[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference 7.286::fddi

Title:FDDI - The Next Generation
Moderator:NETCAD::STEFANI
Created:Thu Apr 27 1989
Last Modified:Thu Jun 05 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2259
Total number of notes:8590

607.0. "Fail modes of FDDIcontroller 700" by LARVAE::WILLIAMS_G () Fri Jun 12 1992 12:29

    Hi,
    
    Can anyone tell me what, if any, are the failure modes of the
    DECcontroller 700 FDDI Turbochannel interface ?
    
    Are there set failure modes ? My customer is happy if there is a 
    problem with the board and it fails totally (no network impact). 
    However, if the interface fails and goes into some random transmit mode 
    then this is a potential disaster for the network.
    
    Thanks,
    
    Gary.
T.RTitleUserPersonal
Name		DateLines
607.1KONING::KONINGPaul Koning, A-13683Fri Jun 12 1992 13:5115
Same general answer as for Ethernet (see Ethernet conference).

In FDDI the situation is, if anything, better yet.  If your station were
to fail in such a way that it "babbles" continuously and allows nothing
at all from others to get through, that will trigger a recovery mechanism
("Trace" -- see the FDDI primer for an outline) that removes your station
from the network.  The kinds of failures that do this will also prevent
the station from reconnecting, so the station will stay off until fixed.

Failures that simply send lots of traffic (which is the worst that any
software failure can do) don't matter a whole lot; other stations will
still be able to send data.  Among other things, that will allow network
management to talk to the concentrator to turn off your port!

	paul
607.2ThanksLARVAE::WILLIAMS_GMon Jun 15 1992 11:4414
    Paul,
    
    Thanks for the info, much more positive than with an Ethernet card.
    
    I understand your point(s) about heavy loads from one controller not 
    stopping the network. In the case I am working with the network would 
    not sustain too much of this type of thing and still allow response 
    times to be met - but that's another story.
    
    Thanks again,
    
    Gary.
607.3KONING::KONINGPaul Koning, A-13683Mon Jun 15 1992 14:428
Re response time: keep in mind that there's a strict bound on how long it
takes to transmit.  Given N stations each offering an "infinite" load,
the worst case delay for your first message to get out is N * TTRT.
That's with async service, and ignoring errors such as loss of token.

For more detail on this, look at the paper on this topic by Raj Jain.

	paul