Temporal-Aware GPU Resource Allocation for Distributed LLM Inference via Reinforcement Learning
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912589450051584 |
|---|---|
| author | Du, Chengze Yu, Zhiwei Xu, Heng Wang, Haojie liu, Bo Li, Jialong |
| author_facet | Du, Chengze Yu, Zhiwei Xu, Heng Wang, Haojie liu, Bo Li, Jialong |
| contents | The rapid growth of large language model (LLM) services imposes increasing demands on distributed GPU inference infrastructure. Most existing scheduling systems follow a reactive paradigm, relying solely on the current system state to make decisions, without considering how task demand and resource availability evolve over time. This lack of temporal awareness in reactive approaches leads to inefficient GPU utilization, high task migration overhead, and poor system responsiveness under dynamic workloads. In this work, we identify the fundamental limitations of these instantaneous-state-only scheduling approaches and propose Temporal Optimal Resource scheduling via Two-layer Architecture (TORTA). TORTA introduces a spatiotemporal scheduling framework that captures both long-term workload patterns and short-term execution constraints. It adopts a two-layer design: a macro-level scheduler leverages reinforcement learning and optimal transport to coordinate inter-region task distribution, while a micro-level allocator refines task-to-server assignments within each region to reduce latency and switching costs. Experimental results across multiple network topologies show that TORTA reduces average inference response time by up to 15\%, improves load balance by approximately 4-5\%, and cuts total operational cost by 10-20\% compared to state-of-the-art baseline methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_10259 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Temporal-Aware GPU Resource Allocation for Distributed LLM Inference via Reinforcement Learning Du, Chengze Yu, Zhiwei Xu, Heng Wang, Haojie liu, Bo Li, Jialong Distributed, Parallel, and Cluster Computing Networking and Internet Architecture The rapid growth of large language model (LLM) services imposes increasing demands on distributed GPU inference infrastructure. Most existing scheduling systems follow a reactive paradigm, relying solely on the current system state to make decisions, without considering how task demand and resource availability evolve over time. This lack of temporal awareness in reactive approaches leads to inefficient GPU utilization, high task migration overhead, and poor system responsiveness under dynamic workloads. In this work, we identify the fundamental limitations of these instantaneous-state-only scheduling approaches and propose Temporal Optimal Resource scheduling via Two-layer Architecture (TORTA). TORTA introduces a spatiotemporal scheduling framework that captures both long-term workload patterns and short-term execution constraints. It adopts a two-layer design: a macro-level scheduler leverages reinforcement learning and optimal transport to coordinate inter-region task distribution, while a micro-level allocator refines task-to-server assignments within each region to reduce latency and switching costs. Experimental results across multiple network topologies show that TORTA reduces average inference response time by up to 15\%, improves load balance by approximately 4-5\%, and cuts total operational cost by 10-20\% compared to state-of-the-art baseline methods. |
| title | Temporal-Aware GPU Resource Allocation for Distributed LLM Inference via Reinforcement Learning |
| topic | Distributed, Parallel, and Cluster Computing Networking and Internet Architecture |
| url | https://arxiv.org/abs/2507.10259 |