AWS Spot Instances are spare compute capacity offered by AWS at discounts of up to 90% compared to on-demand pricing. When running Kubernetes clusters, using spot instances can dramatically lower infrastructure costs. For example, a typical 10-node cluster in us-east-1 can save between $858.48 and $981.12 per month by switching from on-demand to spot instances (as of February 2025).
The primary risk is that AWS can reclaim spot instances with just two minutes' notice, potentially disrupting workloads. Not all workloads are suitable for spot instances—critical databases, payment systems, and applications with strict uptime requirements should avoid spot usage. Proper planning and resilient architecture are essential to mitigate these risks.
Workloads that are interruption-tolerant, stateless, or can recover state externally are ideal for spot instances. Examples include CI/CD pipelines, development and testing environments, batch processing, machine learning training jobs, and horizontally scalable stateless web applications.
Critical databases, payment processing systems, authentication services, session management, single-instance stateful applications, and workloads with strict timing or startup requirements should not run on spot instances due to their sensitivity to interruptions.
To use spot instances, configure your EKS cluster with dedicated spot node groups. Tools like Karpenter can automate node provisioning and handle spot interruptions. It's essential to set up Node Termination Handlers, monitoring, and resilient storage solutions to ensure workloads are gracefully drained and rescheduled during interruptions.
Karpenter is an open-source Kubernetes node provisioning tool that rapidly responds to spot instance interruptions by provisioning new nodes and maintaining workload availability. It automates instance selection, scaling, and interruption handling, making it easier to maximize cost savings while maintaining reliability.
The AWS Node Termination Handler monitors for spot instance interruption notices and ensures that Kubernetes workloads are gracefully drained before the instance is terminated. This helps maintain application availability and prevents data loss during spot instance reclamation.
Persistent storage should be configured to allow volumes to detach and reattach across nodes and availability zones. Using zone-aware StatefulSets and multi-AZ StorageClasses ensures data accessibility even when spot instances are interrupted and workloads are rescheduled.
Effective monitoring is crucial. Tools like Prometheus, AWS CloudWatch, and custom alerting for spot interruptions, price spikes, and capacity constraints are recommended. These tools provide early warnings and help maintain reliability during spot instance events.
Testing is essential. Use the AWS Fault Injection Simulator to trigger spot instance interruptions and validate that your workloads are rescheduled and data remains accessible. Regularly monitor and review failover scenarios to ensure resilience.
Delivery Hero transitioned 90% of its Kubernetes workloads to spot instances, achieving a 70% reduction in infrastructure costs and handling 4x-5x traffic spikes. ITV migrated 75% of workloads to EKS spot instances, reducing costs by 60% and deployment times from 40 to 4 minutes, saving $150,000 annually.
Best practices include starting with non-critical workloads, implementing comprehensive monitoring, using pod disruption budgets, maintaining instance type diversity, and regularly testing failover scenarios. Regular reviews and optimizations are also essential for ongoing success.
Choosing a diverse mix of instance types increases reliability by reducing the risk of capacity shortages. Automated tools like Karpenter or autonomous platforms can optimize instance selection based on cost, availability, and performance, adjusting in real time as conditions change.
Manual management requires hands-on selection and monitoring. Automated tools (like Karpenter) use rules to manage nodes and scaling. Autonomous platforms (like Sedai) use machine learning to analyze usage, predict needs, and implement optimizations automatically, maximizing savings and reliability with minimal intervention.
EKS Auto Mode requires separate node groups for spot and on-demand workloads. You cannot mix instance types within the same node group, so you must explicitly define which workloads run on spot instances by creating dedicated spot node groups.
Check that the AWS Node Termination Handler is correctly installed and configured. Ensure IAM permissions are set, and monitor logs for errors. Validate that your workloads have appropriate pod disruption budgets and that your cluster autoscaler is functioning as expected.
Recommended resources include the AWS Spot Instance Advisor, EKS Workshop Spot Guide, Karpenter Documentation, and the AWS Node Termination Handler GitHub repository.
Sedai's autonomous cloud optimization platform uses machine learning to analyze usage patterns, predict capacity needs, and automatically select the optimal mix of spot and on-demand instances. This maximizes cost savings, improves reliability, and reduces manual intervention for Kubernetes clusters on AWS.
