people

Bo Peng 彭博

Dr. Bo Peng is an assitant professor in SJTU, a member of the Shanghai Key Laboratory of Scalable Computing and Systems. Bo Peng received his B.E. and M.E. and Ph.D. degrees from Shanghai Jiao Tong University in 2016, 2019 and 2023, and was awarded the Outstanding Graduates of Shanghai three times. His research interests mainly include virtualization, cloud computing, and heterogeneous memory and storage systems. He has published papers in international conferences/journals, including TC, TACO, USENIX ATC, IPDPS, etc. He won the honor of the National Scholarship for graduate students twice in 2018 and 2021. He was the technological backbone of two National R&D Projects and one National Science Fund Project.

Publications

Conference

  • Bo Peng, Cheng Guo, Jianguo Yao, Haibing Guan
    LPNS: Scalable and Latency-Predictable Local Storage Virtualization for Unpredictable NVMe SSDs in Clouds, in Proceedings of 2023 USENIX Annual Technical Conference (USENIX ATC 23), 785-800 (CCF A Conference)
  • Bo Peng, Jianguo Yao, Yaozu Dong, Haozhong Zhang, Yu Xu, Haibing Guan, MDev-NVMe: A NVMe Storage
    Virtualization Solution with Mediated Pass-Through”, USENIX ATC 2018 (CCF A Conference)
  • Bo Peng, Jianguo Yao, Zhengwei Qi, and Haibing Guan, HybridPass: Hybrid Scheduling for Mixed Flows in
    Datacenter Networks”, IPDPS 2018 (CCF B Conference)

Journal

  • Bo Peng, Yaozu Dong, Jianguo Yao, Fengguang, Wu, and Haibing Guan, FlexHM: A Practical System for Heterogeneous Memory with Flexible and Efficient Performance Optimizations, ACM Transactions on Architecture
    and Code Optimization，20(1): 13:1-13:26 (2023) (CCF A Journal)
  • Bo Peng, Jianguo Yao, Yaozu Dong, and Haibing Guan, MDev-NVMe: Mediated Pass-Through NVMe
    Virtualization Solution with Adaptive Polling, IEEE Trans. Computers 71(2): 251-265 (2022) (CCF A Journal)
  • Bo Peng, Ming Yang. Jianguo Yao, Haibing Guan: A Throughput-Oriented NVMe Storage Virtualization with
    Workload-Aware Management, IEEE Trans. Computers 70(12): 2112-2124 (2021) (CCF A Journal)
  • Bo Peng, Jianguo Yao, Xin Xu, Proactive Coordination for Low-Congestion Multi-Path Datacenter Networks”,
    Journal of Systems Architecture 101, (2019) (CCF B Journal)

Patents

  • 姚建国，彭博，管海兵；“一种云数据中心网络中混合数据流协同调度方法” ；ZL201710414936.2; 已授权；授权日期：2021.05.11。
  • 姚建国，郭成，彭博，陈义全，王一静; “时延可预测的 NVMe 存储虚拟化方法和系统”; ZL 202011620231.4; 已授权；授权日期：2023.01.31。
  • 姚建国，杨铭，彭博， 陈振东; “一种输入/输出 I/O 请求处理方法和主机设备”; CN202010397767.8; 已公开；公开时间：2021.11.16。
  • 姚建国, 吴志成, 彭博, 管海兵, “可扩展性的 NVMe 存储虚拟化方法和系统”; CN202111507944.4; 已公开；公开时间：2022.03.04。
  • 姚建国; 张子扬; 管海兵; 彭博; “基于指纹的块粒度数据去重系统和方法”; CN202210009039.4; 已公开；公开时间：2022.04.29。
  • 姚建国, 谢正平, YIREN RONNIE HUANG, 唐杰, 彭博；“拥有即时处理能力的NVMe虚拟化方法、系统、介质及设备”，CN202211684110.5，已申请，申请日期：2022.12.27。
  • 姚建国, 王文杰, YIREN RONNIE HUANG, 唐杰, 彭博；“完全硬件卸载的NVMe存储加速系统”； CN202211684098.8；已申请；申请日期：2022.12.27。

Researches

High-performance Storage Virtualization

  • High-performance Mediated Pass-Through Virtualization with adaptive polling.
    The first kernel state NVMe virtualization method greatly reduces the cost of active polling and improves the scalability of storage virtualization scheme.

  • Research on hybrid deployment of high-density containers and security containers for storage virtualization
    The first high-density virtualization scheme for a single NVMe disk device that breakthroughs the 256 limit of BDF, supporting the mixed deployment of 1000+quantity containers and security containers for high-performance serverless storage applications.

Cloud Storage QoS

  • High throughput QoS storage virtualization scheduling research.
    NVMe storage native virtualization with zero throughput under multi tenant shared cloud storage scenario.

  • Latency-predictable NVMe virtualization research.
    The first multi-tenant cloud storage virtualization method with predictable latency for normal and unpredictable NVMe disks.

Heterogeneous Memory System and Virtualization:

  • High-performance Transparent Heterogeneous Main Memory System.
    An Optane AEP supported practical heterogeneous memory management system and memory
    virtualization with customized high-performance memory management and performance control.

Others:

  • Mixed Flows Scheduling in Data Center Network.
    Main achievements: The first mixed flow collaborative scheduling scheme in the data center network effectively reduces the network delay cost caused by mixed flow network congestion and improves the utilization of network throughput rate.
×