OpScale is presented, a practical operator-level orchestration framework of profiling, provisioning, placement, and runtime serving that attains SLOs with up to 36.3% fewer GPUs and 28% less power, or achieves 44% higher throughput under fixed cost budgets.
Abstract
Achieving cost efficiency while meeting strict user-facing SLOs (e.g., time-to-first-token) remains a fundamental challenge for cloud GPU clusters serving large language models (LLMs). Autoscaling is the key mechanism for cluster resource management, yet a basic system design question is open for serving LLMs: what should be the unit of scaling? Existing approaches primarily treat the entire model as a monolithic scaling unit--simple but unable to capture the fine-grained dynamics of inference workloads. As a result, such coarse-grained scaling often leads to either SLO violations under bursty demand or significant GPU under-utilization. Our characterization reveals substantial operator heterogeneity, exposing operator-level elasticity as a viable scaling primitive. We present OpScale, a practical operator-level orchestration framework of profiling, provisioning, placement, and runtime serving. OpScale is designed to tackle the high complexity and the space explosion problem, arising from operating at this finer granularity. Evaluated with production traces on up to 40 A100s and 24 GB200s, OpScale attains SLOs with up to 36.3% fewer GPUs and 28% less power, or achieves 44% higher throughput under fixed cost budgets.
Online large language model (LLM) serving has become the backbone of modern AI applications, powering diverse downstream services through shared hardware clusters. However, modern serving systems frequently encounter highly dynamic workloads characterized by severe workload skewness, where a small fraction of model instances receives the vast majority of traffic. Existing instance-level scaling mechanisms are limited by coarse-grained resource adjustment: scaling up requires the cold-start of full-model replicas, incurring substantial latency, while scaling down leaves the system vulnerable to performance degradation during sudden traffic surges. The key insight of this work is that LLM serving offers a unique opportunity for fine-grained scaling. In this paper, we propose CoCoScale, a layer-wise dynamic scaling mechanism that selectively expands the parallelism of hot layers onto idle resources reclaimed from underutilized devices, enabling elastic data parallelism without altering model architectures or adding hardware overhead. Evaluations demonstrate that CoCoScale significantly reduces cold start latency by 97.9%-99.3% compared to traditional scale up. Under production traces, CoCoScale reduces average latency by 20.7\%--28.1\% and achieves full Service Level Objective (SLO) attainment, demonstrating superior dynamic adaptability and resource efficiency.
Jingfeng Wu, Yiyuan He, Minxian Xu et al.· 0 citations
CELLServe formalizes SLO-constrained joint resource provisioning as an optimization problem with a dedicated algorithm, and introduces an opportunistic instance merging strategy for decode phase functions to reclaim fragmented resources.
Zejian Wang, Nan Lin, Zinuo Cai et al.· ACM Transactions on Architec...· 0 citations
LLM serving systems are provisioned for peak load to meet strict latency targets, leaving substantial GPU compute idle whenever traffic falls below peak. We present DeltaServe, a host-agnostic co-serving design that converts this idle inference capacity into LoRA fine-tuning throughput while preserving inference service-level objectives (SLOs). DeltaServe integrates with existing inference engines through a compact hook interface that requires only multi-LoRA batching support. It exploits the shared execution structure of inference prefill and LoRA fine-tuning forward passes, and uses an SLO-aware scheduler to admit and execute fine-tuning only when sufficient inference headroom is available. The scheduler is driven by a CUDA-graph-aware latency model calibrated offline and refined online. We integrate DeltaServe with vLLM, SGLang, and S-LoRA. On a production trace from Company X, DeltaServe on vLLM delivers 2.9x higher fine-tuning throughput than LLMStation at 100% inference SLO compliance, versus 85% for LLMStation. It also achieves 39% higher fine-tuning throughput than a baseline running vLLM+torchtune, using no additional hardware and maintaining full SLO compliance.
Jiaxuan Chen, Jianshu She, Ye Yuan et al.· 0 citations
Production large language model (LLM) serving uses continuous batching to maximize GPU utilization. While shared compute and memory resources improve throughput, they entangle the energy footprints of concurrent requests and make per-request energy physically unobservable. It is thus critical to develop a fair method for disaggregating device-level energy to individual requests. We tackle this challenge by developing a new approach to energy accounting in LLM serving. We introduce KV (Key-Value) volume, a physically grounded metric that captures the spatiotemporal footprint of a request’s KV cache occupancy, and show that energy per KV volume (EPV) provides a stable and reproducible signature for modeling serving energy. Building on EPV, we develop a state-aware energy model and a game-theoretic attribution method, and further build WattsOnLLM, a lightweight system for practical online energy accounting in existing LLM serving platforms with request-level granularity. We demonstrate that conventional heuristics deviate from the Shapley fairness target by 51.6% to 95%. In contrast, WattsOnLLM achieves a macroscopic energy estimation error as low as 1.9% and approximates the Shapley fairness target within 5.25% with very low overhead.
Xianyi Yuan, Hanlong Liao, Kunming Zhang et al.· Asia-Pacific Workshop on Net...· 0 citations
Large Language Model (LLM) serving has become a critical cloud workload, and realistic traces are essential for motivating and benchmarking serving systems. However, existing LLM serving workload studies remain limited in scale and scope. They often observe short time periods and provide limited visibility into how users interact with models in production. As a result, they do not fully capture how LLM serving workloads evolve over time or how user-model interactions shape production traffic. In this work, we further the understanding of real-world LLM serving workloads through both a global characterization and a longitudinal study of a one-year production trace from Chutes. Unlike prior studies, our trace captures full production behavior across many models and users, including both popular and long-tail models. We analyze the workload from aggregate, temporal, model-level, and user-level perspectives, revealing workload evolution and user-model structure that are typically hidden behind aggregate views. To support future research, we will release the full one-year trace with the paper, enabling downstream studies of production behavior without relying on sampled or synthetically generated workloads.
William Nixon, Jon Durbin, Florian Standhartinger et al.· 0 citations
Existing Large Language Model (LLM) inference systems often rely on static model placement and scheduling policies, which struggle to handle heterogeneous and dynamic real-world workloads. The key challenge is to adapt serving strategies to workload fluctuations while keeping reconfiguration overhead minimal. In this paper, we present OrionInfer, an adaptive LLM serving system that aligns inference strategies with real-time demand. OrionInfer introduces three key techniques: (1) runtime switching between data parallelism and tensor parallelism with negligible overhead; (2) an efficient inference pipeline that preserves batching efficiency during parallelism transitions; and (3) live-migration-based load balancing to alleviate memory pressure and improve resource utilization. Evaluations across multiple model scales show that OrionInfer delivers robust performance under diverse serving scenarios. In end-to-end serving, it reduces average TTFT by up to 25% over DP-priority configurations under low loads and lowers P99 tail latency by 50%--90% over TP-priority configurations under most high-traffic settings. In disaggregated prefill serving, OrionInfer improves prefill completion time (PCT) SLO attainment by up to 16.5 percentage points over DP-priority static baselines and reduces P99 PCT by up to 74.7% over TP-priority static baselines. Compared with dynamic baseline, OrionInfer provides better tail-latency stability, reducing P99 PCT by 38.6%--40.8% while avoiding the extra memory footprint.
Jingqi Feng, Guang Yang, Yukai Huang et al.· Proceedings of the 32nd ACM...· 0 citations