Spot Optimizer¶
Spot Optimizer identifies workloads that can safely run on Spot (preemptible) instances and provides guidance to migrate them — with interruption handling and fallback strategies built in.
When to use it¶
- You want to reduce compute costs by using discounted Spot capacity.
- You need to evaluate which workloads are suitable for Spot without risking production stability.
- You want guardrails and fallback plans for Spot usage.
Open Spot Optimizer¶
Go to Optimization > Spot Optimizer from product navigation.
Evaluate workload suitability¶
Spot Optimizer analyzes workloads based on:
- Statelessness — can the workload resume without data loss if interrupted?
- Fault tolerance — does the application handle instance termination gracefully?
- Interruption history — how often has this Spot pool been interrupted?
Workloads are rated for Spot suitability. Do not move production-critical, stateful workloads to Spot without a tested fallback strategy.
Plan a Spot migration¶
- Review the workload list and identify High suitability candidates.
- Open a workload to review its suitability rating, estimated savings, and recommended instance types.
- Plan Availability Zone (AZ) diversification to reduce interruption risk.
- Configure fallback to On-Demand if Spot capacity is unavailable.
- Implement with your infrastructure team and test interruption handling before production use.
Best practices¶
- Start with batch, CI/CD, dev, or test workloads — these tolerate interruption best.
- Use multiple instance types and AZs to improve Spot availability.
- Implement checkpointing or retry logic in stateful workloads before moving to Spot.
- Monitor Spot interruption rates after migration and adjust pool selection if needed.
Related pages¶
- Commitment Optimizer — combine Spot with Reserved Instance coverage
- Recommendations — broader list of compute optimization opportunities