--- title: Kubernetes Spot Instances: How to Cut Compute Costs... description: Kubernetes Spot Instances offer 60-90% compute cost savings but require careful configuration to avoid reliability issues. Teams running Spot-heavy setups save... canonical: https://daily.dev/posts/kubernetes-spot-instances-how-to-cut-compute-costs-without-gambling-on-reliability-izpc2uk55 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Kubernetes Spot Instances: How to Cut Compute Costs Without Gambling on Reliability | daily.dev og:description: Kubernetes Spot Instances offer 60-90% compute cost savings but require careful configuration to avoid reliability issues. Teams running Spot-heavy setups save... og:url: https://daily.dev/posts/kubernetes-spot-instances-how-to-cut-compute-costs-without-gambling-on-reliability-izpc2uk55 og:image: https://api.daily.dev/og/posts/izpc2Uk55.png og:image:alt: Kubernetes Spot Instances: How to Cut Compute Costs Without Gambling on Reliability og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Kubernetes Spot Instances: How to Cut Compute Costs Without Gambling on Reliability **[Cast AI](https://daily.dev/sources/castai)** · 14 min read · 0 upvotes · 0 comments ## Summary Kubernetes Spot Instances offer 60-90% compute cost savings but require careful configuration to avoid reliability issues. Teams running Spot-heavy setups save 77% on average vs 59% for mixed fleets, per the 2025 Cast AI Kubernetes Cost Benchmark. Key safety mechanisms include PodDisruptionBudgets (critical to prevent simultaneous replica eviction), correct terminationGracePeriodSeconds tuned per cloud (90s for AWS, 25s for GCP/Azure), and AWS Node Termination Handler configured for your node group type. Karpenter users get native Spot interruption handling without NTH, plus SpotToSpotConsolidation (v0.34.0+) for continuous cost optimization. Workload suitability matters: stateless services, batch jobs, CI/CD runners, and checkpointed ML training are safe; stateful databases, control-plane components, and payment processors are not. The post also covers cloud-specific Spot market behavior, GCP Preemptible vs Spot VM differences, and how Cast AI automates predictive rebalancing 1-3 hours ahead of interruptions. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Where should terminationGracePeriodSeconds be set for Kubernetes pods running on spot instances, and does placement matter? It must be set at the pod spec level, not inside the container spec. Placing it inside the container spec is a common YAML mistake that causes Kubernetes to silently ignore the value and fall back to the default 30 seconds, which can be too short for AWS's 2-minute or too long for GCP/Azure's roughly 30-second spot interruption notice windows. _Teams tuning spot instance shutdown timing can track Kubernetes configuration gotchas like this on daily.dev._ ### Do I need to run AWS Node Termination Handler alongside Karpenter for spot instance interruptions? No, Karpenter handles spot interruptions natively and NTH is unnecessary when using it. Karpenter watches EventBridge and IMDS for interruption notices, cordons the affected node, and provisions a replacement without a separate DaemonSet. NTH is only needed for clusters using Cluster Autoscaler or self-managed node groups, where IMDS polling mode suits self-managed groups and Queue Processor mode (via EventBridge and SQS) suits EKS managed node groups. _Engineers weighing Karpenter against Cluster Autoscaler for spot handling can follow updates like this on daily.dev._ ### What is the difference between GCP Preemptible VMs and GCP Spot VMs for Kubernetes workloads? Preemptible VMs use an older API with a hard 24-hour maximum runtime and an effective hard shutdown around 15 seconds despite a 30-second notice, while Spot VMs are the current recommendation with no 24-hour limit and the same 30-second best-effort notice. New GKE workloads should use Spot VMs instead of Preemptible VMs to avoid the runtime cap and tighter shutdown window. _Developers choosing between GCP VM types for cost savings can track these distinctions on daily.dev._ ## Similar posts on daily.dev - [Karpenter and Spot Instances: Cut Compute Cost Without Gambling on Reliability](https://daily.dev/posts/karpenter-and-spot-instances-cut-compute-cost-without-gambling-on-reliability-i83oaztmh) · Cast AI · 0 upvotes · 0 comments - [How Do I Reduce Kubernetes Costs Without Hurting Reliability?](https://daily.dev/posts/how-do-i-reduce-kubernetes-costs-without-hurting-reliability--11xjt53gx) · Fairwinds Blog · 1 upvotes · 0 comments - [Tier Your Apps, Cut Your Costs: A Practical Framework for Spot Instances in