[c-nsp] pros and cons for IPTV multicast in rosen-mvpn vs GRT

[Oliver Boehmer (oboehmer)]

>> easier to reduce loss-of-connectivity after link/node failures using IGP
>Fast Convergence compared to BGP-based convergence.
>Yes the convergence time would be slower I suppose as a mere addition of
>another protocols to the picture.
>Though if you consider the whole LoC timeframe the IGP or BGP convergence
>is
>one of the smallest portions -compared to the PIM convergence time.

well, that really depends on the failure and platform, but BGP
control-plane convergence is generally a magnitude higher than PIM. So I
would not discount this if you're after sub-second convergence. IOS-XR and
halfway recent IOS /XE releases use an source-based, event-triggered RPF
check and will be able to send out the required PIM Joins quite fast, so
the time difference of BGP vs. IGP will make a difference..
Core link failures would still be handled quickly, here we talk about
converging the MDT S,G..
 
> 
>
>> Why do you consider putting it into a VRF?
>Well the main concern my boss has is the exposure of core/global routing
>table to set-top-boxes -as a potential attack vector

Hmm, if the global routing is limited to the IP-TV Vlan, locking it down
via generic ACLs is easy as the traffic sources and type will be very
limited.

	oli
 
>-----Original Message-----
>From: Oliver Boehmer (oboehmer) [mailto:oboehmer at cisco.com]
>Sent: Monday, February 18, 2013 4:25 PM
>To: Adam Vitkovsky; cisco-nsp at puck.nether.net
>Subject: Re: [c-nsp] pros and cons for IPTV multicast in rosen-mvpn vs GRT
>
> 
>>Are there any cons for running IPTV in draft-rosen-mvpn as opposed to
>>global routing table
>
>current implementation makes it generally easier to reduce
>loss-of-connectivity after link/node failures using IGP Fast Convergence
>compared to BGP-based convergence when the mcast sources/TV headends are
>visible in BGP. You can also use p2mp-TE-FRR when mcast is in the global
>routing table (not sure if things have changed there). MoFRR is also
>targeted for PIM deployments in the global table, and some live-live
>approaches might not work as well if the sources aren't visible from the
>core/P nodes.
>
>Why do you consider putting it into a VRF?
>
>I acknowledge that reasoning is highly dependent on specific network
>topology and requirements..
>
>	oli
>
>