Container Insights for AKS

Container Insights is a feature of Azure Monitor that provides performance monitoring for containerized workloads deployed to Azure Kubernetes Service (AKS) clusters, Azure Container Instances (ACI), and self-managed Kubernetes clusters running on Azure Stack or on-premises. It collects memory and processor metrics from controllers, nodes, and containers, as well as container inventory and log data. The primary goal is to give you visibility into the health and performance of your Kubernetes infrastructure without requiring you to manually configure monitoring agents or build custom dashboards.

For the AZ-104 exam, you need to understand that Container Insights is the recommended monitoring solution for AKS clusters. It is not a management tool — it does not automate scaling or healing. It is purely an observability tool that provides data to Azure Monitor, which you can then use to create alerts, triggers, or integrate with other tools.

Container Insights works by deploying a containerized Log Analytics agent (the Azure Monitor agent for containers) as a DaemonSet on each node in your AKS cluster. The agent is based on the OMS (Operations Management Suite) agent and is responsible for collecting:

The agent sends this data to a Log Analytics workspace every 60 seconds by default. The workspace stores the data in tables such as Perf, ContainerInventory, ContainerLog, KubeEvents, and KubePodInventory. You can query these tables using Kusto Query Language (KQL) to gain insights.

The data collection pipeline: 1. The agent on each node queries the kubelet API for pod and container statistics. 2. The agent also reads container logs from the Docker socket or containerd socket. 3. The agent sends the data to the Log Analytics workspace via HTTPS on port 443. 4. The workspace ingests the data and makes it available for querying.

To enable Container Insights on an existing AKS cluster using Azure CLI:

az aks enable-addons --addons monitoring --name <cluster-name> --resource-group <rg-name> --workspace-resource-id <workspace-id>

To verify the agent is running:

kubectl get daemonsets --namespace=kube-system | grep omsagent

To check the agent logs:

kubectl logs -n kube-system -l app=omsagent

To query performance data in Log Analytics:

Perf
| where TimeGenerated > ago(1h)
| where ObjectName == "K8SNode"
| summarize avg(CounterValue) by Computer, CounterName

Container Insights integrates with:

Container Insights is a powerful monitoring tool for AKS that integrates seamlessly with Azure Monitor. For the AZ-104 exam, focus on understanding its purpose, components, configuration, and how it differs from other monitoring solutions. Remember that it is a monitoring-only feature, requires a Log Analytics workspace, uses the OMS agent as a DaemonSet, and collects data every 60 seconds.

A large e-commerce company runs its microservices on AKS across multiple nodes. They need to monitor resource usage to ensure consistent performance during peak shopping seasons. They enable Container Insights on their production cluster. The operations team creates dashboards showing CPU and memory usage per service. They set up alerts for when any container exceeds 80% CPU for more than 5 minutes. During a flash sale, an alert triggers — a payment service container is hitting 95% CPU. The team drills into the container logs via Container Insights and discovers a memory leak. They scale up the deployment and schedule a fix. Without Container Insights, they would have only noticed the issue when customers complained about payment failures.

A SaaS provider manages 50 AKS clusters across different regions. They need a centralized view of all clusters. They configure each cluster to send data to a single Log Analytics workspace. Using Container Insights, they create a workbook that aggregates metrics from all clusters, showing overall health, regional performance, and cost trends. They set up alerts for node failures and high memory usage. When a cluster in West US experiences a node failure, they get an alert and quickly redeploy the node. The centralized workspace also allows them to run cross-cluster queries to identify which clusters are underutilized, helping them optimize costs.

A development team deploys a new version of their app to a staging AKS cluster. After deployment, they notice increased latency. They use Container Insights to examine the cluster map and see that one node has significantly higher CPU usage than others. They drill into that node and see that a specific pod is consuming excessive memory. They check the container logs for that pod and find repeated error messages about a database connection timeout. The team realizes the new version has a bug causing excessive database queries. They roll back the deployment. Container Insights helped them pinpoint the root cause in minutes rather than hours.