SCORPIO: Serving the Right Requests at the Right Time for Heterogeneous SLOs in LLM Inference

TL;DR

SCORPIO is an SLO-aware LLM serving system that adaptively manages requests to improve throughput and SLO compliance by exploiting heterogeneity in service objectives.

Contribution

It introduces a novel SLO-oriented scheduling framework with adaptive admission control and batching, specifically designed for heterogeneous SLOs in LLM inference.

Findings

System goodput increased by up to 14.4X.

SLO adherence improved by up to 46.5%.

Effective handling of diverse SLOs in LLM serving environments.

Abstract

Existing Large Language Model (LLM) serving systems prioritize maximum throughput. They often neglect Service Level Objectives (SLOs) such as Time to First Token (TTFT) and Time Per Output Token (TPOT), which leads to suboptimal SLO attainment. This paper introduces SCORPIO, an SLO-oriented LLM serving system designed to maximize system goodput and SLO attainment for workloads with heterogeneous SLOs. Our core insight is to exploit SLO heterogeneity for adaptive scheduling across admission control, queue management, and batch selection. SCORPIO features a TTFT Guard, which employs least-deadline-first reordering and rejects unattainable requests, and a TPOT Guard, which utilizes a VBS-based admission control and a novel credit-based batching mechanism. Both guards are supported by a predictive module. Evaluations demonstrate that SCORPIO improves system goodput by up to 14.4X and SLO adherence by up to 46.5% compared to state-of-the-art baselines.