[c-nsp] Design issue for customer with dual MPLS links

[David Freedman]

jack daniels wrote:
> Hi Guys,
> 
> This is a paticular design issue I'm facing with customer where I have a lot
> of constrainits .
> 
> Topology
> 
> 
> MPLS CLOUD (ISP1)
> MPLS CLOUD (ISP2)
> |
> |
> |
> |
> |
> |
> CE1
> CE2
> |
>                     |
> |--------------------------------PIX525
> (CLUSTER)---------------------------------
>                                     |
>                                     |
>                                     |
>                                   LAN ( 6509 catalyst switch - runnning
> HSRP)
> 
> 
> Issue - I want  to go out via ISP1 and come back via ISP1 ......Backup is
> CE2
> 
> When traffic reaches PIX cluster how will it decide whether ISP1 is UP and
> traffic is not blackholed......How will PIX cluster decide to FWD traffic to
> ISP2.
> 
> Now for this solution , constraints are -
> 
> 1) I cant run HSRP on CE1 and CE2

Do you manage CE1/2 ? if not, how about getting the two ISPs to
co-operate and do HSRP/VRRP between eachother? this situation is not
impossible and I know of many examples (at least in Europe) where this
takes place.

> 2) Cant run run dynamic routing on PIX

PIX does RIP and I think some BGP now, no?

Even so if you don't manage CE1/2 you would have to get both managing
parties to enable this.

> 3) IP SLA also can't also be used on PIX cluster

Are the pix in L2 or L3 mode?
if in L2 could you not do IPSLA from the 6509 such to decide which CE
for egress?

If the pix are in L3 I suppose you could subdivide them into contexts
for ISP1/2 and have each of these statically tied to CE1/2 and then use
IPSLA on 6509 to fail between these? (yes, it would be messy and painful
managing both sets of rules)

Dave.


> 
> Regards
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>