single point of entry instead of needing to deal with multiple individual service addresses, and then use rules or path matching to direct traffic appropriately. The ingress endpoint itself still needs to be exposed via one of the above methods but can provide feature-rich management of traffic routing into the cluster.
NGINX Ingress
By default, Charmed Kubernetes sets up the NGINX Ingress Controller,
which can be customized via the config options on the worker charm.
Ingress
resources can then be used to configure routes for specific
applications.
Istio Ingress
By deploying the Istio bundle into your cluster, you can use the
Istio traffic management features. You can then relate Kubernetes
charms which support the ingress
relation to automatically manage VirtualServices
,
manually manage VirtualServices
for your applications, or even use
Ingress
resources.
Multiple Ingress Controllers
Multiple ingress controllers or gateways can be used at the same time, typically with
different configuration or exposure. The networking.k8s.io/v1
defines an
IngressClass
resource to select between controllers, but some
controllers may rely on annotations, such as Istio's
kubernetes.io/ingress.class
annotation.
See the guide to contributing or discuss these docs in our public Mattermost channel.