EKS Auto Mode Spot Strategies Quiz

Related Document: Spot Instance Strategies

Multiple Choice Questions

1. What is the optimal strategy to distribute Spot instance interruption risk?

• A) Use only a single instance type
• B) Use diverse instance families, generations, and sizes
• C) Use only On-Demand
• D) Select only the cheapest instances

Show Answer

Answer: B) Use diverse instance families, generations, and sizes

Explanation: Spot instances experience interruptions per capacity pool. Allowing diverse instance types enables acquiring instances from multiple capacity pools, distributing the interruption risk.

spec:
  template:
    spec:
      requirements:
        # Diverse instance families
        - key: karpenter.k8s.aws/instance-category
          operator: In
          values: ["m", "c", "r", "i", "d"]
        # Diverse generations
        - key: karpenter.k8s.aws/instance-generation
          operator: In
          values: ["5", "6", "7"]
        # Diverse sizes
        - key: karpenter.k8s.aws/instance-size
          operator: In
          values: ["large", "xlarge", "2xlarge"]
        # Diverse architectures
        - key: kubernetes.io/arch
          operator: In
          values: ["amd64", "arm64"]

2. What is the Karpenter label key that distinguishes between Spot and On-Demand instances?

• A) node.kubernetes.io/capacity-type
• B) karpenter.sh/capacity-type
• C) eks.amazonaws.com/instance-type
• D) karpenter.k8s.aws/spot-or-ondemand

Show Answer

Answer: B) karpenter.sh/capacity-type

Explanation: This label allows specifying Spot/On-Demand instances in Pod nodeAffinity or NodePool requirements.

# Setting Spot instance preference in Pod
affinity:
  nodeAffinity:
    preferredDuringSchedulingIgnoredDuringExecution:
      - weight: 100
        preference:
          matchExpressions:
            - key: karpenter.sh/capacity-type
              operator: In
              values: ["spot"]

3. What is the default warning time given before a Spot instance is interrupted?

• A) 30 seconds
• B) 2 minutes
• C) 5 minutes
• D) 10 minutes

Show Answer

Answer: B) 2 minutes

Explanation: AWS Spot instances are given a 2-minute warning before being reclaimed. Workloads must terminate gracefully during this time.

Spot Interrupt Handling Best Practices:

• Set Pod's terminationGracePeriodSeconds to 2 minutes or less
• Implement SIGTERM handler in applications
• Prefer stateless workloads
• Implement checkpointing mechanism (for batch jobs)

4. Which workload type is NOT recommended for Spot instances?

• A) Batch processing jobs
• B) Stateless web servers
• C) Single-instance databases
• D) Development/test environments

Show Answer

Answer: C) Single-instance databases

Explanation: Single-instance databases can experience availability issues during interrupts, making them unsuitable for Spot.

Workloads suitable for Spot:

• Batch processing / Big data analytics
• CI/CD pipelines
• Stateless web servers (Auto Scaling)
• Development/test environments
• Container-based microservices

Workloads requiring On-Demand:

• Databases
• Message queues
• Cluster management components
• Long-running stateful jobs

5. How do you configure Spot-first selection when mixing Spot and On-Demand in NodePool?

• A) spotPriority: high
• B) NodePool priority setting via weight value
• C) capacityPriority: spot
• D) preferSpot: true

Show Answer

Answer: B) NodePool priority setting via weight value

Explanation: Create multiple NodePools and specify priority with weight values. Higher weight is used first.

# Spot-first NodePool (weight: 100)
apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: spot-first
spec:
  weight: 100  # High priority
  template:
    spec:
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["spot"]
---
# On-Demand fallback NodePool (weight: 10)
apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: ondemand-fallback
spec:
  weight: 10  # Low priority
  template:
    spec:
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["on-demand"]

6. What is the maximum savings rate for Spot instances compared to On-Demand?

• A) 30-40%
• B) 50-60%
• C) 70-90%
• D) 95% or more

Show Answer

Answer: C) 70-90%

Explanation: Spot instances can achieve up to 70-90% cost savings compared to On-Demand.

Cost Optimization Strategy Combinations:

Strategy	Expected Savings
Spot instances	70-90%
Graviton (ARM)	~20%
Spot + Graviton	Up to 90%

However, Spot savings rates vary depending on instance type and availability zone.