[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

106.0. "Bridge forwarding Database" by FRAMBO::WILTRUD () Mon Aug 13 1990 09:31

    Hallo,
    
    again me with another question concerning the DECbride 500. How large
    can the forwarding Database of the DECbridge 500 be? Specially on the
    FDDI-side it might be necessary to have a lot of entries. (Imagine 100
    Ethernet LANs with 1024 stations) Does the Bridge learn every Stations
    on all Ethernets which are connected to the FDDI-Backbone via a Bridge?
    Is the bridge still fast enough if the forwarding Database is really big?
    If the bridge receives a FDDI-packet and it doesn't know the
    destination adress will the Bridge forwarding the packet to the
    Ethernet? Wouldn't be this mechanism dangerous because the Ethernet
    could be fill up with such packets?
    
    Again Thanks for any help
    
    Wiltrud

T.R	Title	User	Personal Name	Date	Lines
106.1		CVG::PETTENGILL	mulp	`Tue Aug 14 1990 23:19`	25
	I don't know if anything is said about this in the 802.1 spec covering bridges, but DEC has currently settled on 16,000 nodes as the maximum for an extended LAN. This is double the number in our initial bridge product, the LANbridge 100. In one sense, this number is not really the maximum as this is only the maximum number of addresses that any one bridge can contain. Due to locality of traffic and idle stations and other factors, one might expect that the number of addresses that any one bridge would see in a period on the order of an hour would be less than 16,000 even if the total number of nodes was several times greater. However, let's assume that 16,000 nodes is the maximum? Other than speculation, do you have an argument that suggests that this number is too small for an extended LAN. Don't be fooled by numbers like 200km rings and 100km over single mode fiber; these don't imply connecting all the nodes in a 1000 square miles together into a single entended LAN is feasible. The geography limits are much smaller for the next decade, at least. I'm not suggesting that you're wrong; when I presented a model for a large extended LAN I felt that a good upper bound was 100,000, but that was based on a next generation LAN used to connect everything together including disk servers and user terminals for a major DEC metro area, eg, greater Maynard, or greater Nashua. Howewer, I think 16,000 is a number that we can live with for some time and it can also be increased before there is an actual need to get there.
106.2	Forewarding packets because Destination unknown?	FRAMBO::WILTRUD		`Wed Aug 15 1990 03:57`	8
	Thanks, I think your right, 16,000 nodes are a lot of stations. What's with the second question. Is it possible that the load on a Ethernet-segment get to high because the FDDI-Bridge forwards packets while the Bridge doesn't know the destination? Wiltrud
106.3		KONING::KONING	NI1D @FN42eq	`Wed Aug 15 1990 19:41`	7
	Maybe. But as .1 indicates, that can only occur (except as a transient) if a particular bridge has more than 16k addresses to contend with. So it's impossible if the extended LAN has fewer than that, and unlikely unless it has a LOT more than that. I very much doubt you will see more than one or two such networks in the next few years. paul