Sedai provides autonomous optimization for cost, performance, and availability, proactive issue resolution, full-stack cloud coverage (compute, storage, data across AWS, Azure, GCP, Kubernetes), release intelligence, and plug-and-play implementation. It also supports multiple operation modes: Datapilot (observability), Copilot (one-click optimization), and Autopilot (fully autonomous execution).
Yes, Sedai integrates with Kubernetes autoscalers such as HPA/VPA and Karpenter, as well as monitoring tools like Prometheus, CloudWatch, Datadog, and Azure Monitor. This ensures seamless optimization and observability for Kubernetes environments.
Unlike traditional tools that rely on static rules or manual adjustments, Sedai uses machine learning to autonomously optimize cloud resources based on real application behavior. This results in up to 50% cost savings, 75% latency reduction, and 6X productivity gains, with minimal manual intervention.
Release Intelligence tracks changes in cost, latency, and errors for each deployment, helping teams improve release quality, minimize risks, and ensure smoother deployments. Companies like Freshworks have benefited from this feature to optimize their AWS Lambda platforms.
Sedai integrates with Infrastructure as Code (IaC), IT Service Management (ITSM), and compliance workflows. Every optimization is constrained, validated, and reversible, ensuring safe operations and compliance with enterprise-grade governance standards.
Sedai is SOC 2 certified, demonstrating adherence to stringent industry standards for data protection and compliance. More details are available on the Sedai Security page.
Sedai can reduce cloud costs by up to 50%, improve application performance by reducing latency up to 75%, and deliver up to 6X productivity gains by automating routine tasks. Customers like Palo Alto Networks saved $3.5 million, and KnowBe4 achieved 50% cost savings in production using Sedai.
Sedai is trusted by leading organizations such as Palo Alto Networks, HP, Experian, KnowBe4, Expedia, CapitalOne Bank, GSK, and Avis. These companies use Sedai to optimize cloud environments and improve operational efficiency.
Sedai's platform is used across industries including cybersecurity, IT, financial services, healthcare, travel, e-commerce, SaaS, and digital commerce. Case studies include Palo Alto Networks (cybersecurity), HP (IT), Experian (financial services), Expedia (travel), and Belcorp (retail/e-commerce).
KnowBe4 achieved 50% cost savings and saved $1.2 million on AWS bills. Palo Alto Networks saved $3.5 million and reduced Kubernetes costs by 46%. Belcorp reduced AWS Lambda latency by 77%. More case studies are available on the Sedai resources page.
Sedai is designed for platform engineers, IT/cloud operations, technology leaders (CTO, CIO, VP Engineering), site reliability engineers (SREs), and FinOps professionals in organizations with significant cloud operations across industries such as cybersecurity, IT, finance, healthcare, travel, and e-commerce.
Sedai's setup process is quick and efficient—just 5 minutes for general use cases and up to 15 minutes for specific scenarios like AWS Lambda. For complex environments, timelines may vary. Personalized onboarding and extensive documentation are available to support implementation.
Sedai offers plug-and-play implementation, agentless integration via IAM, and a 30-day free trial. Customers can access detailed documentation, a community Slack channel, and personalized onboarding sessions for a smooth start.
Sedai provides detailed technical documentation covering features, setup, and usage. Access it at docs.sedai.io/get-started. Additional resources, including case studies and datasheets, are available on the Sedai resources page.
Sedai integrates with monitoring tools (CloudWatch, Prometheus, Datadog, Azure Monitor), Kubernetes autoscalers (HPA/VPA, Karpenter), IaC and CI/CD tools (GitLab, GitHub, Bitbucket, Terraform), ITSM (ServiceNow, Jira), notification tools (Slack, Microsoft Teams), and various runbook automation platforms.
Sedai stands out with 100% autonomous optimization, proactive issue resolution, application-aware intelligence, full-stack cloud coverage, release intelligence, and rapid plug-and-play implementation. Unlike competitors that rely on static rules or manual adjustments, Sedai uses machine learning for continuous, outcome-focused optimization.
