用大模型智能调优系统参数，让服务器性能提升超70%且不踩坑。
TuxBot: Semantic-Aware Online OS Tuning with Large Language Models
- 结合系统状态与历史记录，用大模型生成有语义的调优建议。
- 在13个真实负载上平均性能提升72.5%，比非大模型方法高153.3%。
- 适合运维人员和系统工程师，尤其在缺乏应用指标时仍表现优异。
在线操作系统调优可提升长时服务性能，但现有控制器对运行主机适配差。它们将调度、功耗、内存和I/O控制视为黑箱变量，仅优化标量奖励，忽略调参间的结构关系，在应用指标缺失时失效，还可能导致服务陷入持续退化的状态。我们提出TuxBot，一种基于大语言模型的主机侧稳态系统调优框架。TuxBot将调参模式、遥测数据、当前配置、近期动作-响应历史及检索到的过往运行记录整合为紧凑决策上下文。快速循环提出低延迟更新，慢速循环定期重设搜索策略，所有提议变更经类型化验证后才作用于内核或sysctl接口。该设计使控制器能理解系统控制语义与间接性能信号，同时约束模型成本、延迟与权限。我们在五个基准套件的13个真实负载上评估，调优最多41个Linux参数。整体上，相比默认设置性能提升72.5%，比最强非大模型基线高出153.3%。一次30窗口会话模型调用成本约0.20美元。仅依赖主机级指标时，仍比基于应用目标的基线高93.7个百分点，且避免了结构盲探引发的严重性能退化区域。
原文摘要 · Abstract (English)
Online OS tuning can improve long-running services, but existing controllers are poorly matched to live hosts. They treat scheduler, power, memory, and I/O controls as black-box variables and optimize a scalar reward. This view ignores cross-knob policy structure, breaks down when application metrics are unavailable, and can send a running service into degraded regions that persist after the bad setting is removed. We present TuxBot, a host-side framework for steady-state OS tuning with bounded language-model guidance. TuxBot turns knob schemas, telemetry, current configuration, recent action--response history, and retrieved prior runs into a compact decision context. A fast loop proposes low-latency updates, a slower loop periodically revises the search strategy, and every proposed change passes through typed validation before reaching kernel or sysctl interfaces. This lets the controller reason about OS-control meaning and indirect performance signals while keeping model cost, latency, and authority constrained. We evaluate TuxBot on 13 live workloads from five benchmark suites while tuning up to 41 Linux parameters. Across the suite, TuxBot improves stable-phase performance by 72.5% over default settings and by 153.3% relative to the strongest non-LLM baseline. A 30-window session costs about $0.20 in model calls. With only host-level metrics, TuxBot still outperforms baselines given direct application objectives by 93.7 percentage points, while avoiding severe degraded regions reached by structure-blind exploration.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。