{"id":284,"date":"2026-08-08T08:13:27","date_gmt":"2026-08-08T08:13:27","guid":{"rendered":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/"},"modified":"2026-08-08T08:13:27","modified_gmt":"2026-08-08T08:13:27","slug":"kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache","status":"publish","type":"post","link":"https:\/\/wp.spain2.com\/es\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/","title":{"rendered":"DRA de Kubernetes en 2026: Planificaci\u00f3n de GPU Sin el Dolor de Cabeza del Device Plugin"},"content":{"rendered":"<p>GPUs used to be the domain of research labs and big tech. In 2026, an SMB with a single GPU instance runs LLM inference, fine-tuning jobs, and batch transcription \u2014 and pays per second for the privilege. That makes GPU efficiency a real budget line, not a footnote. And for years, the way Kubernetes handed out GPUs made efficiency almost impossible: the device plugin model gave every workload a whole GPU or nothing at all.<\/p>\n<p>That changed with <strong>Dynamic Resource Allocation (DRA)<\/strong>. It went stable in Kubernetes v1.35 and is enabled by default. If you run GPU workloads on Kubernetes \u2014 or you&#8217;re about to \u2014 DRA is the most important scheduler change in years. Here&#8217;s what it is, how to use it, and whether your SMB should adopt it now.<\/p>\n<h2>Why the Device Plugin Model Hurts SMBs<\/h2>\n<p>Since Kubernetes 1.8, special hardware has been exposed through device plugins. On an NVIDIA cluster that means running the NVIDIA device plugin as a DaemonSet, labeling GPU nodes, and requesting a card like this:<\/p>\n<pre><code>resources:\n  limits:\n    nvidia.com\/gpu: 1   # one whole GPU, or nothing<\/code><\/pre>\n<p>That works, but it&#8217;s crude in exactly the ways that cost SMBs money:<\/p>\n<ul>\n<li><strong>All-or-nothing allocation.<\/strong> A pod that needs 4 GB of VRAM to serve a small model consumes an entire 24 GB card. On a two-node GPU cluster, that&#8217;s the difference between running two workloads and six.<\/li>\n<li><strong>Static decisions.<\/strong> The kubelet hands out devices at container start. The scheduler never sees GPU health, memory, or topology, so you compensate with taints, labels, and guesswork.<\/li>\n<li><strong>Vendor lock-in at the API level.<\/strong> The device plugin API is effectively per-vendor. Moving from NVIDIA to AMD means rewriting how you request hardware.<\/li>\n<li><strong>No sharing, no metadata.<\/strong> Nothing tells Kubernetes &#8220;any GPU with at least 16 GB of memory&#8221; or &#8220;a GPU on the same PCIe switch as this NIC.&#8221;<\/li>\n<\/ul>\n<p>None of this is fatal at one GPU node. It becomes fatal at three \u2014 which is exactly where SMBs land once AI workloads stop being an experiment.<\/p>\n<h2>How DRA Works: DeviceClasses, ResourceClaims, and ResourceSlices<\/h2>\n<p>DRA replaces the device plugin model with a general, vendor-neutral device API built around four pieces:<\/p>\n<ul>\n<li><strong>DeviceClass<\/strong> \u2014 an admin-defined category of devices, with CEL selection rules (&#8220;NVIDIA GPU with at least 16 GiB memory&#8221;).<\/li>\n<li><strong>ResourceClaim \/ ResourceClaimTemplate<\/strong> \u2014 what a workload asks for. Claims can be per-pod or shared by several pods.<\/li>\n<li><strong>ResourceSlice<\/strong> \u2014 a live inventory of devices published by the driver (&#8220;node-a has 2 A10s, node-b has 4 L4s&#8221;).<\/li>\n<li><strong>A DRA driver<\/strong> \u2014 vendor software (NVIDIA&#8217;s ships in the GPU Operator) that publishes slices, allocates devices, and exposes them to containers via the Container Device Interface (CDI).<\/li>\n<\/ul>\n<p>The flow is simple: the scheduler matches each claim against available slices, picks a node, the driver prepares the device, and CDI bind-mounts the right files into your container. No taints, no labels, no kubelet restarts.<\/p>\n<p>Two 2026 details worth knowing: DRA went <strong>stable in v1.35<\/strong> (enabled by default), and v1.36 added <strong>prioritized request lists<\/strong> \u2014 you can say &#8220;prefer a big GPU, fall back to two small ones.&#8221; That&#8217;s the kind of flexibility that makes GPU nodes schedulable the way CPU nodes are.<\/p>\n<h2>A Working Example: Requesting a GPU with DRA<\/h2>\n<p><strong>Step 1 \u2014 install a DRA driver.