paper-with-me

Papers

Block: Balancing Load in LLM Serving with Context, Knowledge and Predictive Scheduling

2025-08-05 · Wei Da, Evangelia Kalyvianaki arxiv

This paper presents Block, a distributed scheduling framework designed to optimize load balancing and auto-provisioning across instances in large language model serving frameworks by leveraging contextual information from incoming requests. Unlike popular model serving systems that rely on monolithic and heuristic task schedulers, Block operates as a fully distributed, stateless, and predictive scheduling system to achieve low overhead, reliability, and scalability. It leverages the deterministic and predictable characteristics of LLM inferences, such as host configurations, response lengths, and hardware performance, to make scheduling decisions based on accurately predicted metrics. Evaluation on a 12 GPUs cluster shows that Block significantly outperforms heuristic schedulers, boosting serving capacity by up to 16.7\% and reducing P99 tail latency by up to 49.5\%. These performance gains remain consistent across diverse models, workloads and configurations. Code and data are open-sourced.

📄 PDF Abstract BibTeX arXiv:2508.03611

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving

2026-07-01 · Sangjin Choi, Sukmin Cho, Yifan Xiong, Ziyue Yang 외 hf

In prefill-decode (PD) disaggregated LLM serving, each request is assigned to a decode worker after prefill. Existing decode routers balance only load; for mixture-of-experts (MoE) models this is incomplete: equally load…

Load Balancing Mixture of Experts with Similarity Preserving Routers

2025-06-16 · Nabil Omi, Siddhartha Sen, Ali Farhadi

Sparse Mixture of Experts (MoE) models offer a scalable and efficient architecture for training large neural networks by activating only a subset of parameters ("experts") for each input. A learned router computes a dist…

Mixture-of-Experts

EWSJF: An Adaptive Scheduler with Hybrid Partitioning for Mixed-Workload LLM Inference

2026-01-29 · Bronislav Sidik, Chaya Levi, Joseph Kampeas arxiv

Serving Large Language Models (LLMs) under mixed workloads--short, latency-sensitive interactive queries alongside long, throughput-oriented batch requests--poses a fundamental scheduling challenge. Standard First-Come, …

HERALD: High-Throughput Block Diffusion LLM Serving via CPU-GPU Cooperative KV Cache Retrieval

2026-06-19 · Omin Kwon, Doyeon Kim, Jongseok Park, Seung Yul Lee 외 arxiv

Diffusion LLMs (dLLMs) improve GPU utilization over autoregressive decoding by generating multiple tokens per forward pass, but their KV cache still grows linearly with context, limiting throughput at long contexts. KV c…

Communication Load Balancing via Efficient Inverse Reinforcement Learning

2023-03-22 · Abhisek Konar, Di wu, Yi Tian Xu, Seowoo Jang 외

Communication load balancing aims to balance the load between different available resources, and thus improve the quality of service for network systems. After formulating the load balancing (LB) as a Markov decision pro…

reinforcement-learningReinforcement LearningReinforcement Learning (RL)