Layer two/three protocols?

dclarke

Hey Guys,

I'm new to this forum so I will start of by saying Hi everyone!

Well where do I begin I completed my CCENT last September and I am now moving on to my ICND2 exam to become CCNA qualified. Now I find most things pretty easy to pick up but this is just doing my head in! I know that Routing protocols are layer 3 and WAN protocols like PPP, HDLC and Frame Relay are layer 2 protocols but I what I don't understand is how Routing protocols like OSPF can form a neighbour relationship over a packet switched networks like frame relay?.. Now I guess my problem is seeing how they can form a relationship if the location they are reaching is say 15 hops away?

P.S. I am sincerely sorry if this question has been answered before, or if this seems like a stupid question.

SharkDiver

I don't think that 2 OSPF routers that are 15 hops away can become neighbors.

For both OSPF and EIGRP, it is a requirement that both router's interfaces be in the same subnet.

One exception to this is MetroEthernet, and that is because MetroE is a Layer 2 protocol.

Yeah, right after I typed this, I thought about the fact that Frame Relay is also a layer 2 protocol.

Ignore what I said.

Forsaken_GA

dclarke wrote: »

Well where do I begin I completed my CCENT last September and I am now moving on to my ICND2 exam to become CCNA qualified. Now I find most things pretty easy to pick up but this is just doing my head in! I know that Routing protocols are layer 3 and WAN protocols like PPP, HDLC and Frame Relay are layer 2 protocols but I what I don't understand is how Routing protocols like OSPF can form a neighbour relationship over a packet switched networks like frame relay?.. Now I guess my problem is seeing how they can form a relationship if the location they are reaching is say 15 hops away?

The frame relay network is transparent to the routers. They don't know what's in the cloud, and they don't care. The PVC that gets delivered over frame-relay appears to be a directly connected link. The same concept applies to MPLS. If I deliver a Layer 2 VPN circuit from Seattle to Atlanta, as far as the routers in Seatte and Atlanta are concerned, it's a direct layer 2 connection.

The frame relay cloud is essentially a big switch in the sky. If I put 15 switches between two routers, and trunked them all together so that their access ports were in the same vlan, they'd appear directly connected. Layer 2 hop != Layer 3 hop

networker050184

To add to what Foresaken laid out, the Layer 3 header is not touched while the packets traverse the frame relay cloud. Once the packet arrives at the distant router, the layer two header is stripped and the original L3 packet is processed by the router. I like to think of this in terms of actual bits to get a better understanding.

SharkDiver

Forsaken,

I am assuming that there are several ways to set up MPLS then?
Every one that I've set up, the CE router was only neighbors with the PE router.

Thanks!

networker050184

There are multiple ways of setting up an MPLS VPN. You have your standard L3VPN which is what you are familiar with and there are also a few varieties of L2VPNs that create a layer 2 connection across the MPLS network.

SharkDiver

Networker,

Thanks for the info.

Is it hard to get a vendor in California to give you an IP address in the same subnet as a different vendor gives you in New York?

networker050184

If you have a L2 VPN you can use what ever IPs you want.

Now getting an MPLS VPN to work across multiple carries is a whole different story. Plenty of providers have points set up with each other, but most likely you will end up with a L3VPN in this scenario. I haven't seen any carriers the do L2VPN with each other, but I haven't seen it all!

SharkDiver

OK, Thanks.
I have yet to see one like that. Our MPLS is always a point-to-point with a /30 mask between the customer's router and our router in the cloud.

Forsaken_GA

SharkDiver wrote: »

OK, Thanks.
I have yet to see one like that. Our MPLS is always a point-to-point with a /30 mask between the customer's router and our router in the cloud.

Right, it's the most common way to deliver private circuits these days. Delivering layer 2 circuits bring other issues with them (imagine a broadcast storm spanning your wan links.. not pretty). In my experience, they're only used when the customer has a specific need to maintain layer 2 adjacency across the WAN, ie for VMotion or High Availability in a VMWare cluster.