<\/strong> On NVIDIA hardware, the GPU Operator enables DRA with a single flag:<\/p>\n<pre><code>helm repo add nvidia https:\/\/helm.ngc.nvidia.com\/nvidia\nhelm repo update\n\nhelm install gpu-operator nvidia\/gpu-operator \\\n  --namespace gpu-operator --create-namespace \\\n  --set driver.enabled=true \\\n  --set toolkit.enabled=true \\\n  --set dra.enabled=true<\/code><\/pre>\n<p><strong>Step 2 \u2014 define a DeviceClass.<\/strong> This one matches any NVIDIA GPU:<\/p>\n<pre><code>apiVersion: resource.k8s.io\/v1\nkind: DeviceClass\nmetadata:\n  name: nvidia-gpu\nspec:\n  selectors:\n    - cel:\n        expression: |\n          device.driver == \"nvidia.com\" &amp;&amp;\n          device.attributes[\"nvidia.com\"].type == \"gpu\"<\/code><\/pre>\n<p>Attribute names come from your driver. Run <code>kubectl get resourceslices -o yaml<\/code> to see exactly what fields your driver publishes, then write selectors against them \u2014 for example <code>device.attributes[\"nvidia.com\"].memory &gt;= 16Gi<\/code>.<\/p>\n<p><strong>Step 3 \u2014 request the device.<\/strong> A ResourceClaimTemplate plus a Deployment that references it:<\/p>\n<pre><code>apiVersion: resource.k8s.io\/v1\nkind: ResourceClaimTemplate\nmetadata:\n  name: gpu-claim\nspec:\n  spec:\n    devices:\n      requests:\n        - name: gpu\n          exactly:\n            deviceClassName: nvidia-gpu\n---\napiVersion: apps\/v1\nkind: Deployment\nmetadata:\n  name: llm-inference\nspec:\n  replicas: 1\n  selector:\n    matchLabels:\n      app: llm-inference\n  template:\n    metadata:\n      labels:\n        app: llm-inference\n    spec:\n      resourceClaims:\n        - name: gpu\n          resourceClaimTemplateName: gpu-claim\n      containers:\n        - name: inference\n          image: nvcr.io\/nvidia\/pytorch:24.12-py3\n          resources:\n            claims:\n              - name: gpu<\/code><\/pre>\n<p>Kubernetes creates a claim per pod automatically. Verify with:<\/p>\n<pre><code>kubectl get resourceclaims\nkubectl describe resourceclaim gpu-claim-&lt;pod-name&gt;<\/code><\/pre>\n<p>Want several pods to share one device (common for sharded inference)? Create a single claim with <code>allocationMode: All<\/code> and <code>adminAccess: true<\/code>, then reference it by name from each pod&#8217;s <code>spec.resourceClaims<\/code>.<\/p>\n<h2>What DRA Changes for SMBs \u2014 and When to Hold Off<\/h2>\n<p><strong>It cuts GPU spend.<\/strong> Request 8 GiB of VRAM instead of a whole card and two jobs pack onto one device. On hourly-priced GPU instances that is real money \u2014 the same math we covered in <a href=\"https:\/\/wp.spain2.com\/your-kubernetes-bill-is-out-of-control-how-smbs-can-cut-k8s-costs-by-50-without-sacrificing-reliability\">cutting Kubernetes costs<\/a>, applied at the device level.<\/p>\n<p><strong>It makes autoscaling smarter.<\/strong> Because claims are scheduler-aware, DRA pairs naturally with node autoscaling: <a href=\"https:\/\/wp.spain2.com\/karpenter-in-2026-smarter-kubernetes-node-autoscaling-for-smbs\">Karpenter<\/a> provisions the node, DRA picks the device. No more manually tainting GPU nodes to keep jobs off them.<\/p>\n<p><strong>It removes vendor-specific glue.<\/strong> The same YAML requests NVIDIA, AMD, or Intel accelerators \u2014 only the driver changes.<\/p>\n<p><strong>When to hold off:<\/strong> DRA needs a recent cluster (v1.35+), so an upgrade is a prerequisite \u2014 see our <a href=\"https:\/\/wp.spain2.com\/your-kubernetes-cluster-is-falling-behind-a-practical-upgrade-strategy-for-smbs\">practical cluster upgrade guide<\/a>. Driver maturity varies; NVIDIA&#8217;s is the most battle-tested. And if you run a single GPU node with one workload, the device plugin still works fine \u2014 keep it, and migrate when the second node arrives. During migration both models can coexist, so move inference workloads first and compare utilization before you commit.<\/p>\n<p><strong>A sensible adoption path:<\/strong> upgrade the cluster \u2192 install the driver with DRA enabled \u2192 define one DeviceClass \u2192 convert a single workload \u2192 watch <code>kubectl get resourceslices<\/code> and GPU utilization for a week \u2192 expand.<\/p>\n<p>DRA is one of those rare infrastructure changes that simplifies rather than complicates. It replaces taints, labels, and vendor plugins with one declarative API \u2014 and for SMBs watching GPU bills grow, it&#8217;s the difference between guessing and knowing exactly what your cluster can run.