Re: "Jon William Toigo": Storage over Ethernet/IP

[Ray Hoare <hoare+@pitt.edu>]

At 08:50 AM 5/26/00 -0400, Dave Nagle wrote:

>FYI
>------- Forwarded Message
>
>I am seeking a few points of clarification:
>
>1.  Fibre Channel folks have attempted to explain to me why TCP/IP could =
>NEVER be a viable interconnect for block level storage operations.  They =
>claim:
>
>a.  TCP is too CPU intensive and creates too much latency for storage =
>I/O operations.


Approximately 40% of sending TCP/IP messages is software overhead.
There is a 400ns access time for just touching a PCI card within a *PC*

>b.  The IP stack is too top heavy and processing packet headers is too =
>slow to support storage I/O operations.

This may be true for a PC/Workstation.  Direct hardware support is needed.

>c.  The maximum throughput of a GE TCP/IP connection is 768 Mps, which =
>is too slow to support storage I/O operations.

The maximum throughput of a PCI bus is 33MHz*32bits/cycle = 1056 Mbps (132 
MBytes/s)
*PEAK*
The disk in my computer works just fine.

Ray



---------------------------------------------------------
Raymond R. Hoare, Assistant Professor
348 Benedum Hall
Department of Electrical Engineering
University of Pittsburgh
Pittsburgh, PA 15261, USA
Email: hoare@pitt.edu
Voice: (412) 624-5836
Fax: (412) 624-8003
http://www.pitt.edu/~hoare