Saved in:
Bibliographic Details
Main Authors: Yousefijamarani, Zahra, Wang, Xinglu, Wang, Qian, Heisler, Morgan Lindsay, Shabani, Taha, Gholipour, Niloofar, Yassini, Parham, Chang, Hong, Chen, Kan, Zhang, Qiantao, Bai, Xiaolong, Wang, Jiannan, Xiong, Ying, Zhang, Yong, Fan, Zhenan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2508.15919
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913058092220416
author Yousefijamarani, Zahra
Wang, Xinglu
Wang, Qian
Heisler, Morgan Lindsay
Shabani, Taha
Gholipour, Niloofar
Yassini, Parham
Chang, Hong
Chen, Kan
Zhang, Qiantao
Bai, Xiaolong
Wang, Jiannan
Xiong, Ying
Zhang, Yong
Fan, Zhenan
author_facet Yousefijamarani, Zahra
Wang, Xinglu
Wang, Qian
Heisler, Morgan Lindsay
Shabani, Taha
Gholipour, Niloofar
Yassini, Parham
Chang, Hong
Chen, Kan
Zhang, Qiantao
Bai, Xiaolong
Wang, Jiannan
Xiong, Ying
Zhang, Yong
Fan, Zhenan
contents Large language model (LLM) serving faces the dual challenge of meeting strict user-specific service-level objectives (SLOs) while minimizing computational cost under dynamic, multi-task workloads. Existing approaches either rely on static scheduling policies or focus on single-task settings, limiting their applicability in real-world deployments with heterogeneous requests, variable prompt lengths, and elastic scaling requirements. We present HFX, a production LLM serving system that jointly optimizes request scheduling and elastic scaling across model replicas to satisfy diverse SLOs. HFX introduces a \textbf{scheduler} that performs proactive budget estimation and prioritization to ensure SLO compliance for both new and in-flight requests. HFX also integrates a \textbf{scaler} that supports fast device-to-device (D2D) weight transfer, reducing cold-start latency. Additionally, the system supports both colocated and disaggregated prefill/decode deployments, enabling adaptation to diverse workload patterns and cloud environments. Through extensive experiments on multi-task workloads, we demonstrate consistently higher SLO attainment, lower end-to-end latency, and lower NPU usage cost by up to 4.44$\times$, 65.82\%, and 49.81\%, respectively, compared to state-of-the-art systems. Our results highlight the effectiveness of SLO-aware scheduling and scaling in practical LLM serving, providing a robust framework for cost-efficient and SLO-compliant deployments.
format Preprint
id arxiv_https___arxiv_org_abs_2508_15919
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle HFX: Joint Design of Algorithms and Systems for Multi-SLO Serving and Fast Scaling
Yousefijamarani, Zahra
Wang, Xinglu
Wang, Qian
Heisler, Morgan Lindsay
Shabani, Taha
Gholipour, Niloofar
Yassini, Parham
Chang, Hong
Chen, Kan
Zhang, Qiantao
Bai, Xiaolong
Wang, Jiannan
Xiong, Ying
Zhang, Yong
Fan, Zhenan
Distributed, Parallel, and Cluster Computing
Artificial Intelligence
Large language model (LLM) serving faces the dual challenge of meeting strict user-specific service-level objectives (SLOs) while minimizing computational cost under dynamic, multi-task workloads. Existing approaches either rely on static scheduling policies or focus on single-task settings, limiting their applicability in real-world deployments with heterogeneous requests, variable prompt lengths, and elastic scaling requirements. We present HFX, a production LLM serving system that jointly optimizes request scheduling and elastic scaling across model replicas to satisfy diverse SLOs. HFX introduces a \textbf{scheduler} that performs proactive budget estimation and prioritization to ensure SLO compliance for both new and in-flight requests. HFX also integrates a \textbf{scaler} that supports fast device-to-device (D2D) weight transfer, reducing cold-start latency. Additionally, the system supports both colocated and disaggregated prefill/decode deployments, enabling adaptation to diverse workload patterns and cloud environments. Through extensive experiments on multi-task workloads, we demonstrate consistently higher SLO attainment, lower end-to-end latency, and lower NPU usage cost by up to 4.44$\times$, 65.82\%, and 49.81\%, respectively, compared to state-of-the-art systems. Our results highlight the effectiveness of SLO-aware scheduling and scaling in practical LLM serving, providing a robust framework for cost-efficient and SLO-compliant deployments.
title HFX: Joint Design of Algorithms and Systems for Multi-SLO Serving and Fast Scaling
topic Distributed, Parallel, and Cluster Computing
Artificial Intelligence
url https://arxiv.org/abs/2508.15919