Kubernetes Engine offers integrated support for two types of cloud load balancing for a publicly accessible application:
You can create TCP load balancers by specifying
type: LoadBalanceron a Service resource manifest. Although a TCP load balancer works for HTTP web servers, they are not designed to terminate HTTP(S) traffic as they are not aware of individual HTTP(S) requests. Kubernetes Engine does not configure any health checks for TCP load balancers.
You can create HTTP(S) load balancers by using an Ingress resource. HTTP(S) load balancers are designed to terminate HTTP(S) requests and can make better context-aware load balancing decisions. They offer features like customizable URL maps and TLS termination. Kubernetes Engine automatically configures health checks for HTTP(S) load balancers.
If you are exposing an HTTP(S) service hosted on Kubernetes Engine, HTTP(S) load balancing is the recommended method for load balancing.
Before you beginTake the following steps to enable the Kubernetes Engine API:
- Visit the Kubernetes Engine page in the Google Cloud Platform Console.
- Create or select a project.
- Wait for the API and related services to be enabled. This can take several minutes.
- Enable billing for your project.
Install the following command-line tools used in this tutorial:
gcloudis used to create and delete Kubernetes Engine clusters.
gcloudis included in the Google Cloud SDK.
kubectlis used to manage Kubernetes, the cluster orchestration system used by Kubernetes Engine. You can install
gcloud components install kubectl
Set defaults for the
To save time typing your project ID and Compute Engine zone
options in the
gcloud command-line tool
gcloudcommand-line tool, you can set default configuration values by running the following commands:
gcloud config set project [PROJECT_ID] gcloud config set compute/zone us-central1-b
Create a container cluster
Create a container cluster named
loadbalancedcluster by running:
gcloud container clusters create loadbalancedcluster
Step 1: Deploy a web application
Create a Deployment using the sample web application container image that listens on a HTTP server on port 8080:
kubectl run web --image=gcr.io/google-samples/hello-app:1.0 --port=8080
Step 2: Expose your Deployment as a Service internally
Create a Service resource to make the
reachable within your container cluster:
kubectl expose deployment web --target-port=8080 --type=NodePort
When you create a Service of type NodePort with this command, Kubernetes Engine makes your Service available on a randomly- selected high port number (e.g. 32640) on all the nodes in your cluster.
Verify the Service was created and a node port was allocated:
$ kubectl get service web NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE web NodePort 10.35.245.219 <none> 8080:32640/TCP 5m
In the sample output above, the node port for the
web Service is
Also, note that there is no external IP allocated for this Service. Since the
Kubernetes Engine nodes are not externally accessible by default, creating
this Service does not make your application accessible from the Internet.
To make your HTTP(S) web server application publicly accessible, you need to create an Ingress resource.
Step 3: Create an Ingress resource
Ingress is a Kubernetes resource that encapsulates a collection of rules and configuration for routing external HTTP(S) traffic to internal services.
On Kubernetes Engine, Ingress is implemented using Cloud Load Balancing. When you create an Ingress in your cluster, Kubernetes Engine creates an HTTP(S) load balancer and configures it to route traffic to your application.
While the Kubernetes Ingress is a beta resource, meaning how you describe the Ingress object is subject to change, the Cloud Load Balancers that Kubernetes Engine provisions to implement the Ingress are production-ready.
The following config file defines an Ingress resource that directs traffic to
apiVersion: extensions/v1beta1 kind: Ingress metadata: name: basic-ingress spec: backend: serviceName: web servicePort: 8080
To deploy this Ingress resource, download
kubectl apply -f basic-ingress.yaml
Once you deploy this manifest, Kubernetes creates an Ingress resource on your
cluster. The Ingress controller running in your cluster is responsible
for creating an HTTP(S) Load Balancer to route all external HTTP
traffic (on port 80) to the
web NodePort Service you exposed.
Step 4: Visit your application
Find out the external IP address of the load balancer serving your application by running:
$ kubectl get ingress basic-ingress NAME HOSTS ADDRESS PORTS AGE basic-ingress * 203.0.113.12 80 2m
Point your browser to the external IP address of your application and see a plain text HTTP response like the following:
Hello, world! Version: 1.0.0 Hostname: web-6498765b79-fq5q5
You can visit Load Balancing on GCP Console and inspect the networking resources created by the Ingress controller.
Step 5: (Optional) Configuring a static IP address
When you expose a web server on a domain name, you need the external IP address of an application to be a static IP that does not change.
