[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

303.0. "FDDI RECONFIGURATION TIMINGS IN CASE OF A FAILURE" by BRSSWS::VERMEIREN () Fri Jul 19 1991 08:50

	Hi,

I am looking for some figures about the reconfiguration time on a FDDI ring 
in case of a failure, add or remove of a DAS station etc. 
I guess that this will be depending on the size of the network, number of 
stations and so, but I would like to known minimum, maximum and maybe an
average of the reconfiguration time.

Thanks in advance.
Etienne Vermeiren
Decsite Belgium
T.RTitleUserPersonal
Name		DateLines
303.1It can varyJUMP4::JOYGet a life!Fri Jul 19 1991 11:2210
    From a conversation I had with Paul Konig, this time can vary,
    depending on vendor implementations, ie roving MAC, etc. If "roving"
    MACs are used the reconfig time might vary from 2 to 16 microseconds.
    Using the standard insertion/removal method that we and many other
    vendors use, I believe its pretty consistent around 8 microseconds.
    
    Paul, please correct me if I got this wrong.
    
    Debbie
303.2KONING::KONINGEesti vabaks!Mon Jul 22 1991 11:578
Not quite... that should read milliseconds, not microseconds!

Also, for the roving MAC case we're really guessing, since there IS no
standard for what those do (see discussions earlier in this file).  One
plausible set of guesses gives reconfiguration times as long as 
165 milliseconds.  Compare that with what the standard gives you...

	paul
303.3what is a plausible value ?BRSSWS::VANDENBERGHEFri Aug 09 1991 10:3516
			Hi,

		Some questions about the figures:

	A reconfiguration will depends of different problems and parameters.

	Error detection, wrapping, scrub and claim processes will take some
	time.
	Does this plausible reconfiguration time of 165 milliseconds embrace
	all the processes or is it a part of it.

	Regards,
	Robert
303.4KONING::KONINGEesti vabaks!Fri Aug 09 1991 13:1912
Those numbers are for  the whole process except for error detection.  In other
words, I start counting at the moment that the node decides there is an error,
or other reason for reconfiguring.  The delay from the beginning of a problem
to the detection of that problem varies a lot.  It depends on the type of
error and its severity.  (For example, cut cable -- "no light" -- is detected
in much less than a millisecond.  Excessive error rates -- LEM reject -- are
detected very rapidly if the error rate is way over the limit, but it can
take multiple seconds if the error rate is only slightly over the limit.
Of course in the latter case the detection delay is not a major problem, since
the network doesn't suffer a whole lot in the meantime.)

	paul