Questions about FCIP

["Kavi, Prabhu" <prabhu_kavi@tenornetworks.com>]

I have some questions about the Fibre Channel over TCP/IP I-D.
I understand IP networks well, but am a Fibre Channel newbie and
am asking these questions to learn more about Fibre Channel's 
requirements rather than to challenge the document.  Here are
my questions:

1.  Why Fibre Channel over TCP, rather than over UDP?  Is it 
    because of an expectation of lossy IP networks?  Or is it
    equally valid to put Fibre Channel over UDP (and therefore
    can be the subject of another I-D).
2.  What are the typical values of the ED_TOV and R_A_TOV timers? 
    Do I understand this right to say that these timers effectively
    limit the latency across the IP network?
3.  What is an acceptable loss ratio of Fibre Channel datagrams
    that still allows the SAN devices to deliver appropriate
    end user performance (order of magnitude estimate is 
    sufficient).

The reasons I ask are:

* I believe it is invalid to assume that IP networks are 
  necessarily more lossy than Fibre Channel networks.  A 
  well-engineered IP network, using a combination of both 
  Diff-Serv and MPLS, can achieve (close to) zero loss.
* Stacking one reliable delivery protocol on top of another can
  possibly lead to interesting performance problems after a loss
  occurs, depending upon the timer values for TCP and Fibre Channel.

Thanks in advance for your answers.

Prabhu Kavi
----------------------------------------------------------------------
Prabhu Kavi                     Phone:  1-978-264-4900 x125 
Director, Adv. Prod. Planning   Fax:    1-978-264-0671
Tenor Networks                  Email:  prabhu_kavi@tenornetworks.com
100 Nagog Park                  WWW:    www.tenornetworks.com
Acton, MA 01720