{"id":30034,"date":"2025-07-13T19:23:02","date_gmt":"2025-07-13T16:23:02","guid":{"rendered":"https:\/\/hgpu.org\/?p=30034"},"modified":"2025-07-13T19:23:02","modified_gmt":"2025-07-13T16:23:02","slug":"kis-s-a-gpu-aware-kubernetes-inference-simulator-with-rl-based-auto-scaling","status":"publish","type":"post","link":"https:\/\/hgpu.org\/?p=30034","title":{"rendered":"KIS-S: A GPU-Aware Kubernetes Inference Simulator with RL-Based Auto-Scaling"},"content":{"rendered":"<p>Autoscaling GPU inference workloads in Kubernetes remains challenging due to the reactive and threshold-based nature of default mechanisms such as the Horizontal Pod Autoscaler (HPA), which struggle under dynamic and bursty traffic patterns and lack integration with GPU-level metrics. We present KIS-S, a unified framework that combines KISim, a GPU-aware Kubernetes Inference Simulator, with KIScaler, a Proximal Policy Optimization (PPO)-based autoscaler. KIScaler learns latency-aware and resource-efficient scaling policies entirely in simulation, and is directly deployed without retraining. Experiments across four traffic patterns show that KIScaler improves average reward by 75.2%, reduces P95 latency up to 6.7x over CPU baselines, and generalizes without retraining. Our work bridges the gap between reactive autoscaling and intelligent orchestration for scalable GPU-accelerated environments.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Autoscaling GPU inference workloads in Kubernetes remains challenging due to the reactive and threshold-based nature of default mechanisms such as the Horizontal Pod Autoscaler (HPA), which struggle under dynamic and bursty traffic patterns and lack integration with GPU-level metrics. We present KIS-S, a unified framework that combines KISim, a GPU-aware Kubernetes Inference Simulator, with KIScaler, [&hellip;]<\/p>\n","protected":false},"author":351,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[11,3],"tags":[1782,20,2081,176,854],"class_list":["post-30034","post","type-post","status-publish","format-standard","hentry","category-computer-science","category-paper","tag-computer-science","tag-nvidia","tag-nvidia-geforce-rtx-3080","tag-package","tag-task-scheduling"],"views":1043,"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/30034","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/users\/351"}],"replies":[{"embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=30034"}],"version-history":[{"count":0,"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/30034\/revisions"}],"wp:attachment":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=30034"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=30034"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=30034"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}