Frame Relay and it's Relation to Routing

Oh man! Ad-blocking software has been detected! :'(

This website is run by the community, for the community... and it needs advertisements in order to keep running. Blocking our ads means your killing our stats!
Please disable your ad-block, or become a premium member to hide all advertisements and this notice.

Something that I'm having a little trouble getting my head around with is the nature or the relationship between frame relay and routing protocols.

I've been able to succesfully configure and raise point-to-point frame relay interfaces using Boson software, but from there I'm not sure on the best way to pass dynamic routing over the top of it, and haven't been able to find any documentation that tells me exactly what I'm trying to figure out.

I know that frame relay interfaces don't by default broadcast routing announcements. I also know that static maps can be set up using the "frame-relay map ip" comand, but I'm not sure how you'd go abuot ensuring that routing protocols are enabled by default. I did find the command "ip ospf network broadcast" which could be run from an interface, but to my mind it would be logical for there to be something that was interface specific rather than protocol specific with regards to this. This command didn't work on the Boson software, but I'm aware of it having some bugs anyway so this isn't necessarily conclusive.

I'm also a little confused about the difference between using the above command and setting up a static route. The above command allows you to specify a destination IP address and local DLCI to be broadcasted (out the interface presumably). I'm not sure why this is used to specify a single IP and not a network address for that matter. Obviously it's not being used for routing purposes as I'm visualising it (i.e. layer 3).

You could represumably set up a static route specifying the remote network and the next hop interface but this is obviously problematic if you don't want to rely on static routes.

I know I'm not quite visualising the relationship correctly, despite reading over both the McGraw Hill book and the Sybex book, and various other bits of documentation. Would anybody mind setting me right a little?

Cheers.

Alex.

 

Oh man! Ad-blocking software has been detected! :'(

This website is run by the community, for the community... and it needs advertisements in order to keep running. Blocking our ads means your killing our stats!
Please disable your ad-block, or become a premium member to hide all advertisements and this notice.

The frame relay map command is intended to show the connection between a dlci and the ip address of the far router. i.e. a single machine; it is not a routing command. Of course, because it is a direct connection, the router will add the route to that endpoint in its routing table. But it won't add anything that end-point itself is connected to, unless you either set up a static route or use a routing protocol that is transmitted over the link. If it is a point to point link, there is usually no problem with enabling a routing protocol over the link.

The problem is with point to multipoint links, and concerns avoiding routing loops. The usual 'split horizon' rule means that when one of the links reports on its route, it will not be advertised back out of the same interface, so the other end-points will never see it, and so the routing table will never be properly set up for the whole network. But if you disable split horizon you can run into looping and 'count to infinity' problems.

The usual way of dealing with this is, as you tried, using subinterfaces and disabling split horizon - but this doesn't work with some protocols, which won't accept subinterfaces, but treat the link as a single one, and therefore disallow split horizon from being disabled for it.

Where it does work, you have to specify that the link accepts routing broadcast messages by including the option 'broadcast' at the end of the frame relay map command.

Things are more complex for OSPF, try here.

 

Thanks very much for taking the time to reply.

Right, I've since experimented a little more and found that indeed, using interfaces rather than sub-interfaces there's no problem. I was thinking that the whole idea of using subinterfaces was to set up a series of individual point-to-point links, much in the way VLANs trunk ports attached to routers work. Perhaps that's where I'm going wrong in my thinking?

So the purpose of using "frame-relay map ip" is more to speed up the convergence process, and assist with OSPF configuration than having anything to do with the configuration of later 3 routing? I suspect I've been associating the "broadcast" parameter with broadcasting routing table broadcasts, wrongly? Or have I?

The documentation I have seems to be a little contradictory. If the broadcast does refer to layer 3 protocols, it would seem that the broadcast parameter refers to the interface's inclination to listen for routing broadcasts, not sent them out.

I still don't quite understand why the layer 3 address needs to be mapped to the DLCI this being the case (my thinking being that as I understand it the router thinks you've basically got a point to point link, with the frame relay cloud handling routing using a combination of physical ports and DLCI addresses rather than layer 3 - as long as the routing protocol knows which interface to pass things out through). Cisco documentation states that Frame Relay is a purely a layer 2 protocol (unlike X.25 for example) so I'm not quite sure where this fits into the scheme of things, as I've been able to successfully set up a static route using the sub-interface (s0/0.xxx) as the next hop.

--

So, my question essentially remains, how do I enable dynamic layer 3 routing? Is it simply a case of using a compatible routing protocol and specifying the endpoint and broadcast options for the sub-interface. Should I specify the sub-interface as a point-to-point link or multipoint (assuming I've got a simple 3 router, fully meshed topology, which I do, for the sake of argument)? I have been specifying each sub-interface as point-to-point, is this incorrect?

Thanks in advance for any further help anybody can give, I feel like I'm right on the brink of understanding this, and it's driving me mad.

 

'Broadcast' allows the passage of routing updates over the link, and that is what it is for.

When using subinterfaces, use point to point links if what you want is a spoke/wheel configuration, with all traffic going via the central point. With multipoint you can get communication between points on the 'wheel' that don't go via the central router.

With a multi-linked topology you have to have some way of mapping the L2 to the L3 addresses: in ethernet this is done with ARP, with frame relay it is done with 'inverse arp' - but if you disable inverse arp, you can replace it with a static mapping of the L2 and L3 addresses for that link by using the 'map' command.

You don't need inverse arp or a map command with single point to point, or subinterface point to point links, but use the frame relay interface dlci command to address the link.

But with multipoint links on subinterfaces, you do need either inverse arp or the map command so that packets know where to go.

Try herefor a better explanation than mine.