[c-nsp] BGP v4 multicast

[Phil Mayers]

Gert Doering wrote:
> Hi,
> 
> On Wed, Oct 01, 2008 at 04:50:49PM +0100, Phil Mayers wrote:
>> Just a quick question; what's the best syntax for causing an eBGP 
>> session with v4 multicast AF to originate routes?
> 

<sorry - missed the CC to list>

> Same way you currently do it with IPv4 unicast, just put the statements
> under "ipv4 multicast".
> 
> As in - if you're doing "redistribute static route-map foo", you can
> just put the same statement under "address-fam ipv4 multicast" as well.

Ok, that's what I wasn't clear on - whether the "v4 multi" AF would read
the normal unicast routing table, or whether it required "ip mroute"
statements to Null0 or something vile.

> 
>>   address-family ipv4 multicast
>>     neighbor 192.168.1.1 activate
>>     network 10.2.0.0 netmask 255.255.0.0
>>     network 10.3.0.0 netmask 255.255.0.0
> 
> This will work fine.
> 
>> ...and will it originate 10.2.0.0/16 & 10.3.0.0/16 via multicast to the 
>> eBGP peer? 
> 
> It will not "originate via multicast", but it will send these two prefixes

Yes sorry - typo

> in the IPv4 multicast address-family - via the normal TCP/IPv4 unicast 
> BGP session. Yes.
> 
>> Is this recommended?
> 
> Well, we have just removed all IPv4 multicast BGP from our network, since
> it doesn't work on "the global Internet" :-) - but technically, use whatever

Interesting; does that mean you have abandoned ASM and are pushing SSM,
or abandoned multicast completely (although given the lack of
SSM-enabled apps, they may be one and the same)

Sadly I am stuck supporting use of the "wonderful" AccessGrid
technologies (sigh) which more or less mandate multicast.

> works for you in IPv4 unicast to bring the prefixes into BGP.
> 
> gert
>