From: Hall, Keith (KHall@xxxxxxxxxxxxxxxxx)
Date: Tue Nov 14 2000 - 15:53:00 GMT-3
Good question! Let me pose one back to you... ;)
In the case of r3-r2-r4 over NBMA from the scenario presented, r3 and r4 are
going through the same r2 interface/subinterface with ip split-horizon
disabled. If r3-r2-r4 use serial interfaces, these are different interfaces
- correct?
In the non-NBMA mode, there is a clear line of traffic flowing "downstream".
In NBMA mode, there should be an explicit join to start the flow of traffic.
"[W]hen each join is received from NBMA neighbors, PIM stores each neighbor
IP address/interface in the outgoing interface list for the group. When a
packet is destined for the group, the software replicates the packet and
unicasts (data-link unicasts) it to each neighbor that has joined the
group."
http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/12cgcr/np1_c
/1cprt1/1cmulti.htm
"[O]nly PIM WAN neighbors that have joined for the group will get packets
sent to as data link unicasts"
http://www.cisco.com/warp/public/cc/techno/protocol/ipmu/prodlit/mcmds_an.ht
m
If I haven't explicitly "joined" a group or caused an incidental
registration somehow in the direction of the other router (ex. for another
downstream router), how can a automatically join all configured multicast
groups on the network? Isn't that Dense Mode PIM? Doesn't that defeat the
whole purpose of Sparse Mode PIM?
Keith T. Hall
Sr. Network Engineer, Service Provider Accounts
Greenwich Technology Partners
3810 Concorde Parkway, Suite 500
Chantilly, VA 20151
(703) 966-1854 Cell
(703) 222-6465 Office
(703) 222-6424 Fax
khall@greenwichtech.com
http://www.greenwichtech.com
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Brian
Sent: Tuesday, November 14, 2000 12:47 PM
To: cciemail
Subject: Re: Repost: Multicast with RP as Spoke
Well I don't believe the pruning is causing the problem. Try setting up
between three routers connected by serial links without running it over a
multipoint interface in the middle. You can ping from the one edge router
to the other edge router which is the RP and has the join-group statement on
ite Ethernet port. Can anyone explain exactly what pim-nbma mode does,
because it doesn't seem to work in this scenerio?
Thanks
----- Original Message -----
From: Hall, Keith <KHall@greenwichtech.com>
To: <ccielab@groupstudy.com>
Sent: Tuesday, November 14, 2000 9:20 AM
Subject: RE: Repost: Multicast with RP as Spoke
> RE: the lack of ping responses with 225.4.4.4 from r3 or 225.3.3.3 from r4
>
> Jeez - I knew this Q would come up.. see pruning! ;)
>
> Remember that r4 and r3 are distant by a router hop. r4 is not
subscribing
> to 225.3.3.3 and r3 is not subscribing to 225.4.4.4 (and therefore I
believe
> it should be pruned under Sparse Mode PIM behavior - contrasted to Dense
> Mode PIM behavior). The incidental responses received via r2 without any
> 'real multicast source' and any 'real multicast subscriber' present is due
> to proximity; not actual subscription.
>
> r3 ------- r2 ------- r4
> 225.3.3.3 225.2.2.2 225.4.4.4
>
> Also, take a look at the overall design. In more complex scenarios, this
> can be quite a challenge to trace. Multicast pings are a little different
> from a standard unicast ping. 225.4.4.4 is on r4 (the RP). 225.4.4.4 is
> figuratively "at the root of the tree" or "at the head of the waters", so
to
> speak. (This is why I originally asked about the multicast source
> placement.) Bad RP placement can have a very detrimental effect on the
> multicast stream. Multicast implementation is very simple in small
> networks, but can be a real pain in the @#$@ in larger ones. (In fact,
I've
> heard of only partially successful or unstable implementations by other
> internetwork integration groups because of the lack of overall design and
> understanding of PIM mechanisms.)
>
>
> RE: Multicast Example
>
> If I understand correctly, you are able to perform a multicast ping in one
> direction??? Check if the 'broadcast' keyword is on r3...
>
> Yesterday, I made up my own OSPF config to test your scenario. The config
> presented was definitely not a full config (ex. no 'frame-relay
> interface-dlci' command) and I couldn't "cut-and-paste" it directly.
> Judging by the complexity of the original config (route-maps, ODR, and
> demand-circuit, et cetera), I would first try running the multicast script
I
> sent over a simplified, standard OSPF over NBMA config without route-maps
or
> other junk. Pls try this first..
>
>
> Keith T. Hall
> Sr. Network Engineer, Service Provider Accounts
> Greenwich Technology Partners
> 3810 Concorde Parkway, Suite 500
> Chantilly, VA 20151
> (703) 966-1854 Cell
> (703) 222-6465 Office
> (703) 222-6424 Fax
> khall@greenwichtech.com
> http://www.greenwichtech.com
>
>
>
>
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:25:45 GMT-3