Abstract Tail latency is a widely-used performance metric in storage stack as an important indicator for the quality of service. Especially with the devices becoming increasingly faster, high tail latency does not only become more pronounced in presence, but also cause greater influences.Existing tools have shown only limited effectiveness, often failing to address the problem across a broad range of storage systems. In this paper, we propose a prediction-based IO rerouting framework, named Slow Re-router to alleviate latency spikes with inter-device cooperation. This proactive approach effectively mitigates latency under heavy workloads, achieving up to a 12x improvement over state-of-the-art optimization techniques. Furthermore, the Slow Re-router framework is adaptable to a variety of scenarios, including both redundant and non-redundant storage configurations, and is compatible with both SSD and HDD.
No takes yet. Share an insight, caveat, or question.
Linxiao Bai (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: