Is it possible to scale up pods dynamically based on input parameters in AKS?

user_2092282 0
Viewed 233

I'm new in the Kubernetes universe and I have some doubts about an implementation that I want to do.

I have the following scenario: I have 200 instances of a worker that executes some business logic and the unique thing that differentiate them is the input parameters.

I was thinking in to use AKS to scale up this infrastructure according to the input parameter and dynamically, only create a new pod when exists the demand for the worker with the input parameter "XYZ".

Simple architecture draft:

enter image description here

I have an API that receives a request and with base in this request, an orchestrator send the request for the correct worker.

So I'd like to know if this type of architecture is possible with AKS and if is a good approach.

azure-aks kubernetes azure
2 Answers

This is one of the scenario where you can use Azure Functions with ACI or with KEDA to autoscore the containers based on the demand.

Use the AKS virtual node to provision pods inside Azure Container Instances that start in seconds. This enables AKS to run with just enough capacity for your average workload. As you run out of capacity in your AKS cluster, scale out additional pods in Azure Container Instances without additional servers to manage.

Here is my blog on Scale Applications with Kuberenetes-based Event Driven AutoScaling

user_17494030
user_113656910
Related