EKS Auto Mode Cost Management Quiz

Related Document: Cost Management

Multiple Choice Questions

1. Approximately what percentage of cost savings can be achieved by including Graviton (ARM) instances for cost optimization?

• A) 5%
• B) 10%
• C) 20%
• D) 50%

Show Answer

Answer: C) 20%

Explanation: AWS Graviton processor-based instances (arm64) are approximately 20% cheaper than comparable x86 instances while delivering excellent performance.

spec:
  template:
    spec:
      requirements:
        # Include Graviton (ARM) instances
        - key: kubernetes.io/arch
          operator: In
          values: ["amd64", "arm64"]

Cost Optimization Strategies:

• Using Graviton (ARM) instances: ~20% savings
• Using Spot instances: Up to 70-90% savings
• Aggressive Consolidation: Consolidate underutilized nodes
• Appropriate resource requests: Prevent overprovisioning

2. What does it mean when CPU limit is set to 500 in NodePool's limits setting?

• A) Maximum CPU per node is 500 cores
• B) NodePool can provision up to 500 vCPU total
• C) Maximum 500m CPU per Pod
• D) Cluster-wide CPU limit

Show Answer

Answer: B) NodePool can provision up to 500 vCPU total

Explanation: NodePool's limits restricts the total amount of resources that NodePool can provision.

apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: compute-optimized
spec:
  limits:
    cpu: 500      # Maximum 500 vCPU
    memory: 1Ti   # Maximum 1TB memory
  template:
    spec:
      requirements:
        - key: karpenter.k8s.aws/instance-category
          operator: In
          values: ["c"]

This helps prevent cost overruns and manage budgets.

3. What Consolidation policy automatically cleans up underutilized nodes to optimize costs?

• A) consolidationPolicy: WhenEmpty
• B) consolidationPolicy: WhenEmptyOrUnderutilized
• C) consolidationPolicy: Always
• D) consolidationPolicy: Aggressive

Show Answer

Answer: B) consolidationPolicy: WhenEmptyOrUnderutilized

Explanation: The WhenEmptyOrUnderutilized policy targets both empty and underutilized nodes for consolidation.

disruption:
  consolidationPolicy: WhenEmptyOrUnderutilized
  consolidateAfter: 5m

Policy Comparison:

• WhenEmpty: Only removes empty nodes (conservative, stability-first)
• WhenEmptyOrUnderutilized: Aggressive cost optimization

Cost savings effect: Consolidate underutilized nodes to reduce total node count and infrastructure costs

4. What is the recommended approach when using Savings Plans with EKS Auto Mode?

• A) Compute Savings Plans for instance family flexibility
• B) EC2 Instance Savings Plans for specific instance types
• C) Use only Reserved Instances
• D) Don't use Savings Plans

Show Answer

Answer: A) Compute Savings Plans for instance family flexibility

Explanation: Auto Mode uses various instance types depending on workloads, making Compute Savings Plans most suitable.

Savings Plans Comparison:

Type	Flexibility	Discount
Compute Savings Plans	Instance family, size, region, OS flexible	Up to 66%
EC2 Instance Savings Plans	Specific instance family fixed	Up to 72%

Recommended Strategy:
1. Analyze baseline workload volume (3-month baseline)
2. Cover 70% of minimum usage with Compute Savings Plans
3. Run remainder flexibly with Spot/On-Demand

5. What is required for Pod-level cost allocation in cost analysis using Kubecost?

• A) Automatically supported without configuration
• B) Add cost-center label to Pods
• C) Install Kubecost agent and cluster integration
• D) AWS Cost Explorer is sufficient

Show Answer

Answer: C) Install Kubecost agent and cluster integration

Explanation: Kubecost is a Kubernetes-native cost monitoring tool that provides detailed cost analysis.

# Kubecost installation (Helm)
helm repo add kubecost https://kubecost.github.io/cost-analyzer/
helm install kubecost kubecost/cost-analyzer \
    --namespace kubecost \
    --create-namespace

Kubecost features:

• Cost analysis by namespace/workload
• Resource efficiency recommendations
• Budget alert settings
• AWS integration for actual cost accuracy

6. What is the best practice to prevent overprovisioning when setting resource requests?

• A) Always set limits and requests equal
• B) Reference VPA recommendations based on actual usage
• C) Set requests as high as possible
• D) Omit requests setting

Show Answer

Answer: B) Reference VPA recommendations based on actual usage

Explanation: Use Vertical Pod Autoscaler (VPA) to analyze actual resource usage and set appropriate request values.

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: my-app-vpa
spec:
  targetRef:
    apiVersion: "apps/v1"
    kind: Deployment
    name: my-app
  updatePolicy:
    updateMode: "Off"  # Only provide recommendations, no auto-apply

Check recommendations:

kubectl describe vpa my-app-vpa

This enables:

• Prevent overprovisioning (cost savings)
• Prevent underprovisioning (performance guarantee)

7. What is the cost-efficient NodePool separation strategy for multi-tier workloads?

• A) Place all workloads in single NodePool
• B) Separate NodePool by tier (frontend/API/batch/ML)
• C) Separate NodePool by instance size
• D) Separate NodePool by availability zone

Show Answer

Answer: B) Separate NodePool by tier (frontend/API/batch/ML)

Explanation: Separating NodePools according to workload characteristics achieves optimal cost/performance balance.

Tier	Strategy	Expected Savings
Frontend	On-Demand + Graviton	~20%
API	Spot mixed + Graviton	~40%
Batch	Spot only + diversification	~70%
ML	Appropriate instance size selection	~30%

# Batch tier example - Maximum cost savings
apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: batch-tier
spec:
  template:
    spec:
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["spot"]  # Spot only
        - key: kubernetes.io/arch
          operator: In
          values: ["amd64", "arm64"]  # Include Graviton

8. What is needed to categorize EKS Auto Mode costs by tag in AWS Cost Explorer?

• A) All tags are automatically applied
• B) Need to set tags field in NodeClass
• C) Configure in AWS Organizations
• D) Tag-based categorization not possible

Show Answer

Answer: B) Need to set tags field in NodeClass

Explanation: Apply cost allocation tags to provisioned nodes through the tags field in NodeClass.

apiVersion: eks.amazonaws.com/v1
kind: NodeClass
metadata:
  name: production-nodeclass
spec:
  tags:
    Environment: production
    Team: platform
    CostCenter: engineering-001
    Project: kubernetes-platform

AWS Cost Explorer setup:

1. Enable Cost Allocation Tags in Billing Console
2. Select desired tag keys
3. Available in Cost Explorer after 24 hours