The MPLS-OPS Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] RE: IGP
This is a little misleading. When you are asking about IGPs, there is typically only one within the service provider network. Plain MPLS or MPLS VPN will work fine with distance vector protocols like EIGRP, but as mentioned a link state routing protocol is needed for MPLS traffic engineering, as each node needs the topology information those protocols provide and OSPF and IS-IS have the opaque LSA extensions necessary for traffic engineering. The question of running lots of IGP instances comes in to play when you are looking at supporting lots of customers on a PE, then lots of customer instances of THEIR IGP may exist on the PE. Roger is correct in that there is currently a 32 routing process limit in IOS and if the PE to CE protocol is OSPF, that is the limit. The majority of large scale commercial deployments use BGP between the CE and PE, where this 32 limit does not come in to play (through the use of address families). One of the concerns with running lots of OSPF instances between the PE and lots of CEs is that CEs that are under the control of the customer enterprise can generate an awful lot of LSAs and as there is no way really to contain the impact of LSA processing on the PE, it's not recommended for large scale deployment, or at least not as optimal. Chris At 08:03 AM 7/4/2002, Roger Clark Williams wrote: Zeevik has it right, but to add emphasis, OSPF and IS-IS are the only IGPs that support MPLS TE tunnels. Keep in mind, however, that Cisco IOS to my understanding supports only 32 routing processes per router, and OSPF must start a separate process each time it is used (unlike RIP and BGP). I am not sure right now whether IS-IS supports the use of address families within a single routing process (anyone?), but OSPF does not. Therefore, OSPF is good as an IGP in an MPLS cloud but has some limitations when used on a PE as both an IGP and a VPN routing protocol.
|
|