Sedai's unique features include 100% autonomous optimization, proactive issue resolution before user impact, application-aware intelligence, release intelligence, and a quick 5–15 minute setup. These capabilities enable measurable cost savings, performance improvements, and operational efficiency not commonly found in other solutions.
Sedai addresses pain points such as cost inefficiencies, operational toil, performance and latency issues, lack of proactive issue resolution, complexity in multi-cloud environments, and misaligned priorities between engineering and FinOps teams. It automates routine tasks, aligns objectives, and ensures reliability and cost efficiency.
Sedai provides actionable insights and autonomous optimization that balance performance and cost efficiency. By automating optimizations and providing visibility into cost and performance impacts, Sedai helps engineering and FinOps teams work toward shared objectives.
John Jamie
Content Writer
February 7, 2025
Featured
Running Kubernetes clusters on spot instances offers substantial cost savings while introducing unique operational considerations. This guide provides a comprehensive approach to implementing and managing Kubernetes workloads on AWS spot instances, combining strategic planning with practical implementation details.
AWS spot instances offer discounts of up to 90% compared to on-demand pricing by utilizing excess AWS capacity. This significant cost reduction comes with the understanding that instances can be reclaimed with just two minutes' notice when AWS needs the capacity back.
Typical cluster configuration:- Region: us-east-1- Instance types: m5.large, m5a.large, m4.large- Base cluster: 10 nodes- Average monthly utilization: 80%
Monthly Cost Comparison:On-Demand m5.large (10 nodes): $1,226.40Spot Instance (mixed): $245.28 - $367.92Potential monthly savings: $858.48 - $981.12
Note: Prices as of February 2025
Not all Kubernetes workloads are suitable for spot instances. While the cost savings are attractive, the potential for instance interruption means you need to carefully evaluate which workloads can handle occasional disruption. Understanding these characteristics is crucial for successful spot instance implementation, as placing the wrong workload on spot instances can lead to service disruptions and reliability issues.
1. Interruption Tolerance - Can handle occasional restarts - Maintain state externally or can recover state - Have reasonable startup times - Can be rescheduled to different nodes
2. Flexible Timing - Not strictly time-critical - Can retry failed operations - Have built-in fault tolerance
Development and Testing- CI/CD pipelines- Development environments- Load testing environments- QA clusters- Integration testing environments
Production Workloads- Stateless web applications- Background job processors- Batch processing systems- Data analysis workloads- Machine learning training jobs- Horizontally scalable services
Not Recommended for Spot- Critical databases- Payment processing systems- Authentication services- Session management services- Single-instance stateful applications- Single-replica deployments- Applications with startup times exceeding 2 minutes- Applications requiring guaranteed graceful shutdown- Workloads with strict timing requirements- Services that cannot tolerate occasional restarts
Successfully running Kubernetes on spot instances requires careful configuration of multiple components. Each component plays a crucial role in managing spot instances effectively, from handling interruptions to ensuring proper scaling. The following sections detail the essential configurations needed, walking through the setup of cluster components, node management tools, monitoring, and volume handling. Follow these implementations in order, as later configurations often depend on earlier ones.
Karpenter provides advanced node provisioning capabilities that are particularly valuable for spot instance management. Unlike traditional auto-scaling, Karpenter can rapidly respond to spot instance interruptions and maintain workload availability through intelligent node provisioning. The following configurations demonstrate how to set up Karpenter to effectively manage spot instances while maximizing cost savings and maintaining reliability.
The AWS Node Termination Handler is a critical component for managing spot instance lifecycles. It monitors for spot instance interruption notices and ensures workloads are gracefully drained before an instance is terminated. This handler is essential for maintaining application availability during spot instance reclamation. The following configuration sets up robust interruption handling with monitoring and notification capabilities.
Effective monitoring is essential when running Kubernetes on spot instances. Without proper monitoring, you might miss critical events like imminent instance terminations or capacity constraints. The following monitoring configuration provides early warning systems for spot instance interruptions, price spikes, and capacity issues, allowing you to take proactive measures before problems affect your applications.
Managing persistent storage with spot instances presents unique challenges. When instances are interrupted, you need to ensure that your persistent volumes can be properly detached and reattached to new nodes, potentially in different availability zones. The following configurations demonstrate how to set up resilient storage solutions that can handle spot instance interruptions while maintaining data accessibility.
