{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,17]],"date-time":"2026-03-17T08:27:49Z","timestamp":1773736069783,"version":"3.50.1"},"reference-count":14,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,12,15]],"date-time":"2025-12-15T00:00:00Z","timestamp":1765756800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,12,15]],"date-time":"2025-12-15T00:00:00Z","timestamp":1765756800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/100000001","name":"National Science Foundation","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,12,15]]},"DOI":"10.1109\/ants66931.2025.11429804","type":"proceedings-article","created":{"date-parts":[[2026,3,16]],"date-time":"2026-03-16T20:11:03Z","timestamp":1773691863000},"page":"1-6","source":"Crossref","is-referenced-by-count":0,"title":["C-KASH: Custom Metric-based Kubernetes Autoscaler and Scheduler for Heterogeneous GPU-backed LLM Inference"],"prefix":"10.1109","author":[{"given":"Abhishek","family":"Muthukumar","sequence":"first","affiliation":[{"name":"Purdue University,School of Applied and Creative Computing,West Lafayette,IN,47907"}]},{"given":"Tanay","family":"Maheshwari","sequence":"additional","affiliation":[{"name":"Purdue University,School of Applied and Creative Computing,West Lafayette,IN,47907"}]},{"given":"Deepak","family":"Nadig","sequence":"additional","affiliation":[{"name":"Purdue University,School of Applied and Creative Computing,West Lafayette,IN,47907"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Horizontal Pod Autoscaler","year":"2017"},{"key":"ref2","article-title":"Monitoring GPUs in Kubernetes with DCGM","year":"2020"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.3390\/s24175551"},{"key":"ref4","article-title":"AAPA: An Archetype-Aware Predictive Autoscaler with Uncertainty Quantification for Server-less Workloads on Kubernetes","author":"Zhang","year":"2025"},{"key":"ref5","article-title":"Custom Pod Autoscaler","year":"2024"},{"key":"ref6","article-title":"GPU Sharing Scheduler for Kubernetes Cluster","year":"2024"},{"key":"ref7","article-title":"NVIDIA Open Sources Run:ai Scheduler to Foster Community Collaboration","year":"2025"},{"key":"ref8","article-title":"Exploring GPU Sharing in Kubernetes with NVIDIA KAI Scheduler and Exostellar SDG","author":"Shen","year":"2025"},{"key":"ref9","article-title":"Beware of Fragmentation: Scheduling GPU-Sharing Workloads with Fragmentation Gradient Descent","volume-title":"Proceedings of the 2023 USENIX Annual Technical Conference","author":"Weng"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/ccgrid64434.2025.00015"},{"key":"ref11","article-title":"LLMariner: Autoscaling for LLM Inference","year":"2024"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/ICC52391.2025.11161235"},{"key":"ref13","article-title":"Lessons Learned from the Chameleon Testbed","volume-title":"Proceedings of the 2020 USENIX Annual Technical Conference (USENIX ATC \u201920)","author":"Keahey"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.229"}],"event":{"name":"2025 IEEE International Conference on Advanced Networks and Telecommunications Systems (ANTS)","location":"Delhi, India","start":{"date-parts":[[2025,12,15]]},"end":{"date-parts":[[2025,12,18]]}},"container-title":["2025 IEEE International Conference on Advanced Networks and Telecommunications Systems (ANTS)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11429671\/11429680\/11429804.pdf?arnumber=11429804","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,17]],"date-time":"2026-03-17T06:18:02Z","timestamp":1773728282000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11429804\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,15]]},"references-count":14,"URL":"https:\/\/doi.org\/10.1109\/ants66931.2025.11429804","relation":{},"subject":[],"published":{"date-parts":[[2025,12,15]]}}}