I/O Architecture

RESEARCH ON SOLID-STATE DISK (SSD)

While storage capacity and CPU processing power have experienced rapid growth in the past, improvement in data bandwidth and access times of hard disk drives (HDD) has not kept pace. As a result, we are seeing an ever widening speed gap between CPUs and disk IO. Disk arrays can improve overall IO throughput but random access latency is still very large because of mechanical operations involved. Large buffers and deep cache hierarchy can improve latency but the access time reduction has been very limited so far because of poor data locality at the disk storage level. There is a number of research projects currently going on in HPCL:

  • Using SSD as a cache in storage hierarchy
    1. Jin Ren and Qing Yang, “I-CASH: Intelligently Coupled Array of SSD and HDD” in The 17th IEEE International Symposium on High Performance Computer Architecture, 2011 (HPCA’11). 
    2. Jing Yang and Qing Yang, “A New Metadata Update Method for Fast Recovery of SSD Cache,” The 8th IEEE International Conference on Networking, Architecture, and Storage (NAS 2013), July, 2013, Xi’an, China.
    3. Jin Ren and Qing Yang,  “ A New Buffer Cache Design Exploiting both Temporal and Content Localities”, The 30th International Conference on Distributed Computing Systems, ICDCS 2010. Genoa, Italy, June 21-25, 2010.
  • New nonvolatile storage devices
    1. Jeffrey Carvell, Ruihua Cheng, and Qing Yang, “INDUCED MAGNETO-ELECTRIC COUPLING AT FERROELECTRIC/FERROMAGNETIC INTERFACE” Journal of Applied Physics, 113, 17C715 (2013).http://dx.doi.org/10.1063/1.4794873.
    2. “Multiferro-Heterostructure Thin Film Having Tunable Magnetic Coupling At Room Temperature”, US PTO patent pending, 12/906,684. (With Ruihua Cheng).