I'm not sure how load balancing works with Ingress.
If I understand correctly, what happens is actually something like this:
I fail to see how the load balancing is performed.
What is wrong in the above scheme that I have drawn?
Can you help me rectify it?
Notes:
- The following answer tells me that the Ingress controller itself is of type 'loadbalancer': Ingress service type
- I use kind ClusterIP because I don't want to expose the loadbalancer to the outside world. The following article does not support this claim, where the load balancer would be provided by the service:
https://medium.com/google-cloud/kubernetes-nodeport-vs-loadbalancer-vs-ingress-when-should-i-use-what-922f010849e0
The ClusterIP services themselves perform load balancing. The naming can be confusing as LoadBalancer services are not the only services that involve load balancing - LoadBalancer actually means something more like 'cloud provider please create an external load balancer and point it at this service'. The kubernetes ClusterIP services also load-balance across Pods in different Nodes using the kube-proxy. If you don't want kubernetes to do load balancing then you have to specifically disable it by creating a headless service.
It seems like the first scheme you drew is correct. But I think you get confused in terminology. Particularly in the difference between ingress and ingress-controller.
Ingress is a type of resources in k8s (like Service, Deployment, ReplicaSet etc). We use ingress if we want to expose some services to an external world with binding to some path and host (i.e. myapp.com/api -> my-api-service).
The job of ingress-controller is to handle creation/update/deletion of ingress resources and implement all the functionality needed for ingress. Under the hood ingress-controller is a simple deployment exposed as LoadBalancer or NodePort service depending on where k8s is deployed. And image-controller forwards received request further to one of pods of service which matches host and path in some of the deployed ingress resources.
I got curious and thought: why would I need an Ingress for load balancing on layer 7 if the only thing it does is forward the traffic to a Service that implements the load balancing on layer 4?
Most Ingress controller implementations I looked up, talk to the Kubernetes API server to keep track of all Pods associated with a Service. Instead of forwarding traffic to the Service, they skip the intermediary and directly forward to the Pods. Since an ingress controller operates on layer 7, it enables a more application-oriented load balancing.
Related
I use MetalLB and Nginx-ingress controller to provide internet access to my apps.
I see that in most configurations, the service is set to ClusterIP, as the ingress will send traffic there.
My question is: does this end up with double load balancing, that is, one from MetalLB to my ingress, and another from my ingress to the pods via ClusterIP?
If so, is this how it is supposed to be, or is there a better way?
Metallb doesn't receive and forward any traffic, so
from MetalLB to my ingress
doesn't really make sense. Metallb just configures kubernetes services with an external ip and tells your surrounding infrastructure where to find it. Still with your setup there will be double load-balancing:
Traffic reaches your cluster and is load-balanced between your nginx pods. Nginx handles the request and forwards it to the application, which will result in a second load-balancing.
But this makes total sense, because if you're using an ingress-controller, you don't want all incoming traffic to go through the same pod.
Using an ingress-controller with metallb can be done and can improve stability while performing updates on you application, but it's not required.
Metallb is a solution to implement kubernetes services of type LoadBalancing when there is no cloud provider to do that for you.
So if you don't need layer 7 load-balancing mechanism you can instead of using a service of type ClusterIP with an ingress-controller just use a service of type LoadBalancing. Metallb will give that service an external ip from your pool and announce it to it's peers.
In that case, when traffic reaches the cluster it will only be load-balanced once.
I've just deployed websocket based echo-server on AWS EKS. I see it's running stable and okay but when I was searching for implementation details I was finding only articles that were saying something about nginx ingress controller or AWS application loadbalancer and a lot of troubles with them.
Do I miss anything in my current, vanilla config? Do I need the AWS ALB or nginx ingress controller?
Thank you for all the replies.
All the best.
Do I miss anything in my current, vanilla config?
You probably exposed your echo-server app using service type - ClusterIP or NodePort which is fine if you only need to access your app locally in the cluster (ClusterIP) or using your node IP address (NodePort).
Do I need the AWS ALB or nginx ingress controller?
They both are different things, but they have similar common goal - to make your websocket app available externally and distribute traffic based on defined L7 routing routes. It's good solution if you have multiple deployments. So you need to answer yourself if you need some kind of Ingress Controller. If you are planning to deploy your application into production you should consider using those solutions, but probably it may be fine with service type LoadBalancer.
EDIT:
If you are already using service type LoadBalancer your app is already available externally. Ingress controller provides additional configuration possibilities to configure L7 traffic route to your cluster (Ingress Controllers are often using LoadBalancer under the hood). Check this answer for more details about differences between LoadBalancer and Ingress.
Also check:
Choosing the Right Load Balancer on Amazon: AWS Application Load Balancer vs. NGINX Plus
Configuring Kubernetes Ingress on AWS? Don’t Make These Mistakes
WebSocket - Deploy to Kubernetes
LoadBalancer vs Ingress
I don't know if I missed something, but I can't seem to find any posts/doc that is related to my question. Maybe I misunderstand the type ingress in kubernetes, but is it possible to define multiple ingresses that use the same LoadBlancer? Having to start one LoadBalancer for every ingress is costly.
One of the benefit of using ingress it helps to avoid creating an external LoadBalancer for each LoadBalancer type service. On many cloud providers some of the ingress controllers will create the corresponding external Load Balancer resource for each ingress resource. But using Nginx Ingress controller you need one loadBalancer to expose the Nginx Ingress controller itself. Then create multiple ingress resource and have multiple backends. All the backends are served by same external Load Balancer.
