Exam AI-100 topic 1 question 9 discussion

Is this part of AI-100? I don't remember learning anything related to this...

The question asks about nodes so the answer is Cluster Auto-scaler. If it asks about pods then the answer will be horizontal pod auto-scaler

To handle the unpredictable application load in an Azure Kubernetes Service (AKS) cluster that occasionally requires more than the maximum of 32 nodes, the recommended scaling method would be B. cluster autoscaler.

ACI: D is answer. https://learn.microsoft.com/en-us/azure/architecture/solution-ideas/articles/scale-using-aks-with-aci "Use the AKS virtual node to provision pods inside ACI 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 ACI, without any additional servers to manage"

autoscaler is to scale the cluster up/down. So as per the question it can't go beyond 32 nodes and the question is what you do when we need more than 32 nodes? surely autoscaler won't do this job.

Correct answer is cluster autoscaler. Ref : https://docs.microsoft.com/en-us/azure/aks/cluster-autoscaler#:~:text=cluster%20autoscaler%20settings.-,Update%20an%20existing%20AKS%20cluster%20to%20enable%20the%20cluster%20autoscaler,count%20and%20%2D%2Dmax%2Dcount%20.

can the answer be A?

The ans is (D)Azure container Instance: To rapidly scale your AKS cluster, you can integrate with Azure Container Instances (ACI). https://docs.microsoft.com/en-us/azure/aks/concepts-scale

My answer would be D. Azure Container Instances, since maximum of the cluster is 32 and "occasionally and unpredictably" we need resources above that value. Proof: https://azure.microsoft.com/en-us/services/container-instances/#overview Elastic bursting with AKS ACI provides fast, isolated compute to meet traffic that comes in spikes, without the need to manage servers. For example, Azure Kubernetes Service (AKS) can use the Virtual Kubelet to provision pods inside ACI 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 ACI without any additional servers to manage.

In the question: "The cluster supports a maximum of 32 nodes. You discover that occasionally and unpredictably, the application requires more than 32 nodes." To me, this is a clear indication an additional cluster is needed, therefore B.

“Cluster autoscaler is typically used alongside the horizontal pod autoscaler. When combined, the horizontal pod autoscaler increases or decreases the number of pods based on application demand, and the cluster autoscaler adjusts the number of nodes as needed to run those additional pods accordingly.”

The correct answer is B. If you check https://docs.microsoft.com/en-us/azure/aks/concepts-scale, particularly the section "Cluster Auto-scaler", you'll find the following excerpt: "If the cluster autoscale determines that a change is required, the number of nodes in your AKS cluster is increased or decreased accordingly" On the reason why A. Horizontal pod auto-scaler, it becomes clear on the same web page that this scaler is limited by the number of nodes in the cluster, therefore, not acceptable

I think the correct answers are "A" and "B". The horizontal pod auto-scaler increase/ decrease the number of replicas of an app that has been deployed, while the cluster auto-scaler increase/ decrease the number of nodes in the cluster. Any of these two options could fit well in the required scenario.

Please ignore, this was for the previous question

I think the answer is Visual Studio Code... https://code.visualstudio.com/docs