PIM-BiDir ECMP

[Harindha Fernando]

Hi All,

in PIM-BiDir ECMP is there any significance of the IP address of a shared media (LAN) to elect the DF ? isn't it the router which is having best IGP cost to RP be elected as DF irrespective of interface IP addresses ?

http://www.cisco.com/c/en/us/td/docs/ios/12_4t/ip_mcast/configuration/guide/mctlsplt.html

in the above document 

If bidir-PIM is used in the topology illustrated in Figure 3, a process called DF election would take place between Router 2, Router 3, and Router 4 on Ethernet interface 1. The process of DF election would elect one router for each RP to forward traffic across Ethernet interface 1 for any groups using that particular RP, based on the router with the highest IP address configured for that interface. Even if multiple RPs are used (for example one for G1 and another one for G2), the DF election for those RPs would always be won by the router that has the higher IP address configured on Ethernet interface 1 (either Router 3 or Router 4 in this topology). The election rules used for DF election are virtually the same as the election rules used for the PIM assert process, only the protocol mechanisms to negotiate them are more refined for DF election (in order to return the results more expediently). As a result, when bidir-PIM is used in this topology, load splitting would always occur across Ethernet interface 1.

how can we have multicast load splitting on PIM-BiDir network if we have only one RP for all the groups ?

Rgds

Harin

[Harindha Fernando]

Hi All,

Is this group active or I have posted something not related to CCDE ??

Rgds

Harin

[Marcus Ueda]

Hi,

i dont know how  PIM-BiDir works,

sorry...

--
You received this message because you are subscribed to the Google Groups "CCDE Group Study" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ccdegroupstud...@googlegroups.com.
To post to this group, send email to ccdegro...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Marcus S. Ueda

CCDA

[Ethan Banks]

Harin, the group is alive, but "active" is not the word to describe the activity level here. I think many folks pursuing CCDE are quite busy, and find it difficult to respond.

I understand the question you are asking, but cannot offer a response. I would need to create a lab exercise and do some testing to come up with a definitive answer. I work with multicast very little in my current production environment.

/Ethan

On 23 Jul 2014, at 8:01, Harindha Fernando wrote:

Hi All,

Is this group active or I have posted something not related to CCDE ??

Rgds
Harin

On Tuesday, July 22, 2014 2:10:41 PM UTC+3, Harindha Fernando wrote:

Hi All,

in PIM-BiDir ECMP is there any significance of the IP address of a shared
media (LAN) to elect the DF ? isn't it the router which is having best IGP
cost to RP be elected as DF irrespective of interface IP addresses ?

http://www.cisco.com/c/en/us/td/docs/ios/12_4t/ip_mcast/configuration/guide/mctlsplt.html

in the above document

If bidir-PIM is used in the topology illustrated in Figure 3

http://www.cisco.com/c/en/us/td/docs/ios/12_4t/ip_mcast/configuration/guide/mctlsplt.html#wp1063486,

a process called DF election would take place between Router 2, Router 3,
and Router 4 on Ethernet interface 1. The process of DF election would
elect one router for each RP to forward traffic across Ethernet interface 1
for any groups using that particular RP, based on the router with the
highest IP address configured for that interface. Even if multiple RPs
are used (for example one for G1 and another one for G2), the DF election
for those RPs would always be won by the router that has the higher IP
address configured on Ethernet interface 1 (either Router 3 or Router 4 in
this topology). The election rules used for DF election are virtually the
same as the election rules used for the PIM assert process, only the
protocol mechanisms to negotiate them are more refined for DF election (in
order to return the results more expediently). As a result, when bidir-PIM
is used in this topology, load splitting would always occur across Ethernet
interface 1.

how can we have multicast load splitting on PIM-BiDir network if we have
only one RP for all the groups ?

Rgds
Harin

--

[Jesse L. CCIE#14661]

Harin,
A portion of the answers you seek can be found here in the BIDR PIM Deployement Guide [http://www.cisco.com/c/dam/en/us/products/collateral/ios-nx-os-software/multicast-enterprise/prod_white_paper0900aecd80310db2.pdf]

"The algorithm to elect the designated forwarder is straight forward, all the PIM neighbors in the subnet advertise their unicast route to the rendezvous point and the router with the best route is elected."

[Harindha Fernando]

Hi Jesse/Ethan,

Thanks for your reply, yes as per the RFC 5015 it checks only the cost to reach RP from each PIM-BiDir Router, and the router with best metric to RP will be elected as DF. 

but in PIM-SM RFC4601 the highest IP address will be selected as DR only if it's having equal cost to RP as per section 4.6.3

In this particular ECMP scenario for PIM-BiDir should be using PIM-SM DR election mechanism to elect the DF for PIM-BiDir in the LAN as the cost to reach RP will be same from the R3 and R4 in Figure 3 (http://www.cisco.com/c/en/us/td/docs/ios/12_4t/ip_mcast/configuration/guide/mctlsplt.html#wp1060457)

I could not test it in a LAB.. 

Rgds

Harin

[dip]

Hi , 

I agree that in the  case of PIM Bi-dir ECMP scenario which you mentioned , if the cost to RP is same then the router with highest IP address will be elected as the DF for that segment .

Thanks

Dip

[Harindha Fernando]

But then it's not based on PIM-BiDir DF election process and it's only based on PIM-SM DR election based on assert isn't it?

So in PIM-BiDir ECMP scenario there is no role for DF.

Rgds
Harin

[dip]

So yes at a high level there are similarities between PIM DF and SM assert process and you can say that in case of shared medium the outcome is same.  

 

In case of Pim BiDir ,selection of the DF isn't a data driven event. Rather, for each RP, a DF is elected on each subnet, and on each point-to-point link, before any packets arrive. This is similar to PIM DR election process in SM .

In PIM SM assert process, the trigger is that upstream routers receive same S,G stream from there RPF interfaces and forwards to the  common ethernet segment .They receive that S,G packet on there shared ethernet segment which is in there OIL list and hence it triggers the assert process.So to me PIM assert becomes a data driven event.

Also in the case of PIM Bidir, elected DF node serves as DF for all multicast groups per RP and as you already know that its based on the lowest unicast cost to the RP, and IP address is used to break the tie. In case of P2P link, RPF neighbor is the DF.

So in case of PIM BiDir , if i have three ECMP paths to the RP from three different upstream neighbors on three different P2P links , DF will be elected based on highest IP address which you can say is similar to what will happen with PIM SM as well. 

Difference comes in whenlets say I enable  "ip multicast multipath  s-g-hash next-hop-based " . In case of PIM SM , you will see that based on your hash , (*,G) entries are load balanced across the three upstream neighbors. In case of PIM BiDir, there is no effect since DF is elected per RP for multiple multicast groups so DF election process still remains the same i.e. lowest unicast cost to the RP, and IP address to break the tie. Hence out of three upstream neighbors you will only have one  DF in case of Pim BiDir.

Hoping at this point i haven't confused you anymore .

[Kartick K R]

HI,

Thanks for the detailed explanation.
Just want to be clear, is my conclusion correct:
In PIM Bidir case, load balancing will not happen using the ECMP paths.

Regards,
Kartick

[dip]

yes sir .In order to avoid forwarding loops traffic flows along only one shared tree.

[dip]

argh ...i meant ,your conclusion is correct . For BiDir groups traffic flows through only one shared tree to avoid forwarding loops.