From the docs of Nginx Ingress
In this section you can find a common usage scenario where a single
load balancer powered by ingress-nginx will route traffic to 2
different HTTP backend services based on the host name
For a sample microservice based architecture deployed on Google kubernetes engine, I need help to validate my understanding :
We know services are supposed to load balance traffic for pod replicaset.
When we create an nginx ingress controller and ingress definitions to route to each service, a loadbalancer is also setup automatically.
had read somewhere that creating nginx ingress controller means an nginx controller (deployment) and a loadbalancer type service getting created behind the scene. I am not sure if this is true.
It seems loadbalancing is being done by services. URL based routing is
being done by ingress controller.
Why do we need a loadbalancer? It is not meant to load balance across multiple instances. It will just
forward all the traffic to nginx reverse proxy created and it will
route requests based on URL.
Please correct if I am wrong in my understanding.
A Service type LoadBalancer and the Ingress is the way to reach your application externally, although they work in a different way.
Service:
In Kubernetes, a Service is an abstraction which defines a logical set of Pods and a policy by which to access them (sometimes this pattern is called a micro-service). The set of Pods targeted by a Service is usually determined by a selector (see below for why you might want a Service without a selector).
There are some types of Services, and of them is the LoadBalancer type that permit you to expose your application externally assigning a externa IP for your service. For each LoadBalancer service a new external IP will be assign to it.
The load balancing will be handled by kube-proxy.
Ingress:
An API object that manages external access to the services in a cluster, typically HTTP.
Ingress may provide load balancing, SSL termination and name-based virtual hosting.
When you setup an ingress (i.e.: nginx-ingress), a Service type LoadBalancer is created for the ingress-controller pods and a Load Balancer in you cloud provider is automatically created and a public IP will be assigned for the nginx-ingress service.
This load balancer/public ip will be used for incoming connection for all your services, and nginx-ingress will be the responsible to handle the incoming connections.
For example:
Supose you have 10 services of LoadBalancer type: This will result in 10 new publics ips created and you need to use the correspondent ip for the service you want to reach.
But if you use a ingress, only 1 IP will be created and the ingress will be the responsible to handle the incoming connection for the correct service based on PATH/URL you defined in the ingress configuration. With ingress you can:
Use regex in path to define the service to redirect;
Use SSL/TLS
Inject custom headers;
Redirect requests for a default service if one of the service failed (default-backend);
Create whitelists based on IPs
Etc...
A important note about Ingress Load balancing in ingress:
GCE/AWS load balancers do not provide weights for their target pools. This was not an issue with the old LB kube-proxy rules which would correctly balance across all endpoints.
With the new functionality, the external traffic is not equally load balanced across pods, but rather equally balanced at the node level (because GCE/AWS and other external LB implementations do not have the ability for specifying the weight per node, they balance equally across all target nodes, disregarding the number of pods on each node).
An ingress controller(nginx for example) pods needs to be exposed outside the kubernetes cluster as an entry point of all north-south traffic coming into the kubernetes cluster. One way to do that is via a LoadBalancer. You could use NodePort as well but it's not recommended for production or you could just deploy the ingress controller directly on the host network on a host with a public ip. Having a load balancer also gives ability to load balance the traffic across multiple replicas of ingress controller pods.
When you use ingress controller the traffic comes from the loadBalancer to the ingress controller and then gets to backend POD IPs based on the rules defined in ingress resource. This bypasses the kubernetes service and load balancing(by kube-proxy at layer 4) offered by kubernetes service.Internally the ingress controller discovers all the POD IPs from the kubernetes service's endpoints and directly route traffic to the pods.
It seems loadbalancing is being done by services. URL based routing is being done by ingress controller.
Services do balance the traffic between pods. But they aren't accessible outside the kubernetes in Google Kubernetes Engine by default (ClusterIP type). You can create services with LoadBalancer type, but each service will get its own IP address (Network Load Balancer) so it can get expensive. Also if you have one application that has different services it's much better to use Ingress objects that provides single entry point. When you create an Ingress object, the Ingress controller (e.g. nginx one) creates a Google Cloud HTTP(S) load balancer. An Ingress object, in turn, can be associated with one or more Service objects.
Then you can get the assigned load balancer IP from ingress object:
kubectl get ingress ingress-name --output yaml
As a result your application in pods become accessible outside the kubernetes cluster:
LoadBalancerIP/url1 -> service1 -> pods
LoadBalancerIP/url2 -> service2 -> pods
This is more a design question than an issue. We have deployed in our company our own Kubernetes infrastructure and we are trying to use ingresses and NGINX ingress controller to externally expose our services, but since it is not a cloud environment such as GCP or AWS, we can't use service type "LoadBalancer". Should we just expose our ingress controller through a service type "NodePort"? Is that the normal way to go for production environments (non-cloud)?
From what I've read in another post, one suitable recommendation is to use NodePort, and manually point yet another external load balancer to the port on your Kubernetes nodes.
It just seems that exposing the ingress controller through this mechanism is somehow not very practical or robust (e.g. you don’t know what port your service is going to be allocated, and the port might get re-allocated at some point, etc.)
Is there any other mechanism maybe to expose the ingress controller to the external world?
The Loadbalancer service approach is one way to do it but behind it it's nothing more than a nodeport on the cluster.
Even if you use a service that create a LB on cloud provider, the LB needs to have a target port to communicate with the cluster.
When using a nginx-ingress that will mostly handle web requests, it's common usage to put an ingress in front of a nodeport service.
So with this I think using NodePort services is a good idea to do what you want ;)
This is my opinion, I'm interested if anyone else has another way to do it.
You can specify the port via nodePort in the service. Then it would not be random.