[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

153.0. "Availability with DECconcentrators" by TOOIS1::MIRGHANE () Mon Oct 15 1990 08:41

    We are now answering an RFP where some 15 subsystems could be linked
    together using FDDI.
    Each subsystem is made up of one to 30 systems depending on the
    subsystem.
    In our vision a subsystem is based on computers (such as DS5000/200)
    with at least a double FDDI link per computer. 
    These systems will be linked together using FDDI DECconcentrators with an 
    architecture of trees and subtrees where necessary.
    
    Our problem is availability. The customer can't afford loosing a
    subsystem (or part of a subsystem which is 8 nodes) just because a 
    concentrator has a problem. This seems to be an unacceptable single point 
    of failure.
    
    Knowing the computers such as DS5000/200 have the possibility to accept
    3 FDDI links; can we connect each DS5000 to 2 FDDI concentrators?
    How will the network software (TCP/IP, NFS, DCE, EMA) manage 2
    connexions to 2 concentrators? will there be automatic switch in case
    of a concentrator problem or permanent connection to 2 FFDI links?
    
    
    Thank you for helping us solve this problem
    
    
    Soumetty.

T.R	Title	User	Personal Name	Date	Lines
153.1	dual homing	BAGELS::LEVY		`Mon Oct 15 1990 13:53`	17
	re: <<< Note 153.0 by TOOIS1::MIRGHANE >>> >Knowing the computers such as DS5000/200 have the possibility to accept >3 FDDI links; can we connect each DS5000 to 2 FDDI concentrators? Yes. This topology is called "dual homing." >How will the network software (TCP/IP, NFS, DCE, EMA) manage 2 >connexions to 2 concentrators? Only one of the connections is active at any point in time, so the upper layer protocols don't need to know about it. >will there be automatic switch in case >of a concentrator problem or permanent connection to 2 FFDI links? Automatic failover is provided by CMT built into the concentrators.
153.2	not a "dual homing"	EVTAI1::GROSSETETE		`Tue Oct 16 1990 08:10`	15
	re. -1, sorry, but I don't think this is a "dual homing" configuration (see previous note 147.2 & .3). If you'll have 2 FDDI adapters (SAS implementation) in your DECstation 5000-200, you'll be dependant of the protocols you'll use. ie : DECnet Phase IV on Ultrix can only use one of the controller at a time since it's an end node implementation DECnet Phase V will be different since you could benefit of the "Multi-circuit" function. TCP/IP must be OK. Could an Ultrix expert confirm ??? Best regards Patrick
153.3		ALLZS::MORRISON	The Network IS the System	`Tue Oct 16 1990 12:32`	11
	>How will the network software (TCP/IP, NFS, DCE, EMA) manage 2 >connexions to 2 concentrators? I can't speak for the others, but EMA (or rather the DECmcc implementation of the EMA architecture) doesn't yet handle FDDI. When it does, it will most likely use the same method that we use for Ethernet, where we determine what ports are on the system, and use whichever one works for the particular operation you are doing. In other words, it will handle N ports just fine. Wayne
153.4	not dual homing with DEFZA	BAGELS::LEVY		`Wed Oct 17 1990 18:44`	2
	Of course, you're right Patrick. What I described applies to a DAS adapter for the DS5000, not the DEFZA.
153.5	you mean we're going to have a DAS adapter?	ZPOVC::HWCHOY	It must be Thursday.	`Thu Oct 18 1990 03:57`	1

153.6	No.	BAGELS::LEVY		`Thu Oct 18 1990 19:48`	3
	Nope, I didn't say that. However, the Turbochannel interface is open to third-party developers, so there's the potential that one could be developed outside. (I'm not aware of any efforts in this direction.)