The MPLS-OPS Archive

[Mostazir Rahman <mostazir@yahoo.com>]

Can somebody tell how to unsubscribed to this group?
Thanks.
-Mostazir

--- "Qian, Edward" <Edward.Qian@tekelec.com> wrote:

> Vinay,
>  
> Thanks. It is convincing in yet another perspective
> of "layer inverse"
> as you pointed out.
>  
> I've also received response from other group of
> people. I would like to
> share one of them to this group:
>  
> --------------
> Edward,
> Taking this off-the-list for the moment.
> To the best of my understanding VRRP and MPLS don't
> mix - for the
> following reasons:
> 
> 1.	
> 	VRRP switches do not involve the host in any way,
> the only
> entities that recognize the switch are the routers
> in the VR group and
> the L2 switch between the host and these routers (it
> may notice that
> frames with a certain DMAC address are now forwarded
> to a different
> port). Since the host remains unaware of the switch,
> it cannot take any
> actions on it as well:-)
> 2.	
> 	VRRP can protect outgoing IP traffic because, once
> accepted by
> any router in the VR group, IP packets are  further
> forwarded according
> to their  (global) Destination IP addresses.  
> 3.	
> 	Labels, as opposed to IP addresses, only have
> meaning in the
> scope of the router that has allocated them, bound
> them to some FEC and
> distributed them. (In some cases the scope of a
> label is limited to a
> given interface of the router).  And there is no way
> to force binding of
> a given label to the same FEC in a group of routers.
> (Even manual
> configuration can fail, because the pools of free
> labels in a pair of
> routers may happen to be disjoint).
> 4.	
> 	As a consequence, VRRP CANNOT protect labeled
> traffic.
> 5.	
> 	Note also that router that is part of a VR group
> but does not
> own the virtual IP address will not recognize any
> reference to this
> address anywhere except ARP requests. In particular,
> it will not
> recognize this address in the ERO object in the
> RSVP-TE Path message.
> This means that if you have used this address in the
> ERO object, one of
> the two canhappen:
> 
> 	*	
> 		Master owns this address. In this case the LSP
> will be
> built and will pass through the Master, but slave
> will not be able to
> restore the path if the Master fails
> 	*	
> 		Neither Master nor Slave own this address. In this
> cas
> the LSP will not be built at all.
> 
> So, the answer to your question is, AFAIK: NO, VRRP
> does not work with
> MPLS tunnels.
> Hopefully these notes will be useful.
> With best regards,
> Sasha Vainshtein
> email:     sasha@axerra.com
> <mailto:sasha@axerra.com> 
> --------------
>  
> Regards,
> Edward
> 
> ________________________________
> 
> From: vinaychandra.sham@wipro.com
> [mailto:vinaychandra.sham@wipro.com] 
> Sent: Thursday, September 30, 2004 2:24 AM
> To: Qian, Edward
> Cc: mpls-ops@mplsrc.com
> Subject: RE: [mpls] MPLS with VRRP
> 
> 
> 
> Edward,
> 
>  
> 
> You are basically looking at layer 2.5 reliability
> through a layer 3
> mechanism (VRRP). You can use fast-re-route
> extensions to RSVPTE to
> achieve this same functionality without using VRRP.
> 
>  
> 
> In the set-up given, the slave would reject all the
> RSVP-TE signaling
> message sent on the VMAC/VRIP leaving the salve in
> the cold about the
> tunnel. And when the failover happens the back-up
> router would drop the
> packets.
> 
>  
> 
> Regards,
> 
> Vinay.
> 
> ________________________________
> 
> From: mpls-bounces@lists.ietf.org
> [mailto:mpls-bounces@lists.ietf.org]
> On Behalf Of Qian, Edward
> Sent: Tuesday, September 28, 2004 2:57 AM
> To: mpls@ietf.org
> Subject: [mpls] MPLS with VRRP
> 
>  
> 
> Hello,
> 
>  
> 
> Could someone please share any experience when
> setting up MPLS tunnels
> via VRRP routers?
> 
>  
> 
> Here is a tricky case:
> 
> I have a host connected to two routers (running VRRP
> as a virtual
> default gateway). 
> 
> My host is actually MPLS capable (as a tunnel
> ingress LER) to setup TE
> tunnels via either one of 
> 
> the two neighbor routers to a tunnel egress LER.
> 
> At my host, I choose to use explicit route with
> RSVP-TE ERO to set up
> such a TE tunnel. For the explicit hop list, 
> 
> the first hop would be one of the VRRP routers, say
> I choose the master
> Router A (slave is Router B).
> 
> I should expect to happily send traffic out of my
> host through this TE
> tunnel via Router A if everything runs well.
> 
> Note that Router B has no knowledge of my TE tunnel.
> 
>  
> 
> Assuming Router A has a failure occurred, it is then
> expected that
> Router B should take over as a new master.
> 
> What would happen to my TE tunnel?  My host will
> keep sending labeled
> packets over the Ethernet frames 
> 
> to the same virtual MAC which is also recognized by
> Router B. Router B,
> however, knows nothing about 
> 
> the TE tunnel. So all the TE tunnel traffic will be
> dropped.
> 
> (or just imagining that Router B may even
> intelligently strip off the
> label and perform plain hop-by-hop forwarding to the
> destination,
> 
> that's to say all the packets are leaked out of my
> TE tunnel).
> 
>  
> 
> So my question is: Does VRRP really work with MPLS
> tunnels?
> 
=== message truncated ===

-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml