Cost optimization in Kubernetes - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding Kubernetes Resource Requests and Limits

Which statement best explains how setting resource requests and limits in Kubernetes helps optimize costs?

A Requests and limits automatically scale the number of nodes in the cluster to match pod demand, reducing idle nodes.

B Requests reserve resources for pods, ensuring they get guaranteed CPU and memory, preventing overcommitment and reducing wasted resources.

C Limits allow pods to use unlimited resources, which helps avoid throttling and improves performance at no extra cost.

D Setting requests and limits disables autoscaling, which saves costs by fixing resource usage.

💡 Hint

Think about how reserving resources affects scheduling and cluster utilization.

Resource requests reserve CPU and memory for pods, so the scheduler places pods on nodes that have enough free resources. This prevents overcommitment and helps avoid wasted resources, which saves costs. Limits cap resource usage but do not reserve resources.

💻 Command Output

intermediate

Identifying Idle Nodes with kubectl

You run the command kubectl top nodes and see the following output:

NAME           CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%
node-1         50m          5%     200Mi          10%
node-2         10m          1%     100Mi          5%
node-3         300m         30%    1Gi            50%

Which node is the best candidate for scaling down to save costs?

A node-2 because it has the lowest CPU and memory usage.

B node-3 because it uses the most CPU and memory.

C node-1 because it has moderate CPU and memory usage.

D None, all nodes are equally used.

💡 Hint

Look for the node with the least resource usage to identify idle nodes.

Node-2 has the lowest CPU and memory usage, indicating it is underutilized and a good candidate for scaling down to reduce costs.

🔀 Workflow

advanced

Implementing Cluster Autoscaler for Cost Savings

You want to enable automatic scaling of your Kubernetes cluster nodes to save costs during low demand periods. Which configuration step is essential to allow the Cluster Autoscaler to remove nodes safely?

A Manually delete pods before the autoscaler removes nodes.

B Disable resource requests and limits on all pods to allow flexible scheduling.

C Set pod disruption budgets to ensure critical pods are not evicted during node scale-down.

D Set all nodes to unschedulable to prevent new pods from being scheduled.

💡 Hint

Think about how to protect important pods during node removal.

Pod Disruption Budgets (PDBs) define how many pods can be down during voluntary disruptions like node scale-down. This ensures critical pods remain available and the autoscaler can safely remove nodes without causing downtime.

❓ Troubleshoot

advanced

Troubleshooting High Costs Due to Overprovisioning

Your Kubernetes cluster costs are unexpectedly high. You suspect overprovisioning. Which of the following is the most likely cause?

A Horizontal Pod Autoscaler is scaling pods based on CPU usage.

B Pods have no resource requests or limits, so the scheduler packs them tightly on nodes.

C Cluster Autoscaler is enabled and removing unused nodes automatically.

D Pods have resource requests set too high, causing the scheduler to reserve more node capacity than needed.

💡 Hint

Consider how resource requests affect node allocation.

If pods request more CPU or memory than they actually use, the scheduler reserves more node resources, leading to more nodes being needed and higher costs.

✅ Best Practice

expert

Choosing the Best Strategy for Cost Optimization

Which strategy provides the most effective cost optimization in a Kubernetes environment running variable workloads?

A Use Cluster Autoscaler with right-sized resource requests and limits, combined with Horizontal Pod Autoscaler for workload scaling.

B Set fixed large resource requests for all pods to avoid scheduling delays and keep nodes fully utilized.

C Disable autoscaling and manually add nodes during peak hours to control costs.

D Run all workloads on a single large node to minimize node count.

💡 Hint

Think about combining multiple scaling mechanisms and resource management.

Combining Cluster Autoscaler to adjust node count, proper resource requests and limits to avoid waste, and Horizontal Pod Autoscaler to scale pods based on demand provides the best cost optimization for variable workloads.