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

Conference noted::atm

Title:	atm

Moderator:	NPSS::WATERS

Created:	Mon Oct 05 1992
Last Modified:	Thu Jun 05 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	970
Total number of notes:	3630

897.0. "ATM reducancy link and PNNI?" by SNOFS1::63496::CHIUANDREW () Wed Apr 09 1997 03:14

    Hi,
    
    Could someone give a light how to config ATM network which can give
    link reducancy (in FDDI we can have dual-home or dual ring from the
    VNswitch900EF to Gigaswitch/FDDI), do we have similar thing if customer
    use VNswitch900EA/ATMswitch900/Gigaswitch/ATM?
    
    Secondly, if they use ATMswitch900s as the backbone switch, does it
    support PNNI so that where the primary between ATM900 is down, it can
    automatically re-route to the backup link (i.e. there will be 3 ATM900
    switches, each connected with TWO links and customer will run LANE
    between VN900EA/ATM900). Can it be done? If yes, could someone explain
    how to config ATM900 so that automatically re-route can be done.
    
    
    
    thanks in advance for help!
    Andrew Chiu - NSIS Sydney

T.R	Title	User	Personal Name	Date	Lines
897.1		NPSS::NEWTON	Thomas Newton	`Wed Apr 09 1997 07:03`	37
	If I understand the question, the sort of link redundancy that you're looking for (a link fails, and traffic immediately goes to the backup link without applications being aware of the failure) isn't available in ATM the way that it is in FDDI. The model in ATM is that when a link fails, all the switched circuits through that link are broken, and it is up to the end nodes to redial any important calls. The switches are only responsible for trying to route new calls around failed paths. This carries over to LANE. When a LEC loses a connection to an ELAN, it gets the connection back not by switching to standby circuits, but by restarting the Configure/Join stage. Every vendor's ATM products are this way; this limitation on failover transparency isn't unique to us. Aside: At one time, our ATM switches had a proprietary feature called Resilient Virtual Circuits which DID provide nearly instant failover. Unfortunately, some technical problems proved insurmountable. Now ... o Connecting two links between each pair of ATMswitches sounds like it should work, as long as you understand the nature of ATM fault tolerance ("redial, not transparent"). I don't know enough about our switching/routing implementation to <guarantee> that it will work. You'll have to ask somebody who's got more knowledge of that area of the code about that. o I believe we currently use our own routing algorithm, and/or PNNI Level 0 (e.g., static routing). I believe "full" PNNI support is on the list for some release following Version 2.5.
897.2	Peer group level only	LEMAN::PAIVA	Hawkeye - Network Support @GEO	`Wed Apr 09 1997 08:06`	11
	Re: .1 >o I believe we currently use our own routing algorithm, and/or PNNI > Level 0 (e.g., static routing). I believe "full" PNNI support > is on the list for some release following Version 2.5. As far as I understood it, the first PNNI V1.0 implementation will only support one Peer group. Pedro