From: <nar...@wi...> - 2010-08-26 14:04:12
|
>I think Surya's response explains the situation pretty well. Let us know >if you have further questions. Yes. Surya comments were clear to my questions. I will look more deep in to tipc and come back. thanks. Best regards Naresh Kamboju >Regards, >Al >> -----Original Message----- >> From: Suryanarayana Garlapati [mailto:SGa...@go...] >> Sent: Thursday, August 26, 2010 2:38 AM >> To: nar...@wi...; Stephens, Allan >> Cc: jon...@er...; tip...@li... >> Subject: RE: [tipc-discussion] Request: Unable to config >> second node onsamemachine >> >> Hi Naresh, >> For the scenario 3: >> > scenario 3: two Ethernet interfaces eth0, eth1 >> > >> > Machine-1 Machine-2 >> > |------------- | |---------------| >> > | <1.1.1> | |<1.1.2> | >> > | eth0 | ??? | eth0 | >> > | eth1 | ??? | eth1 | >> > --------------- ---------------- >> >> TIPC expects that there is connectivity between eth0-eth0 and >> eth1-eth1 and there is no intermixing among them(means >> totally distinct Ethernet segements (or independent LANs, if >> you prefer)). >> See the diagram below >> >> Machine-1 Machine-2 >> |------------- | |---------------| >> | <1.1.1> | |<1.1.2> | >> | eth0 | <-------------------> | eth0 | >> | eth1 | <-------------------> | eth1 | >> --------------- ---------------- >> >> Say connectivity between eth0-eth0 is broken, then TIPC >> messaging will be done on the eth1-eth1 link. TIPC >> automatically does the link changeover, when it detects the >> failure on one link. >> >> Regards >> Surya >> >> >> > -----Original Message----- >> > From: nar...@wi... [mailto:nar...@wi...] >> > Sent: Thursday, August 26, 2010 11:54 AM >> > To: all...@wi... >> > Cc: jon...@er...; Suryanarayana Garlapati; tipc- >> > dis...@li... >> > Subject: RE: [tipc-discussion] Request: Unable to config >> second node >> > onsame machine >> > >> > >> > Allan wrote: >> > > >> > >As Jon Maloy stated in his reply, if you want a node to have >> > >redundant TIPC interfaces to Ethernet they have be on totally >> > >distinct Ethernet segements (or independent LANs, if you >> prefer). You >> > >can picture this >> > as: >> > > >> > > |---<1.1.1>---| >> > > | | >> > > |---<1.1.2>---| >> > > | | >> > > |---<1.1.3>---| >> > > eth0 eth1 >> > > LAN LAN >> > > >> > >While you might be able to get some redundancy by eliminating the >> > >distinct segments requirement, as in the following case: >> > > >> > > |---<1.1.1>---| >> > > | | >> > > |---<1.1.2>---| >> > > | | >> > > |---<1.1.3>---| >> > > | | >> > > +-------------+ >> > > single LAN >> > > >> > >you wouldn't be protected in the case that a single defective >> > >Ethernet interface starts babbling away on the LAN and >> corrupts all >> > >traffic >> > being >> > >sent on the LAN. Having the requirement for totally isolated LANs >> > means >> > >that you're protected even in this case. >> > >> > valuable information. thanks. >> > let me give a picture of the same: >> > >> > scenario 1: single Ethernet interface from eth0 >> > >> > Machine-1 Machine-2 >> > |------------- | |---------------| >> > | <1.1.1> | |<1.1.2> | >> > | eth0 |<-------------------->| eth0 | >> > --------------- ---------------- >> > Connection established between (<1.1.1>eth0; <1.1.2>eth0) >> working fine >> > >> > scenario 2: single Ethernet interface from eth1 >> > >> > Machine-1 Machine-2 >> > |------------- | |---------------| >> > | <1.1.1> | |<1.1.2> | >> > | eth1 |<-------------------->| eth1 | >> > --------------- ---------------- >> > Connection established between (<1.1.1>eth1; <1.1.2>eth1) >> working fine >> > >> > scenario 3: two Ethernet interfaces eth0, eth1 >> > >> > Machine-1 Machine-2 >> > |------------- | |---------------| >> > | <1.1.1> | |<1.1.2> | >> > | eth0 | ??? | eth0 | >> > | eth1 | ??? | eth1 | >> > --------------- ---------------- >> > >> > In Scenario 3, what are the probabilities for effective ways of >> > communicating Machines by TIPC. >> > >> > May i know the backup plan for the connection, i mean if connection >> > lost from one Ethernet interface what could be the ways to re- >> > establishing connection curious about the Link Changeover mechanism. >> > >> > Best regards >> > Naresh Kamboju >> > > > >Regards, > > >Al |