Abstract Storage is performance critical for I/O intensive applications deployed in data centers. As one of the recent implementations, media-based cache (MBC) is designed to alleviate the performance bottleneck of hard disk drives (HDD). This paper is motivated by the performance analysis of MBC in HDD and a new queueing model called M/G [ b ] /1/K with multiple state-dependent vacations is developed to formulate the MBC dynamics. Based on the physical mechanism of MBC, a detailed model is derived to capture the characteristics of distributions of service times and vacation times (disk action times). By solving the queueing model, we propose an approach to compute the steady-state distribution and the blocking probability. These analytical results are applied to study the system I/O throughput and the blocking probability of MBC under different workloads. We also analyze the changes of MBC performance with respect to the system parameters in different working scenarios. Typical recommendations of optimized MBC parameters are derived, which can be used to guide the design of MBC.

中文翻译：

---

在具有多个状态相关休假的 M/G[]/1/K 队列上：硬盘驱动器中基于媒体的缓存的实际问题

摘要 存储对于部署在数据中心的 I/O 密集型应用程序的性能至关重要。作为最近的实现之一，基于媒体的缓存 (MBC) 旨在缓解硬盘驱动器 (HDD) 的性能瓶颈。本文受 HDD 中 MBC 的性能分析的启发，并开发了一种称为 M/G [ b ] /1/K 的具有多个状态相关假期的新排队模型来制定 MBC 动态。基于MBC的物理机制，推导出详细的模型来捕捉服务时间和休假时间（磁盘动作时间）的分布特征。通过求解排队模型，我们提出了一种计算稳态分布和阻塞概率的方法。这些分析结果用于研究系统 I/O 吞吐量和 MBC 在不同工作负载下的阻塞概率。我们还分析了不同工作场景下MBC性能相对于系统参数的变化。推导出优化MBC参数的典型建议，可用于指导MBC的设计。