TurboCache: Empowering Switch-Accelerated Key-Value Caches with Accurate and Fast Cache Updates
Xiang Chen, Longlong Zhu, Linying Zheng, Lingfei Cheng, Jianshan Zhang, Xu Yang, Dong Zhang, Xuan Liu, Xiaoming Lu, Xun Yi, Ibrahim Khalil, Albert Y. Zomaya
摘要
Recent key-value (KV) caches are offloaded to programmable switches to offer high query processing performance. However, they suffer from both low accuracy in hot key detection and high latency in cache updates due to the strict limitations on switch registers. We propose TurboCache, a switch-accelerated KV cache with accurate hot key detection and fast cache updates. Our key idea is to leverage the switch recirculation capability to build a novel data structure that caches hot KV pairs. With this hardware-compatible cache data structure, TurboCache designs efficient data plane algorithms that accurately detects new hot keys and quickly updates its cache entirely within switch ASIC pipelines. We have implemented TurboCache on aGbps Tofino switch. Testbed results indicate that TurboCache improves the hot key detection accuracy and decreases the cache update latency of existing KV caches by several orders of magnitude.
问问这篇 Paper
问问你的智能体。
Lune 读过与它相关的顶会 Paper,每个回答都会注明依据哪几篇。
相关 Paper
- Pushing the Limits of In-Network Caching for Key-Value StoresGyuyeong KimNSDI 2025 · 被引用 9 次
- FarReach: Write-back Caching in Programmable SwitchesSiyuan Sheng, Huancheng Puyang, Qun Huang, Lu Tang 等USENIX ATC 2023 · 被引用 22 次
- Turbo: Efficiently Serving Long-Context Large Language Models with In-Network AggregationYing Wan, Yuchen Xu, Chuwen Zhang, Yingsheng Huang 等SIGCOMM 2026
- P4LRU: Towards An LRU Cache Entirely in Programmable Data PlaneYikai Zhao, Wenrui Liu, Fenghao Dong, Tong Yang 等SIGCOMM 2023 · 被引用 18 次
- SketchPipe: Toward Accurate Sketch-based Network Measurement on Multi-Pipeline Switches with Splitless Sketch PlacementXiang Chen, Longlong Zhu, Linying Zheng, Hongyang Du 等NSDI 2026