<\/p>\n<p>Not sure whether DRA is worth the migration for your cluster? We help SMBs make pragmatic Kubernetes and AI-infrastructure decisions \u2014 <a href=\"https:\/\/wp.spain2.com\/reserva-cita\">book a free 30-minute consultation<\/a> and we&#8217;ll map out your path.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Kubernetes DRA is GA in v1.35 and changes how SMBs schedule GPUs. Here is how ResourceClaims and DeviceClasses work, with real YAML.<\/p>","protected":false},"author":0,"featured_media":286,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[2],"tags":[122,121,120,44],"class_list":["post-284","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-devops-engineering","tag-ai-infrastructure","tag-dynamic-resource-allocation","tag-gpu","tag-kubernetes"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.9 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Kubernetes DRA in 2026: GPU Scheduling Without the Device Plugin Headache - SPAIN2.COM<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/wp.spain2.com\/es\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/\" \/>\n<meta property=\"og:locale\" content=\"es_ES\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Kubernetes DRA in 2026: GPU Scheduling Without the Device Plugin Headache - SPAIN2.COM\" \/>\n<meta property=\"og:description\" content=\"Kubernetes DRA is GA in v1.35 and changes how SMBs schedule GPUs. Here is how ResourceClaims and DeviceClasses work, with real YAML.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/wp.spain2.com\/es\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/\" \/>\n<meta property=\"og:site_name\" content=\"SPAIN2.COM\" \/>\n<meta property=\"article:published_time\" content=\"2026-08-08T08:13:27+00:00\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Tiempo de lectura\" \/>\n\t<meta name=\"twitter:data1\" content=\"5 minutos\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/\"},\"author\":{\"name\":\"\",\"@id\":\"\"},\"headline\":\"Kubernetes DRA in 2026: GPU Scheduling Without the Device Plugin Headache\",\"datePublished\":\"2026-08-08T08:13:27+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/\"},\"wordCount\":901,\"publisher\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/wp.spain2.com\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/postA.png\",\"keywords\":[\"ai-infrastructure\",\"dynamic-resource-allocation\",\"gpu\",\"Kubernetes\"],\"articleSection\":[\"DevOps Engineering\"],\"inLanguage\":\"es\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/\",\"url\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/\",\"name\":\"Kubernetes DRA in 2026: GPU Scheduling Without the Device Plugin Headache - SPAIN2.COM\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/wp.spain2.com\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/postA.png\",\"datePublished\":\"2026-08-08T08:13:27+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/#breadcrumb\"},\"inLanguage\":\"es\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"es\",\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/#primaryimage\",\"url\":\"https:\\\/\\\/wp.spain2.com\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/postA.png\",\"contentUrl\":\"https:\\\/\\\/wp.spain2.com\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/postA.png\",\"width\":1600,\"height\":900},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/wp.spain2.com\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Kubernetes DRA in 2026: GPU Scheduling Without the Device Plugin Headache\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/#website\",\"url\":\"https:\\\/\\\/wp.spain2.com\\\/\",\"name\":\"SPAIN2.COM\",\"description\":\"Cloud Consulting That Delivers \u2014 DevOps, SRE &amp; Cloud Infrastructure for SMBs\",\"publisher\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/wp.spain2.com\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"es\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/#organization\",\"name\":\"SPAIN2.COM\",\"url\":\"https:\\\/\\\/wp.spain2.com\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"es\",\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/wp.spain2.com\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/spain2-logo.svg\",\"contentUrl\":\"https:\\\/\\\/wp.spain2.com\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/spain2-logo.svg\",\"caption\":\"SPAIN2.COM\"},\"image\":{\"@id\":\"https:\\\/\\\/wp.spain2.com\\\/#\\\/schema\\\/logo\\\/image\\\/\"}}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"DRA