[c-nsp] Multicast over VPLS

[Zitouni Rachid]

if you do the same on R4 to bridge vc-lsp R3-R4 and vc-lsp R4-R2 (through R5 LSR), you should be faced to 2 problems :

- a loop ...
Possible solution: using 2 dedicated circuits, the R1 router can be attached to vc-lsp R2-R3 (start of the chain) and to vc-lsp R4-R2 (end of the chain). And as R1 is PIM capable, RPF drops the returned traffic.   

- a bottleneck on R4-R5-R2 way :
If R5 becomes down the traffic sent through vc-lsp R4-R2 should use the R4-R3-R2 way 
Possible solution: vc-lsp R4-R2 can be constrained R2-R3-R4 LSP.

 
HiH,
Rachid





-----Message d'origine-----
De : alaerte.vidali at nsn.com [mailto:alaerte.vidali at nsn.com] 
Envoyé : vendredi 30 novembre 2007 15:07
À : Zitouni Rachid; cisco-nsp at puck.nether.net
Objet : RE: [c-nsp] Multicast over VPLS

Great, tks.

And it seems it would solve the waste of bandwidth problem by itself. (no duplicate flows from R2 to R3) 

>>2 ethernet ports connected to each other. The first one is bridged to the vc-lsp R2-R3 and the second one to the vc-lsp R3-R4.


The other option you suggested seems very interesting, as it would solve 2 problems, avoid duplicate same multicast flow and avoid send multicast when there is no interest:

>>igmp snooping on vc-lsp avoiding unnecessary multicast replication.  

You meant contruct vc-lsp between R2 and R3 and a vc-lsp between R2 and R4. You did so because it would not work with VPLS?


Tks,
Alaerte

-----Original Message-----
From: ext Zitouni Rachid [mailto:rzitouni at prosodie.com] 
Sent: Friday, November 30, 2007 11:45 AM
To: Vidali Alaerte (NSN - BR/Rio de Janeiro); cisco-nsp at puck.nether.net
Subject: RE: [c-nsp] Multicast over VPLS

Could be fine if you can bridge between 2 vc lsp with a dedicated "bridge-group"
Or you can use, for example, 2 ethernet ports connected to each other. The first one is bridged to the vc-lsp R2-R3 and the second one to the vc-lsp R3-R4.

HiH,
Rachid
   

-----Message d'origine-----
De : alaerte.vidali at nsn.com [mailto:alaerte.vidali at nsn.com] Envoyé : vendredi 30 novembre 2007 14:31 À : Zitouni Rachid; cisco-nsp at puck.nether.net Objet : RE: [c-nsp] Multicast over VPLS

>>physical ethernet link circuit to bridge traffic between the R2-R3 vc-lsp and the R3-R4 vc-lsp.

I think I did not get it. If I understood your suggestion, on the topology R3 needs to bridge traffic received from R2--R3 to R3--R4 vc-lsp. 
Is that correct? If yes, you are thinking about bridge-group between 2 vc-lsps?


                                             user             user
                                              |                |
Multicast_Server------R1-----R2(gi1)-----(gi1)R3(gi2)-----(gi1)R4(gi2)----(gi1)R5(gi2)
                               |                                                |
                               |________________________________________________|


Tks,
Alaerte

 

-----Original Message-----
From: ext Zitouni Rachid [mailto:rzitouni at prosodie.com]
Sent: Friday, November 30, 2007 11:06 AM
To: Vidali Alaerte (NSN - BR/Rio de Janeiro); cisco-nsp at puck.nether.net
Subject: RE: [c-nsp] Multicast over VPLS

The easiest is to consider a physical ethernet link circuit to bridge traffic between the R2-R3 vc-lsp and the R3-R4 vc-lsp.

HiH,
Rachid 

-----Message d'origine-----
De : alaerte.vidali at nsn.com [mailto:alaerte.vidali at nsn.com] Envoyé : vendredi 30 novembre 2007 13:29 À : Zitouni Rachid; cisco-nsp at puck.nether.net Objet : RE: [c-nsp] Multicast over VPLS

Hi and thanks,

 >>Other way is to make circuit loop on R3 and establish vc-lsps between R2 and R3 then R3 and R4

If I establish VC-LSP from R2---R3, and VC-LSP from R3---R4, how would R3 switch what it received from R2 to R4?
Throught layer 3? 

Br,
Alaerte

-----Original Message-----
From: ext Zitouni Rachid [mailto:rzitouni at prosodie.com]
Sent: Friday, November 30, 2007 7:39 AM
To: Vidali Alaerte (NSN - BR/Rio de Janeiro); cisco-nsp at puck.nether.net
Subject: RE: [c-nsp] Multicast over VPLS


No solution within a single vpls domain wich is by definition a broadcast domain.
If you have a vc-lsp between R2 and R3 and a vc-lsp between R2 and R4 You can optimize using igmp snooping on vc-lsp avoiding unnecessary multicast replication.  
Other way is to make circuit loop on R3 and establish vc-lsps between R2 and R3 then R3 and R4

HiH
Rachid 

-----Message d'origine-----
De : cisco-nsp-bounces at puck.nether.net [mailto:cisco-nsp-bounces at puck.nether.net] De la part de alaerte.vidali at nsn.com Envoyé : jeudi 29 novembre 2007 19:30 À : cisco-nsp at puck.nether.net Objet : [c-nsp] Multicast over VPLS

Hi,

Any information about what draft Cisco is considering/will adopt to solve bandwidth waste issue with Multicast over VPLS?

Besides standard, do you see any solution currently available to avoid PE to send several flows of the same multicast over a single link on ring topology?


Topology:

VPLS between R2/R3/R4/R5
                                   user     user
                                     |       |
Multicast_Server------R1-----R2-----R3-----R4----R5
                                                
By default, R2 will send 3 times the same flow on link R2---R3.

Tks a lot,
Alaerte
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/