Production](https://daily.dev/posts/tier-your-apps-cut-your-costs-a-practical-framework-for-spot-instances-in-production-txov8fzqw) · Cast AI · 0 upvotes · 0 comments - [Best of 2025: CAST AI Report Surfaces Massive Kubernetes Cloud Infrastructure Waste](https://daily.dev/posts/best-of-2025-cast-ai-report-surfaces-massive-kubernetes-cloud-infrastructure-waste-q4oam1c7g) · Cloud Native Now · 0 upvotes · 0 comments - [Intelligent Spot Instance Availability: How Machine Learning Reduces Interruptions by up to 94%](https://daily.dev/posts/intelligent-spot-instance-availability-how-machine-learning-reduces-interruptions-by-up-to-94--jvmxlkdvd) · Cast AI · 1 upvotes · 0 comments --- Tags: [#kubernetes](https://daily.dev/tags/kubernetes), [#finops](https://daily.dev/tags/finops) [View this post on daily.dev](https://daily.dev/posts/kubernetes-spot-instances-how-to-cut-compute-costs-without-gambling-on-reliability-izpc2uk55) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Kubernetes Spot Instances: How to Cut Compute Costs Without Gambling on Reliability","url":"https://daily.dev/posts/kubernetes-spot-instances-how-to-cut-compute-costs-without-gambling-on-reliability-izpc2uk55","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/kubernetes-spot-instances-how-to-cut-compute-costs-without-gambling-on-reliability-izpc2uk55"},"datePublished":"2026-08-10T13:31:49.317Z","dateModified":"2026-09-14T09:08:19.849Z","description":"Kubernetes Spot Instances offer 60-90% compute cost savings but require careful configuration to avoid reliability issues. Teams running Spot-heavy setups save...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b922f8333d9b4ffc13cf9fd83e4eb91e?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b922f8333d9b4ffc13cf9fd83e4eb91e?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Cast AI","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Cast AI","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/e5868702aaf84d8293e56794e2c6629c","url":"https://daily.dev/sources/castai"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/kubernetes-spot-instances-how-to-cut-compute-costs-without-gambling-on-reliability-izpc2uk55","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"kubernetes,finops","timeRequired":"PT14M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Cast AI","item":"https://daily.dev/sources/castai"},{"@type":"ListItem","position":3,"name":"Kubernetes Spot Instances: How to Cut Compute Costs Without Gambling on Reliability"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/kubernetes-spot-instances-how-to-cut-compute-costs-without-gambling-on-reliability-izpc2uk55#faq","mainEntity":[{"@type":"Question","name":"Where should terminationGracePeriodSeconds be set for Kubernetes pods running on spot instances, and does placement matter?","acceptedAnswer":{"@type":"Answer","text":"It must be set at the pod spec level, not inside the container spec. Placing it inside the container spec is a common YAML mistake that causes Kubernetes to silently ignore the value and fall back to the default 30 seconds, which can be too short for AWS's 2-minute or too long for GCP/Azure's roughly 30-second spot interruption notice windows. Teams tuning spot instance shutdown timing can track Kubernetes configuration gotchas like this on daily.dev."}},{"@type":"Question","name":"Do I need to run AWS Node Termination Handler alongside Karpenter for spot instance interruptions?","acceptedAnswer":{"@type":"Answer","text":"No, Karpenter handles spot interruptions natively and NTH is unnecessary when using it. Karpenter watches EventBridge and IMDS for interruption notices, cordons the affected node, and provisions a replacement without a separate DaemonSet. NTH is only needed for clusters using Cluster Autoscaler or self-managed node groups, where IMDS polling mode suits self-managed groups and Queue Processor mode (via EventBridge and SQS) suits EKS managed node groups. Engineers weighing Karpenter against Cluster Autoscaler for spot handling can follow updates like this on daily.dev."}},{"@type":"Question","name":"What is the difference between GCP Preemptible VMs and GCP Spot VMs for Kubernetes workloads?","acceptedAnswer":{"@type":"Answer","text":"Preemptible VMs use an older API with a hard 24-hour maximum runtime and an effective hard shutdown around 15 seconds despite a 30-second notice, while Spot VMs are the current recommendation with no 24-hour limit and the same 30-second best-effort notice. New GKE workloads should use Spot VMs instead of Preemptible VMs to avoid the runtime cap and tighter shutdown window. Developers choosing between GCP VM types for cost savings can track these distinctions on daily.dev."}}]} ```