[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

338.0. "Is there any way to change TRT_Req on DEFZA?" by TKTVFS::IDO (Naoki Ido, CSC/Tokyo, EWB/3FS) Tue Sep 03 1991 07:13

Hi,

I am now trying to connect our FDDIs to some other vendor's. I need
to change TRT_req in DEFZA on the testing because non-DEC vendor requests
it more than 8ms. I could change it on DECbridge500 or DECconcentrator500 with
DECelms but TRT_Neg is still 8ms since DEFZA is working with default 8ms.
I know TRT-Neg can be changed if we remove DS5000 from the ring, but Is there 
no way to change TRT_Req on current ULTRIX V4.2 ?  I also would like to know 
whether it's possible if ULTRIX PhaseV.       

Regards,
Naoki Ido
T.RTitleUserPersonal
Name
DateLines
338.1MIPSBX::thomasThe Code WarriorTue Sep 03 1991 11:552
No possible.  The driver has no support for changing any FDDI parameters.
(Phase V just uses what the driver makes available).
338.28ms is OK! Why change?ZPOVC::LESTERYUNGT&N Mkt/Asia RegionWed Sep 04 1991 10:0812
    FWIW, why would you want to change it.  It has been determined that 8ms
    is a very good value for most FDDI configurations.  See paper by R.Jain
    "Performance Analysis of FDDI Token Ring Networks:  Entity Effect of
    Parameters and Guidelines fro Setting TTRT". Proceedings of the ACM
    SIGCOMM'90 (sept. 1990)
    
    Very informative reading.
    
    Paper is available in ps form.
    
    Regards,
    Lester
338.3...STAR::SALKEWICZIt missed... therefore, I am Wed Sep 04 1991 15:0223
    
    	FWIW, you must have the capability to change these things to
    	be able to claim "interoperability" and "conformance" to certain
    	standards. (ANSI)
    
    	While setting up the DEMFA for the Interop show, we were hit with
    	requuests to change T_Req. Not that  they really needed it
    	changed to make the ring operate; they just wanted to see that it
    	could be changed in case it needs to be. 
    
    	The previous reply correctly states that 8ms is goodness for
    	"most" FDDI networks,.. however,,.. ANSI and our own FDDI
    architecture states that this is a parameter that may need to be
    adjusted,.. and every implementation should provide the means to change
    it. Speaking for Phase IV DECnet over FDDI, the initial release
    with V5.4-3 does not have this capability, but we are busily
    developing it (via extensions to NCP/NICE) for a release in the
    near future. Probably a dash release of 5.5.
    
    	Stay tuned
    
    								/Bill
    
338.4Thank you.TKTVFS::IDONaoki Ido, CSC/Tokyo, EWB/3FSThu Sep 05 1991 08:0311
  
 >   Speaking for Phase IV DECnet over FDDI, the initial release
 >   with V5.4-3 does not have this capability, but we are busily
 >   developing it (via extensions to NCP/NICE) for a release in the
 >   near future. Probably a dash release of 5.5.
    
Well, How about ULTRIX group ?

Thank you,
Naoki    

338.5UPSAR::THOMASThe Code WarriorThu Sep 05 1991 12:231
    Supposedly it will be added in ULTRIX V4.3 but I haven't seen anything.
338.6KONING::KONINGBrivu Latviju!Mon Sep 16 1991 16:1017
Re .0: I think you misunderstood something.

The rule in FDDI is that the smallest value of T_Req (Requested TTRT) is the
value that is used.  This is perfectly normal and doesn't bother anyone.
If another node requested a larger one, it simply yields to the lower value.

You're right in saying that the DS5000 causes the higher T_Req to be ignored
and 8 ms to be used instead.  But this is NOT a problem.  In particular,
there is NO REASON for the T_Req to be the same for all nodes.

8 ms is a good value, for reasons documented in detail in the report mentioned
in other replies.  It is occasionally useful to have a larger value, but
only in very large networks.  For any small to medium size network, increasing
T_req above 8 ms is useless (and, in extreme cases, will hurt, though usually
it makes no difference at all).

	paul