[c-nsp] downlink bgp interconnect best practices

[Nikolay Shopik]

On 31/05/11 14:50, Gert Doering wrote:
> We try to separate "core + uplink" and "customer connection" routers,
> so we can do works on core routers witout affecting customers - and
> vice versa, if we have to reboot a "customer connection" router, we
> know which customers are affected and that nothing else is.

Well in our case all customers fiber/copper terminated in same rack 
where is borders resides. So I just see no point to having additional 
router for customer except for additional redundancy of course. But this 
require router able to handle traffic and having bgp full-view for 
customer. If this router is NOT going to have hold bgp for customer its 
okey probably terminate both bgp sessions at border and connect them via 
L2 vlan/switch as we having now.

>
> So a "text book" scenario would have two "core/uplink" routers here,
> fully meshed with two "customer access" boxes (so there's no single
> switch in between that could break), and customers are then connected
> to one or both of the "access" routers.
>
> uplink/IX  uplink/IX
>      |        |
>      CR1 -- CR2
>       | \  / |
>       |  \/  |
>       |  /\  |
>       | /  \ |
>      AR1    AR2
>       |      |
>       customers
>
>
> Now, reality doesn't always work like this for smaller shops - if the
> customer connects with a bigger pipe than what we have to the access
> routers (like "2 Gbit ether channel uplink"), they get connected to
> the "core" box - because it's sometimes not very economic to roll out
> 10Gig everywhere, just for a single 2G customer...

Our looks like this, where CS1 and CS2 are core switches, 3524XL for 
now, but will be replaced with 3560X. So bgp session goes directly to 
CR2 for now.

  uplink  uplink/uplink/IX/IX
       |        |
       CR1 -- CR2
        | \  / |
        |  \/  |
        |  /\  |
        | /  \ |
       CS1    CS2
        |      |
        customers