选题推荐
============

* 基于机器学习等方法优化 I/O 调度

  * `uFLIP-OC: Understanding Flash I/O Patterns on Open-Channel Solid-State Drives <https://dl.acm.org/doi/abs/10.1145/3124680.3124741>`_
  * `OC-Cache: An Open-channel SSD Based Cache for Multi-Tenant Systems <https://ieeexplore.ieee.org/document/8711079>`_
  * `Host-Level Workload-Aware Budget Compensation I/O Scheduling for Open-Channel SSDs <https://ieeexplore.ieee.org/abstract/document/8863515>`_
  * `Work-in-Progress: A Deep Learning Strategy for I/O Scheduling in Storage Systems <https://ieeexplore.ieee.org/document/9052144>`_

* 面向多租户的 I/O 资源隔离

  * `OC-Cache: An Open-channel SSD Based Cache for Multi-Tenant Systems <https://ieeexplore.ieee.org/document/8711079>`_
  * `OCVM: Optimizing the Isolation of Virtual Machines with Open-Channel SSDs <https://link.springer.com/chapter/10.1007/978-3-030-60245-1_29>`_

* 基于 Zoned Storage 实现文件系统或KVDB

  * `Zoned Storage <https://zonedstorage.io/>`_
  * `ZenFS: RocksDB Storage Backend for ZNS SSDs and SMR HDDs <https://github.com/westerndigitalcorporation/zenfs>`_
  * `Accelerating RocksDB with Zoned Namespaces <https://www.snia.org/sites/default/files/SDC/2019/presentations/NVMe/Holmberg_Hans_Accelerating_RocksDB_with_Zoned_Namespaces.pdf>`_

* 实现外盘算法（如基于 SPDK）

  * 外盘排序

    * `Efficient and Scalable External Sort Framework for NVMe SSD <https://dl.acm.org/doi/abs/10.1109/TC.2020.3041220>`_
    * `Accelerating External Sorting via On-the-fly Data Merge in Active SSD <http://csl.skku.edu/papers/hotstorage14.pdf>`_
    * `KioxiaSort: Sorting 1TB by 89K Joules <http://sortbenchmark.org/KioxiaSort2019.pdf>`_
  
  * 图算法

    * `Large-Scale Graph Processing on Emerging Storage Devices <https://www.usenix.org/conference/fast19/presentation/elyasi>`_
    * `Practicably Boosting the Processing Performance of BFS-like Algorithms on Semi-External Graph System via I/O-Efficient Graph Ordering <https://www.usenix.org/conference/fast22/presentation/yang>`_

  * 训练神经网络

    * `FlashNeuron: SSD-Enabled Large-Batch Training of Very Deep Neural Networks <https://www.usenix.org/conference/fast21/presentation/bae>`_