Pp. 143–154 of the Proceedings

Next: Abstract

Cheating the I/O Bottleneck:
Network Storage with Trapeze/Myrinet

Darrell C. Anderson, Jeffrey S. Chase, Syam Gadde, Andrew J. Gallatin, and Kenneth G. Yocum¹
Department of Computer Science
Duke University
{anderson, chase, gadde, gallatin, grant}@cs.duke.edu

• Abstract
• Introduction
• High-Speed Data Transfer with Trapeze
  • Trapeze Overview
  • Unified Buffering for In-Kernel Trapeze
  • Incoming Payload Table
• Page Transfers in GMS/Trapeze
  • Basic Mechanisms
  • Zero-Copy Reply Handling
  • Delegated RPC
  • Read-Ahead with Nonblocking RPC
    • Nonblocking RPC
    • Read-Ahead from Network Memory
    • Deferred Continuations
• Performance
  • RPC Microbenchmarks
  • GMS/Trapeze File Access Speed
    • File Access Interfaces
    • Sequential Benchmark Results
• Conclusion
• Acknowledgments
• Availability
• Bibliography
• About this document ...