de Kubernetes en 2026: Planificaci\u00f3n de GPU Sin el Dolor de Cabeza del Device Plugin - SPAIN2.COM","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/wp.spain2.com\/es\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/","og_locale":"es_ES","og_type":"article","og_title":"Kubernetes DRA in 2026: GPU Scheduling Without the Device Plugin Headache - SPAIN2.COM","og_description":"Kubernetes DRA is GA in v1.35 and changes how SMBs schedule GPUs. Here is how ResourceClaims and DeviceClasses work, with real YAML.","og_url":"https:\/\/wp.spain2.com\/es\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/","og_site_name":"SPAIN2.COM","article_published_time":"2026-08-08T08:13:27+00:00","twitter_card":"summary_large_image","twitter_misc":{"Tiempo de lectura":"5 minutos"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/#article","isPartOf":{"@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/"},"author":{"name":"","@id":""},"headline":"Kubernetes DRA in 2026: GPU Scheduling Without the Device Plugin Headache","datePublished":"2026-08-08T08:13:27+00:00","mainEntityOfPage":{"@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/"},"wordCount":901,"publisher":{"@id":"https:\/\/wp.spain2.com\/#organization"},"image":{"@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/#primaryimage"},"thumbnailUrl":"https:\/\/wp.spain2.com\/wp-content\/uploads\/2026\/08\/postA.png","keywords":["ai-infrastructure","dynamic-resource-allocation","gpu","Kubernetes"],"articleSection":["DevOps Engineering"],"inLanguage":"es"},{"@type":"WebPage","@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/","url":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/","name":"DRA de Kubernetes en 2026: Planificaci\u00f3n de GPU Sin el Dolor de Cabeza del Device Plugin - SPAIN2.COM","isPartOf":{"@id":"https:\/\/wp.spain2.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/#primaryimage"},"image":{"@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/#primaryimage"},"thumbnailUrl":"https:\/\/wp.spain2.com\/wp-content\/uploads\/2026\/08\/postA.png","datePublished":"2026-08-08T08:13:27+00:00","breadcrumb":{"@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/#breadcrumb"},"inLanguage":"es","potentialAction":[{"@type":"ReadAction","target":["https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/"]}]},{"@type":"ImageObject","inLanguage":"es","@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/#primaryimage","url":"https:\/\/wp.spain2.com\/wp-content\/uploads\/2026\/08\/postA.png","contentUrl":"https:\/\/wp.spain2.com\/wp-content\/uploads\/2026\/08\/postA.png","width":1600,"height":900},{"@type":"BreadcrumbList","@id":"https:\/\/wp.spain2.com\/kubernetes-dra-in-2026-gpu-scheduling-without-the-device-plugin-headache\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/wp.spain2.com\/"},{"@type":"ListItem","position":2,"name":"Kubernetes DRA in 2026: GPU Scheduling Without the Device Plugin Headache"}]},{"@type":"WebSite","@id":"https:\/\/wp.spain2.com\/#website","url":"https:\/\/wp.spain2.com\/","name":"SPAIN2.COM","description":"Cloud Consulting That Delivers \u2014 DevOps, SRE &amp; Cloud Infrastructure for SMBs","publisher":{"@id":"https:\/\/wp.spain2.com\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/wp.spain2.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"es"},{"@type":"Organization","@id":"https:\/\/wp.spain2.com\/#organization","name":"SPAIN2.COM","url":"https:\/\/wp.spain2.com\/","logo":{"@type":"ImageObject","inLanguage":"es","@id":"https:\/\/wp.spain2.com\/#\/schema\/logo\/image\/","url":"https:\/\/wp.spain2.com\/wp-content\/uploads\/2026\/07\/spain2-logo.svg","contentUrl":"https:\/\/wp.spain2.com\/wp-content\/uploads\/2026\/07\/spain2-logo.svg","caption":"SPAIN2.COM"},"image":{"@id":"https:\/\/wp.spain2.com\/#\/schema\/logo\/image\/"}}]}},"_links":{"self":[{"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/posts\/284","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/comments?post=284"}],"version-history":[{"count":0,"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/posts\/284\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/media\/286"}],"wp:attachment":[{"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/media?parent=284"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/categories?post=284"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/wp.spain2.com\/es\/wp-json\/wp\/v2\/tags?post=284"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}