Testing your spot instance configuration is crucial for ensuring your applications can handle interruptions gracefully. Simply setting up the components isn't enough; you need to verify that your system responds correctly to spot instance reclamation. The following section provides tools and procedures for testing your spot instance setup, including simulated interruptions and monitoring practices.
Understanding how organizations successfully implement Kubernetes on spot instances provides valuable insights for your own implementation. Here we examine two major companies that have achieved significant success with spot instances.
Delivery Hero, one of the world's largest food delivery networks, successfully transitioned their entire Kubernetes infrastructure to spot instances, demonstrating that spot instances can work at massive scale.
- Complete transition to spot instances within 6 months- 90% of Kubernetes workloads running on Amazon EKS- Focus on application resilience and high availability
- Multiple instance redundancy - Graceful termination scripts - Production-ready checklists - Termination notice handlers - De-scheduler implementation
- 70% reduction in infrastructure costs- Successfully handling 4x-5x traffic spikes- Managing 390 different applications across 43 countries- Improved focus on business innovation
ITV, the UK's largest commercial broadcaster, implemented spot instances to handle growing viewership while optimizing costs during the pandemic.
- 18-month phased migration - 75% workload migration to EKS - Incremental spot instance adoption
- 60% cost reduction compared to on-demand- $150,000 annual compute savings- Deployment time reduced from 40 to 4 minutes- Increased spot usage from 9% to 24%
- Start with non-critical workloads - Create comprehensive checklists - Focus on application resilience - Implement proper monitoring
- Use mixed instance types - Implement robust auto-scaling - Focus on graceful termination - Maintain redundancy
- Regular review of spot usage - Continuous optimization - Strong monitoring practices - Clear incident response procedures
- Clear migration strategy - Focus on application resilience - Proper auto-scaling implementation - Comprehensive monitoring - Regular optimization reviews
These case studies demonstrate that with proper planning and implementation, spot instances can successfully support large-scale production workloads while delivering significant cost savings, regardless of industry or scale.
Successfully running Kubernetes on spot instances requires more than just initial setup - it requires ongoing operational excellence. These best practices have been gathered from real-world experience running production workloads on spot instances. Following these guidelines will help you maintain reliability while maximizing cost savings.
Choosing the right mix of instance types is a critical success factor for spot instance implementations. There are three main approaches to instance type selection, each offering different levels of operational efficiency:
Traditional hands-on approach using AWS tools to analyze historical pricing and availability data across different instance types and availability zones. This method requires regular monitoring and manual adjustments but provides full control over instance selection:
Tools like Karpenter and AWS Auto Scaling can automatically select instance types based on predefined rules and policies. This approach reduces manual intervention while maintaining control through configuration:
Autonomous cloud optimization tools like Sedai can evaluate the full array of AWS instance types and their costs and recommend (in Copilot mode) or implement (in Autopilot mode) optimal instance selections. These systems use machine learning to:
Similar to instance selection, capacity monitoring and management can be approached at different levels of automation:
Basic Prometheus rules for alerting on capacity issues, requiring manual intervention when problems are detected:
Cluster Autoscaler or Karpenter configurations that automatically respond to capacity changes:
Advanced autonomous platforms can:
EKS Auto Mode simplifies Kubernetes node management but requires careful consideration when using spot instances. While Auto Mode supports spot instances, it requires separate node groups for spot and on-demand workloads - you cannot mix instance types within the same node group. This means you'll need to explicitly define which workloads run on spot instances by creating dedicated spot node groups, as shown in the following configuration example:
Even with proper configuration and monitoring, issues can arise when running Kubernetes on spot instances. The following section covers common problems you might encounter and provides step-by-step resolution procedures. Understanding these troubleshooting patterns will help you maintain system reliability and quickly resolve issues when they occur.
Running Kubernetes on AWS spot instances requires careful planning and robust operational practices. By following this guide's configurations and monitoring recommendations, organizations can achieve significant cost savings while maintaining reliability. Remember to:
1. Start with non-critical workloads2. Implement comprehensive monitoring3. Use pod disruption budgets4. Maintain instance type diversity5. Regular testing of failover scenarios
