[j-nsp] Upstream Traffic Manipulation Question

[William Jackson]

Maybe I am being too basic, but if both remote POPs are from the same
ISP, maybe they have communities you can add to set their localpref?

What about MED on BGP?


William 



-----Original Message-----
From: juniper-nsp-bounces at puck.nether.net
[mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of Stefan Fouant
Sent: 05 October 2009 08:47
To: Walaa Abdel razzak
Cc: juniper-nsp at puck.nether.net
Subject: Re: [j-nsp] Upstream Traffic Manipulation Question

On Sun, Oct 4, 2009 at 9:28 AM, Walaa Abdel razzak
<walaaez at bmc.com.sa>wrote:

> Hi Experts
>
> I need your suggestions for the best design for the following
scenario:
> it's a customer has a public AS and MPLS network, connected to
upstream
> provider through two links in two differnt POP's. The customer
receives the
> default route and few prefixes only from the provider from both POP's.
The
> key here is that the customer needs certain subnets to follow the
default
> route coming from POP1 in the upstream, the downstream must be through
the
> same POP1. Other subnets should follow the default route coming from
POP2 in
> the upstream, the downstream must also be through the same POP2. My
question
> here is not in the downstream as it's easy to adjust through BGP
attributes
> and prefix length, but it's in the upstream direction to the internet.
how I
> can gaurantee that certain prefixes follows the default route through
one
> POP will be back from the same POP? I was thinking about splitting
customer
> traffic into two VRF's and inject default routes with different
prefernces.
> do you think!
>  it's the best or you have another susggestion? The customer has no
other
> provider.
>

For source-based routing of packets, you are going to need to perform
Filter-Based-Forwarding (FBF) which will then cause the packet to get
routed
into a routing-instance of your choosing.  You are going to need to
constrain the set of routes in the routing-instance such that it will
route
out towards the appropriate POP.  The problem here is that you are going
to
need to configure this FBF and routing-instance construct on every
router so
it could get a bit cumbersome.  There are a lot of different ways in
which
you could distribute the routes into your routing-instance - a
consistent
policy using rib-groups could be configured across all your routers to
leak
only certain routes into your routing-instance, or you could set up
additional peerings within your routing-instance/VRF for dynamic
advertisement of the constrained routes.  Depending on the way that the
MPLS
network is set up, you might be able to get away with configuring the
FBF on
only the head-end router and punting that traffic into the appropriate
LSP
for transit across the AS.

I am sure there are other ways... I'd be interested in hearing others'
feedback as well...

Cheers,

-- 
Stefan Fouant
_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp