Why OSPF is not used in MPLS, though it is very efficient protocol

[get_shanu]

Dear all, though OSPF is a very efficient protocol, with a number of features to handle much larger networks,it is not used in MPLS. Can u plz let me know, what special features BGP posses over OSPF, for its selection in MPLS.

[mheusinger]

Hello,

OSPF is not as scalable as BGP is. BGP is for sure more flexible and uses less resources (especially CPU). There are much more options for prefix "marking" (communities), filtering and path selection in BGP than in OSPF.

In short: BGP is a prooven ISP routing protocol scaling to hundred thousands of prefixes. OSPF is more used as IGP for fast convergence and scalability to hundred thousand routes is at least questionable.

Hope this helps! Please rate all posts.

Regards, Martin

[vettrock]

BGP is the defacto internet routing protocol. If you are a ISP, you pretty much have to run BGP to communicate with the rest of the internet. If you are running MPLS in your network, you need a link state IGP, which means you have to run IS-IS or OSPF. BGP and OSPF have different functions, one is an External routing protocol, and the other is an internal routing protocol.

[vladrac-ccna]

Besides you can do QPPB, qos policy propagation over BGP, which its a pretty handy QOS feature.

vlad

[sean]

For MPLS, BGP is only required on the PE routers. On your P routers, you do not have to have BGP running.

[Gary Tong]

BGP is the most scalable routing protocol with abundant attributes.

[milovanov]

In order to run MPLS in your core you need routing conectivity and also a routing protocol. You can use there OSPF but my recomendation is to use ISIS. In order to have also MPLS VPN you need to run MP BGP between the loopbacks of you PE routers. In direction PE->CE you can also have what ever routing protocol you want (also OSPF). If you have OSPF in the core and also OSPF in direction to the customer you need separate OSPF routing processes.

Hope that this helps :))

[olorunloba]

For core MPLS switching, any routing protocol can be used, including static. Service providers have preferred to use link state protocol like OSPF and ISIS and they still do.

However, different MPLS applications have their routing protocol requirements. MPLS Traffic Engineering for example requires a link state protocol. Hence, if you want to do Traffic Engineering, your IGP must be either OSPF or ISIS. This is because for proper operation of MPLS TE, the routers should have a complete topology view of the network, so as to be able to route packet accordingly.

For MPLS VPN, there is a requirement to deploy iBGP on the edge routers. The main requirement stems from label allocation procedure of MPLS. While MPLS routers assigns a label for every route in the FIB, routes learned via BGP are exempted. Rather, for BGP learnt routes, the label for the BGP next-hop of the route is used. This is critical to the operation of the MPLS VPN, because it effectively creates a tunnel and ensures that every router on the network does not have to carry all the VPN routes.

I hope the explanation is clear

[deepak.vyas.ccip]

If u are aware of MPLS u must be knowing about Route Distinguisher i.e. 64 bits value appended to the IP to creat a unique 96 bit prefix. All the routing protocols understand only 32 bit IP, but MP-BGP, an extension of BGP, understands this 96 bit prefix. Hence to use MP-BGP we have to use BGP on the PE routers.

I hope this clears the doubts, but if u do have some thing which leads u to confusion u r free to ask me on cisco forums and do mark me a mail so that i am aware of the updates.

Deepak Vyas

VSNL

deepak.vyas.ccip@gmail.com