[c-nsp] OSPF fast convergence

Justin Shore justin at justinshore.com
Wed May 13 23:12:51 EDT 2009


Phil Mayers wrote:
> Common advice seems to be to make actual link-loss detection fast, in 
> preference to using BFD. That said, I know some people use BFD.
> 
> Assuming you're using LAN cards, you may want to see if you can make 
> router links as routed rather than SVI interfaces. Though routed 
> interfaces are implemented internally as VLANs, presentations I saw from 
> Cisco claim that this:

I prefer to use BFD personally.  Link failure detection without BFD will 
be slow no matter what you do.  FRR doesn't gain you much if it takes 
you several seconds to realize that a link dropped.

I will point out one problem to Walter that may or may not be a big deal 
for his 7600s, depending on they're deployed.  BFD on SVIs is not 
supported or configurable beginning with SRB2.  It worked great but the 
feature was removed.  Search the archives for numerous lengthy 
discussions about the removed capability.  This feature is needed for 
people who deploy their 7600s in pairs and make some access-layer 
connections on them that require a VLAN one of more VLANs to span both 
7600s.  For example our 7600s serve as our core and for a few services 
like FTTH they serve as the L3 edge to our access layer.  We have a 1Q 
trunk on an Etherchannel link (for L1 redundancy) between the 7600s and 
run a FHRP across it (HSRPv2 in our case).  There isn't any other way to 
do this without a 1Q trunk between the chassis.  We also carry one VLAN 
across that trunk and build a L3 relationship across it between the 
7600s.  We've configured BFD on the SVI on both ends and it works great 
and will continue to work great until we upgrade (effectively 
downgrading BFD).  The only way around this is to dedicate a separate 
set of ports for the L3 connection on a dedicated Etherchannel link.  So 
now we're dedicating a minimum of 4 ports (2 per Etherchannel) to 
connecting our 2 chassis together for L2 and L3.  That's a big expense, 
even for GigE interfaces.  Imagine if you needed to do this for 10G 
interfaces.

If your 7600s are isolated islands and only route between the 7600s then 
this isn't a big problem.  However if you also want to carry multple 
VLANs between the 7600s then you need to be aware of these issues.

Justin


More information about the cisco-nsp mailing list