Re: PIM-Spare - Shared and SPT

Hi Mike,

Having configured sparse mode all over the network, the process is as
follows:

[receiver device] --------------[L2 cloud]----------R1-----------[L3
cloud]---------------RP---------------R2--------------[sending device]

1. Receiver IGMP joins. This frame terminates on the first L3 device, say
R1. All L2 devices then build an igmp table with the mac-address and exit
interface. The router hears it and sends PIM (*,G) join upstream towards
the RP. This creates the shared tree from RP to R1. If all is well, this
means if you do a show ip mroute along the path, you have the right (*,G)
output. There isnt any (S,G) entries yet.

2. Sender sends traffic. This normally goes through the PIM-Reg
encapsulation process. However, if RP is directly attached, this step is
skipped.

3. The PIM-Reg encapsulation process is the sending of the mast packets,
encapsulated in unicast from the first router, say R2, closer to the
sender, directly to the RP until the source tree is formed from the RP back
to the first router, R2, near the sender. Once the first unicast
encapsulated mast packet is received, the RP sends it downstream towards
the receiver and at the same time, send back an PIM join back towards R2.
This sending of the PIM join back to R2 creates the SPT tree and as soon as
R2 has this tree completed, declares the PIM-Reg process complete and
immediately switches over to send mcast natively towards the RP.

4. RP sends this traffic down the shared tree from above. This should reach
the receiver.

5. Normally, R1 should trigger a conversion from shared tree to source
tree towards the source, R2. This means (S,G) join from R1 towards R2 on
the network. All of this is subject to the unicast routing table metrics
and rules, as you know, since we are using PIM.

6. When source tree is good, we trim down the shared tree. All traffic
should now flow down the (S,G) entries.

Notice that in step 5, we said normally converts. Sometimes, we can make
it to not convert, and all traffic will simply flow down the shared tree.
This is done by a config on the last hop router -- in this case, R1  by
this line: pim spt-threshold infinity.

Hope that clarifies it up bait. Let me know if it does not make much sense
and I can try again.

Sadiq

On Mon, Dec 12, 2011 at 4:53 AM, ccie routing <mike.ccie_at_gmail.com> wrote:

> Hi,
>
> Can some one point me in the right direction regarding the PIM sparse
> operation. This is what I understand so far....
> Pim sparse has a (*,G) entry between receiver and RP and an (S,G) entry
> between source and RP.
> After the first multicast packet is received PIM sparse forms an (S,G)
> between source and receiver (SPT).
>
> Is this the correct understanding ? I would appreciate if someone can
> direct me to any link that explains pim sparse in depth.
>
>
> Thanks,
> Mike
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
>

--
CCIEx2 (R&S|Sec) #19963
Blogs and organic groups at http://www.ccie.net