[c-nsp] 65xx or 76xx for 'Distribution Layer'?

[Phil Mayers]

On Wed, 2007-10-17 at 10:11 -0400, Tim Durack wrote:
> Until recently the 65xx and 76xx were the same car, just a different
> paint job. Now the BUs are "differentiating" the boxes.
> 
> 65xx will be running the 12.2SX train. This is aimed more at the
> enterprise/DC space. If you want modular IOS, and MPLS features
> targeted more towards the Enterprise, this looks like the way to go.

Some of the high density linecards e.g. 6708 don't and apparently won't
be supported in 7600/12.2SR

> 
> 76xx will be running the 12.2SR train. This looks to be aimed at the
> SP space. Broadband aggregation support, all the fancy VPLS/H-VPLS
> MPLS stuff.

Some of the "high touch" cards (e.g. ES20) won't be supported in 6500.
Also RSP720 is not supported (I wouldn't care, except the CPU speed
increase would be nice to have in some cases)


> 
> Our last core upgrade got us OSPF as an IGP, with some limited areas.
> I regret the mutiple area stuff. In future we will go OSPF with a

I can support this. OSPF areas are not IMHO a good idea in most cases.
Just go for a tiny p2p+loopback OSPF area 0 and iBGP. We moved from
all-OSPF recently and it's been nothing but good for us.

> single area 0 for loopbacks/router-router networks, and BGP for
> "customer" routes (read access switches.) BGP is additional
> complexity, but we need it for globally reachable prefixes anyway. It
> works for the Internet...
>