Routing

Hi ,

Please help me to understand the following:

why routing protocol is defined under instance in L3vpn and why not define globally????

>>>set routing instance site-1 protocol bgp group EBGP type external neighbor 1.1.1.1 peer-as 100

My understanding:

routing instance contains RD, RT, CE facing interface

since, we have CE facing interface under routing instance it means that whatever traffic comes in that CE interface should be placed on the VRF table. but when i define the routing protocol globally ,my CE facing interface receives the packet and should keep it in my VRF table. but this doesn't happens.  

"why routing protocol is defined under instance in L3vpn and why not define globally????"

The reason is that every route learned from the CE has to go into the vrf-table (<vrf-name>.inet.0) and not into the global table (inet.0) to keep routes separate from other VPNs (Customers)

a routing-instance (in this case vrf) consists of routing table(s), logical interfaces AND routing protocols

regards

alexander

Hi Alex,  Thank your for your answer.

But the confusion is now between data traffic and control plane traffic.

Since the CE interface is defined under routing instance, the routing updates we get on CE should automatically go to VRF instance right?? as like the data traffic.

The data traffic which comes on CE interface will automatically go to VRF instance then why not the control plane traffic(routing updates)????

Hi,

There are two ways you have to look at this

1. traffic till ce - pe

2. traffic after reaching Pe

traffic be it be control plane or data plane as you have righly put will land on PE device VRF instance. Now, there is a MP-BGP which runs between Two PE routers globally and MPLS LSP which runs between two PE routers. These global protocols help to transport traffice (control and data of Multiple VRF Instances) while protocols at CE/VRF Level will help to communicate between your CE and PE Devices , hence you have seperation of Global and Local at VRF level so that customers can be identified based on RD/RT

Hi Rakesh,  Thank you so much for sharing your views.

________         ge0/0/1 ___________ 

|  CE        |------------  |        PE                 |

 -------------                    ___________

                                       VRF1 table:

                                         RD 1:1

                                         RT  2:3

                                        CE interface (ge0/0/1)   <<<<---------- this means that, any traffic that comes on the interface ge-0/0/1 should be placed on VRF 1 (even though ebgp protocol is defined in global level, routing updates hits the ge0/0/1 interface so it should obviously be placed in VRF1 table ryt????) if this is the case , why we need to define the protocol configuration under the routing instance

Hi,

Imangine this is the scenario

r1----bgp----r2-----bgp----CE(r3)-------PE1---------------------PE2------------CE(r4)----------r5-----------r6

Now r3 who is the cutomer edge router has already his internal network setup , how will PE1 vrf know about his internal network routes without running dynamic protocol, sure you can write static routes but that isnt managable, hence you need to have a dynamic routing protocol which gets all the customer routes from CE and then redistributes them to PE bgp process.

Thank you rakesh for your explaination 🙂

As Alexander notes, the architecture here is to keep the ce specific routing protocols into the routing instance while the global or signaling protocols are in the root or base routing instance.

For a fuller description and base configuration examples have a look at chapter 12 in the Layer 3 VPN configuration guide.

http://www.juniper.net/techpubs/en_US/junos15.1/information-products/pathway-pages/config-guide-vpns/config-guide-vpns-layer-3.pdf