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

Conference azur::mcc

Title:DECmcc user notes file. Does not replace IPMT.
Notice:Use IPMT for problems. Newsletter location in note 6187
Moderator:TAEC::BEROUD
Created:Mon Aug 21 1989
Last Modified:Wed Jun 04 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:6497
Total number of notes:27359

5480.0. "Architectural question on IP" by MLNCSC::BARILARO () Fri Aug 13 1993 10:56

    Hi,
    
    	A customer of ours would like to know if it's possible to see
    	a multiprotocol router (a Cisco) with N interface (so N IP
    	address) like a single entity.
    	It's  an architectural question, with a lot of pratical aspects
    	involved, for example means that a ping to this router fails only
    	when any of if interfaces are unreachable, that if you've backup
    	link you could continue to export data, etc.etc.
    
    	For what I know of MCC, that's not possible, 1 address = 1 entity.
    	Perhaps MCC/MSU is able to do that, I know it just a little.
    
    	It seems also that if, like this customer, you have defined 1 
    	router with more than 1 couple address/name like this:
    
    UCX> sho host faeta.unipi.it
    
         BIND database
    
    Server:   131.114.2.100    FIGARO.CNUCE.CNR.IT
    
    Host address    Host name
    
    131.114.21.99   FAETA.UNIPI.IT
    131.114.1.4     FAETA.UNIPI.IT
    131.114.248.2   FAETA.UNIPI.IT
    131.114.253.1   FAETA.UNIPI.IT
    
    	both the autotopology, than MCC itself has problem to manage it,
    	entity partially registered, MCC's query returned with error like:
    
    SNMP NETMON_NS:.ip.GARR-gw-mi-infn-it
    AT 13-AUG-1993 14:58:49 Characteristics
    
    No internet name or internet address registered
    
    on UCX it's appear regurarly:
    
    UCX> sho host GARR-gw.cnr.it
    
         BIND database
    
    Server:   131.114.2.130    ICNUCEVX.CNUCE.CNR.IT
    
    Host address    Host name
    
    131.114.1.3     GARR-GW.CNR.IT
    192.12.192.2    GARR-GW.CNR.IT
    192.12.193.14   GARR-GW.CNR.IT
    131.196.64.2    GARR-GW.CNR.IT
    
    And for the little I know MCC check first is dictionary and then the
    naming server (UCX in this case). 
    
    That's all, in base of my MCC's knowledge, i was able to understand.
    
    				Hoping to be wrong.....
    				Ciao Luciano
    
    
T.RTitleUserPersonal
Name
DateLines
5480.1Just a possible solution...ZTOIS1::VISTARenato VISTA, MCS/SPS France/StrasbourgMon Aug 16 1993 12:1141
    
    Luciano,
    
    It is right that using POLYCENTER Network Manager 200 and willing
    managing an multiple IP address as a single global entity is not very
    easy.
    
    Nevertheless, and because you've titled your item in architectural
    term, let me propose you an architectural solution.
    
    To manage CISCO Router (with more than one physical interface/line),
    you have to consider twao main things :
    
    	1) the localisation of your Network Management Platform (NMS)
    (physically, ie "before" ethernet controller, logically in a Internet
    Network, ie the same as your NMS platform)
    
    	2) the monitoring mode, ie :
    
    		. to monitor all equipments constituing the network from
    		your NMS interface to the CISCO/ethernet interface, (ie
    		Controller, DropCable, Transceiver, passive Repeater,
    		DELNIs, ...) you have to use a POLLING mode, based on 
    		IP/ICMP poller (ie ping...) ; so you stay in the same 
    		INTERNET network
    
    		. for other CISCO interfaces (serial,...), you may
    		use the natural asynchronous mode based on generic linkUp &
    		linkDown SNMP/TRAPS
    
    	Whatever you do, the polling mode (ping) is very usefull to be sure
    	that asynchronous events can sent over the physical/logical
    	ethernet/internet network to your NMS platform !!
    
    This kind of monitoring has been set on my own customer sites.
    
    I hope this will help you.
    Renato
    
    
    
5480.2Thanks, but I don't think customer love it..MLNCSC::BARILAROFri Aug 20 1993 05:2245
    Hi Renato,
    
    	Sorry for the delay in my answer.
    
    	Yes, I use the same solution for other customers, using traps or
    	checking the Ifoperstatus of the other interfaces to to see the 
    	status of the other links.
    
    	BUT that's applicable only if you have a single path to reach the
    	entity, or exist an interface that you could consider the most
    	important (for example the ethernet inteface of the remote CISCO
    	that permit you to reach the remote subnet).
    
    	In this customer environment (It's the italian Universities link 
    	to Internet, a network called GARR) there isn't this situation,
    	the network is something like this:
    
    	Ethernet 1
    	 ___
    Eth2  |
    |--	Milan<--------->Turin
    	  |		  |
    	  |		  |
    	Pisa<---------->Florence
    
    
    	So, for example, which interface address I've to use to register
    	the Milan's CISCO if my netmgt system is located in Pisa?
     	What if, from customer point of view, there isn't a more important
    	remote LAN, or more important serial link, but he wants only
    	know the state of his links (and he has some CISCOs with more than
    	20 interfaces all used..., up to 4 different Ethernets and 18
    	serial line).
    	By now, I was forced to register all the IP addresses for the more
    	important CISCOs giving to each one a different name (so forcing
    	the customer to forgot for his BIND services, thing that he doesn't
    	like at all), using the icon "line" to draw the remote links.
    
    	That's is a solution that the customer doesn't love.
    
    	So, solutions, also suggestion of other products (ie MCC/MSU), that
    	could better match its needs should be great.
    
    						Ciao 	Luciano
    
5480.3see also note 4988CSOADM::ROTHFormer K-notes, NOTES11 and Vnotes userFri Aug 20 1993 12:590