By default, Kubernetes Engine allocates ephemeral external IP addresses for HTTP applications exposed through an Ingress. Ephemeral addresses are subject to change. For a web application you are planning for a long time, you need to use a static external IP address.
Note that once you configure a static IP for the Ingress resource, deleting the Ingress will not delete the static IP address associated to it. Make sure to clean up the static IP addresses you configured once you no longer plan to use them again.
Option 1: Convert existing ephemeral IP address to static IP address
If you already have an Ingress deployed, you can convert the existing ephemeral IP address of your application to a reserved static IP address without changing the external IP address by visiting the External IP addresses section on GCP Console.
Option 2: Reserve a new static IP address
Reserve a static external IP address named
gcloud compute addresses create web-static-ip --global
Now you need to configure the existing Ingress resource to use the reserved IP
address. Replace the
basic-ingress.yaml manifest contents with the following
apiVersion: extensions/v1beta1 kind: Ingress metadata: name: basic-ingress annotations: kubernetes.io/ingress.global-static-ip-name: web-static-ip spec: backend: serviceName: web servicePort: 8080
This change adds an annotation on Ingress to use the static IP resource named
web-static-ip . To apply this modification to the existing Ingress, run the
kubectl apply -f basic-ingress.yaml
kubectl get ingress basic-ingress and wait until the IP address of your
application changes to use the reserved IP address of the
It may take a couple of minutes to update the existing Ingress resource, re- configure the load balancer and propagate the load balancing rules across the globe. Once this operation completes, the Kubernetes Engine releases the ephemeral IP address previously allocated to your application.
Step 6: (Optional) Serving multiple applications on a Load Balancer
You can run multiple services on a single load balancer and public IP by configuring routing rules on the Ingress. By hosting multiple services on the same Ingress, you can avoid creating additional load balancers (which are billable resources) for every Service you expose to the Internet.
Create another web server Deployment with the version
2.0 version of the same
kubectl run web2 --image=gcr.io/google-samples/hello-app:2.0 --port=8080
Then, expose the
web2 Deployment internally to the cluster on a
NodePort Service called
kubectl expose deployment web2 --target-port=8080 --type=NodePort
The following manifest describes an Ingress resource that:
- routes the requests with path starting with
- routes all other requests to the
apiVersion: extensions/v1beta1 kind: Ingress metadata: name: fanout-ingress spec: rules: - http: paths: - path: /* backend: serviceName: web servicePort: 8080 - path: /v2/* backend: serviceName: web2 servicePort: 8080
To deploy this manifest, save it to a
fanout-ingress.yaml, and run:
kubectl create -f fanout-ingress.yaml
Once the ingress is deployed, run
kubectl get ingress fanout-ingress to find
out the public IP address of the cluster.
Then visit the IP address to see that both applications are reachable on the same laod balancer:
http://<IP_ADDRESS>/and note that the response contains
Version: 1.0.0(as the request is routed to the
http://<IP_ADDRESS>/v2/and note that the response contains
Version: 2.0.0(as the request is routed to the
Services exposed through an Ingress must serve a response with HTTP 200 status
GET requests on
/ path. This is used for health checking. If your
application does not serve HTTP 200 on
/, the backend will be marked unhealthy
and will not get traffic.
Ingress supports more advanced use cases, such as:
Name-based virtual hosting: You can use Ingress to reuse the load balancer for multiple domain names, subdomains and to expose multiple Services on a single IP address and load balancer. Check out the simple fanout and name-based virtual hosting examples to learn how to configure Ingress for these tasks.
When an Ingress is deleted, the Ingress controller cleans up the associated resources (except reserved static IP addresses) automatically.
To avoid incurring charges to your Google Cloud Platform account for the resources used in this tutorial:
Delete the Ingress: This deallocates the ephemeral external IP address and the load balancing resources associated to your application:
kubectl delete ingress basic-ingress
If you have followed "Step 6", delete the ingress by running:
kubectl delete ingress fanout-ingress
Delete the static IP address: Execute this only if you followed Step 5.
If you have followed “Option 1” in Step 5 to convert an existing ephemeral IP address to static IP, visit GCP Console to delete the static IP.
If you have followed “Option 2” in Step 5, run the following command to delete the static IP address:
gcloud compute addresses delete web-static-ip --global
Delete the cluster: This deletes the compute nodes of your container cluster and other resources such as the Deployments in the cluster:
gcloud container clusters delete loadbalancedcluster