GLOBAL VRF ROUTE LEAKING

[Babak40]

Hi Guys ,

In our core router we have 20 VRFs and by using static route all traffic goes to ISP1 .

Our router is Peered with a Carried Neutral (we name it XX)  as well to make web browsing faster.

The problem is at the moment my global routing table has 60000 bgp routes which has received from XX

and if I want to inject them to each VRF ,it uses 20*60000 route entries which kills the memory 

How can I share global BGP table with Vrfs and send some traffics to XX ?

Thanks 

[Nagendra Kumar Nainar]

Hi,

For each of those 20 VRFs, are you using a static default route?. If so, I think you can configure another vrf specific default static route with NH from global.

Below is an example:

ip route vrf TEST 0.0.0.0 0.0.0.0 10.1.23.3
ip route vrf TEST 0.0.0.0 0.0.0.0 10.1.25.5 global

The above will install 2 default static routes in VRF TEST table with 1 NH as 10.1.23.3 from VRF TEST and other NH 10.1.25.5 from global routing table.

Below is the CEF output:

R2#sh ip cef vrf TEST 0.0.0.0/0 detail
0.0.0.0/0, epoch 0, flags [default route], per-destination sharing
recursive via 10.1.23.3
recursive via 10.1.23.0/24
attached to Ethernet0/0.2
recursive via 10.1.25.5
recursive via 10.1.25.0/24
attached to Ethernet0/0.25
R2#

HTH,

Nagendra

[paul driver]

Hello

just like to add you will need to tell the global rib how to each your vrfs  so you’ll need a static route stating the vrf subnet and point it towards the vrf interface 

[Nagendra Kumar Nainar]

Yes right, for the return traffic from XX to hosts in VRF. Thanks for pointing

[Babak40]

Hi Paul,
I replied in the group please have a look .thanks

[Babak40]

Hi Nagendra ,
The problem is one of them is ISP the other one is not ISP it''s carrier
Neutral it means we can reach to half of our destination through this
carrier faster ,but not all destination.

It means if I put my default route to this carrier half of them will be
dropped .

This carrier has redistributed 60000 routes to bgp and those routed are
most for Facebook,Google,....

So if I go to Facebook from my ISP it takes 10 hops to reach the
destination but using carried takes 5 hops

One solution that comes to my mind is redistribute those bgp routes to my
vrfs but imagine 20*60000 routes ...router will die

Is there any idea ?

[paul driver]

Hello

so would like to push certain vrf internal traffic via the carrier fast path?

[Babak40]

Yes Exactly ,I want to push some VRF's to use the carrier for half of their traffic and ISP for the rest of traffic 

Is there any way to share the bgp table of the router to vrfs without redistribute whole routes to vrfs 

[paul driver]

Hello

sounds like policy based routing would be applicable however depending on how your topology is connected you could manipulate traffic flow via bgp/igp route manipulation 

can you shate a topology 

[rais]

How about asking ISP1 for BGP and pointing default route to XX.

HTH.