Based on the official AWS announcement

What Are Interruptible Capacity Reservations?

On November 25, 2024, AWS announced a new type of EC2 Capacity Reservation called Interruptible Capacity Reservations. This new option provides guaranteed capacity at a significantly lower price point than traditional On-Demand Capacity Reservations (ODCRs), with savings of up to 47%.

The key difference? AWS can reclaim this capacity with a 2-minute warning when they need it for higher-priority workloads. In exchange for accepting this interruption risk, you get substantial cost savings while still maintaining capacity assurance when you need it.

How It Works

Interruptible Capacity Reservations function similarly to standard On-Demand Capacity Reservations, but with one critical distinction:

  • Guaranteed Capacity: When you create an Interruptible Capacity Reservation, AWS guarantees that capacity is available for your use
  • 2-Minute Warning: If AWS needs to reclaim the capacity, you receive a 2-minute notification via Amazon EventBridge
  • Automatic Termination: After the 2-minute warning, any instances using the reserved capacity are automatically terminated
  • Cost Savings: You pay up to 47% less than standard On-Demand Capacity Reservations

Interruption Notification

When AWS needs to reclaim capacity, you'll receive an EventBridge event with the following information:

  • Capacity Reservation ID being interrupted
  • Timestamp of when interruption will occur (2 minutes from notification)
  • Reason for interruption

This gives you a brief window to gracefully shut down workloads, save state, or trigger failover to other capacity.

Cost Comparison

Let's look at how Interruptible Capacity Reservations compare to other EC2 pricing options:

Option Capacity Guarantee Interruption Risk Relative Cost
On-Demand No None Baseline (100%)
On-Demand Capacity Reservation Yes None Same as On-Demand
Interruptible Capacity Reservation Yes 2-min warning Up to 47% less
Spot Instances No 2-min warning Up to 90% less

When to Use Interruptible Capacity Reservations

Interruptible Capacity Reservations are ideal for specific use cases where you need capacity assurance but can tolerate interruptions:

Good Use Cases

  • Batch Processing Jobs: Workloads that can checkpoint and resume, like data processing pipelines or rendering jobs
  • Development/Test Environments: Non-production workloads where brief interruptions are acceptable
  • Stateless Applications: Applications that can quickly failover to other instances without data loss
  • Scheduled Workloads: Jobs that run during specific time windows and can be rescheduled if interrupted
  • Auto-Scaling Groups: Applications with multiple instances where losing one instance doesn't impact availability

Not Recommended For

  • Production Databases: Interruptions could cause data loss or service disruption
  • Single-Instance Applications: No redundancy means interruption = downtime
  • Long-Running Stateful Processes: Processes that can't checkpoint and resume easily
  • Real-Time Applications: Services requiring consistent uptime (use Reserved Instances or On-Demand instead)

How to Get Started

Creating an Interruptible Capacity Reservation is straightforward through the AWS Console, CLI, or API:

Via AWS Console

  1. Navigate to EC2 → Capacity Reservations
  2. Click "Create Capacity Reservation"
  3. Select "Interruptible" as the reservation type
  4. Choose your instance type, availability zone, and quantity
  5. Review the discounted pricing
  6. Create the reservation

Via AWS CLI

aws ec2 create-capacity-reservation \
    --instance-type m5.large \
    --instance-platform Linux/UNIX \
    --availability-zone us-east-1a \
    --instance-count 10 \
    --instance-match-criteria targeted \
    --capacity-reservation-type interruptible

Setting Up EventBridge Notifications

To handle interruptions gracefully, set up an EventBridge rule to capture interruption warnings:

{
  "source": ["aws.ec2"],
  "detail-type": ["EC2 Capacity Reservation Interruption Warning"],
  "detail": {
    "capacity-reservation-id": ["cr-xxxxx"]
  }
}

You can route these events to Lambda functions, SNS topics, or SQS queues to trigger automated responses like:

  • Gracefully shutting down applications
  • Saving application state to S3 or DynamoDB
  • Launching replacement instances in other availability zones
  • Sending alerts to operations teams

Best Practices

To maximize the value of Interruptible Capacity Reservations while minimizing disruption:

1. Design for Interruption

  • Implement checkpointing in your applications
  • Use stateless architectures where possible
  • Store critical data in persistent storage (EBS, S3, RDS)
  • Test your interruption handling regularly

2. Combine with Other Strategies

  • Use Interruptible Capacity Reservations for burst capacity
  • Maintain baseline capacity with Reserved Instances or Savings Plans
  • Implement multi-AZ deployments for high availability
  • Consider Spot Instances for even greater savings on fault-tolerant workloads

3. Monitor and Optimize

  • Track interruption frequency using CloudWatch metrics
  • Monitor cost savings vs. operational overhead
  • Adjust reservation sizes based on actual usage patterns
  • Review interruption logs to identify patterns

4. Automate Response to Interruptions

  • Create Lambda functions to handle EventBridge notifications
  • Implement automatic failover to backup capacity
  • Use Auto Scaling to replace interrupted instances
  • Set up alerting for operations teams

Real-World Example: Batch Processing Pipeline

Let's look at a practical example of how a data analytics company could use Interruptible Capacity Reservations:

Scenario: A company runs nightly ETL jobs processing customer data. Jobs typically take 4-6 hours and can checkpoint every 30 minutes.

Previous Setup:

  • 20x m5.4xlarge On-Demand instances
  • Cost: ~$1.536/hour × 20 instances × 6 hours = $184.32/night
  • Monthly cost: ~$5,530

New Setup with Interruptible Capacity Reservations:

  • 20x m5.4xlarge Interruptible Capacity Reservations
  • Cost: ~$0.814/hour × 20 instances × 6 hours = $97.68/night (47% savings)
  • Monthly cost: ~$2,930
  • Monthly savings: $2,600

Implementation:

  1. Modified ETL jobs to checkpoint state every 30 minutes to S3
  2. Set up EventBridge rule to capture interruption warnings
  3. Created Lambda function to save current state and launch replacement instances
  4. Configured Auto Scaling to maintain desired capacity

Results:

  • Interruptions occur ~2-3 times per month
  • Average recovery time: 5 minutes (load checkpoint + resume)
  • Net savings: $2,600/month with minimal operational overhead

Comparison with Spot Instances

You might be wondering: "How is this different from Spot Instances?" Here's a detailed comparison:

Feature Interruptible Capacity Reservations Spot Instances
Capacity Guarantee Yes - guaranteed when reserved No - subject to availability
Pricing Up to 47% off On-Demand Up to 90% off On-Demand
Interruption Warning 2 minutes 2 minutes
Interruption Frequency Lower (AWS reclaims only when needed) Higher (market-driven)
Best For Workloads needing capacity assurance Highly flexible, fault-tolerant workloads

Key Takeaway: Use Interruptible Capacity Reservations when you need guaranteed capacity but can tolerate occasional interruptions. Use Spot Instances when you want maximum savings and can handle frequent interruptions or lack of availability.

Regional Availability

Interruptible Capacity Reservations are currently available in the following AWS regions:

  • US East (N. Virginia, Ohio)
  • US West (Oregon, N. California)
  • Europe (Ireland, Frankfurt, London, Paris, Stockholm)
  • Asia Pacific (Singapore, Sydney, Tokyo, Seoul, Mumbai)
  • Canada (Central)
  • South America (São Paulo)

AWS is continuously expanding availability to additional regions. Check the AWS What's New page for the latest updates.

Conclusion

Interruptible Capacity Reservations represent a compelling new option in AWS's EC2 pricing portfolio. By accepting a small amount of interruption risk, you can achieve significant cost savings (up to 47%) while still maintaining capacity guarantees.

This option fills an important gap between traditional On-Demand Capacity Reservations (expensive but never interrupted) and Spot Instances (cheap but no capacity guarantee). For workloads that can handle occasional interruptions with proper automation, Interruptible Capacity Reservations offer an excellent balance of cost and reliability.

Key Recommendations:

  • Start with non-critical workloads to test interruption handling
  • Implement robust checkpointing and state management
  • Set up EventBridge notifications and automated responses
  • Monitor interruption frequency and adjust strategy accordingly
  • Combine with other cost optimization strategies for maximum savings

As AWS continues to innovate on pricing models, staying informed about new options like Interruptible Capacity Reservations can help you optimize your cloud costs without sacrificing the reliability your business needs.