Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

Frame relay DLCI number and mac address

Hi

What is wrong with mac address and mac-ip mapping so in frame relay another numbering system(DLCI) is designed?

Is it about the flexiblity that is provided to ISPs? (because mac addreses are constant)

3 REPLIES
Cisco Employee

Frame relay DLCI number and mac address

Hello,

There is nothing wrong with MAC addresses, it's just the Frame Relay is designed around a different approach - assigning the frame a label (DLCI in Frame Relay) that indicates to the immediate next device how the frame should be forwarded further. In this case, the label does not indicate the identity of the recipient but rather only the next hop process. Frame Relay is not the only technology in this aspect: the ATM was another technology built on the identical principle, using the Virtual Path Identifier/Virtual Circuit Identifier (VPI/VCI) as these labels, and also MPLS uses the very same paradigm.

Think of a Frame Relay switch as a huge switching board where each interface has a number of different wires separated by their color (DLCI), and an incoming wire of a particular color on one interface can be connected to a wire of a different color at the outgoing interface. Obviously, this approach works, with the wire colors (DLCIs) being locally significant only, i.e. having unique meaning only between two physically adjacent switches. This allows you to conserve the DLCI space (you can recycle DLCI values on a per-interface basis), and still it allows you to build a switched path through your FR network that is capable of delivering frames between the endpoints. Obviously, you do not need the concept of the globally unique MAC address here.

So once again: Ethernet and Frame Relay come from very different backgrounds and use very different approaches to achieve the same result. This is very common in networking and should not be considered as one technology being inferior to other. It is simply a different take on the same issue, having its particular advantages and disadvantages. There is no such thing as a perfect technology - but I'm sure you know that very well yourself.

Best regards,

Peter

New Member

Frame relay DLCI number and mac address

Thank you very much Peter Paluch

 assigning the frame a label (DLCI in Frame Relay) that indicates to the  immediate next device how the frame should be forwarded further

Also, IP routing mechanism was designed to show next device. But here IP numbering or routing system is not used instead the number that ISP that was given by ISP is used. Because IP routing selects the best path but ISPs don't have to obey this rule. So DLCI numbering is for giving flexibility to ISPs.

Is that right?

Cisco Employee

Frame relay DLCI number and mac address

Hello,

You are welcome

Also, IP routing mechanism was designed to show next device. But here IP  numbering or routing system is not used instead the number that ISP  that was given by ISP is used. Because IP routing selects the best path  but ISPs don't have to obey this rule. So DLCI numbering is for giving  flexibility to ISPs.

Not entirely.

You cannot really compare IP to Frame Relay because they work on different layers. IP is a Layer3 protocol meaning that it is concerned with addressing end hosts and finding a way between any pair of end hosts through a series of intermediary nodes (routers). This is the same for any Layer3 protocol, be it IP, IPv6, IPX, CLNP or others. Logically, if the path between two end hosts goes through a particular sequence of routers, each and every Layer3 protocol must eventually solve the task of selecting the appropriate next hop for the packet to reach its destination host. That is not specific for IP only but for all Layer3 protocols.

However, what is important here is that neither of these Layer3 protocols deals with a more basic issue: once the next hop is determined, how shall the packet be carried from the current node to this next hop? Sure, they are directly connected - that is the point of the next hop - but still, how this packet should be transferred to this next hop in particular? How do we make sure that it is not carried to a different next hop?

This is the purpose of the Layer2 where both Ethernet and Frame Relay belong - to provide the means for directly neighboring hosts to communicate. By directly neighboring hosts, you should understand not only two PCs connected with a crossover cable but rather any constellation in which stations talk to each directly without being aware that there is some network infrastructure between them. In a single Ethernet network, stations do not know and do not care how many switches are between them, nor do they need to know it: in fact, they are not aware of these switches at all. And yet, there may be several switches that need to be traversed before the frame of one station hits the next intended station. The same is valid for Frame Relay switches: they also interconnect a number of devices, and these devices do not neither know nor care how many switches are there - the important thing is that once a frame is sent, it will be delivered through the FR cloud to the exit of the virtual circuit.

Regardless of the Layer2 technology, Ethernet, Frame Relay, IP always considers the communication over this technology a single hop. So your comment about DLCI being able to select worse-path routes is not correct from the IP point of view: IP does not see the possible paths inside a Frame Relay cloud at all, just as it does not see multiple paths in a redundant Ethernet switched network. For IP, it's just a single hop and it does not even care how that hop works.

So once again - do not compare IP and Frame Relay, just as you do not compare IP and Ethernet. Frame Relay is a packet switching technology that uses locally assigned identifiers instead of globally unique addressing. These identifiers do not reflect any shortest paths and instead, they are constructed according to the SP's wishes. By having complete control over creation of these identifiers, SPs can indeed engineer paths as they like. But it would be incorrect to say that IP was replaced with DLCIs. In reality, IP is carried inside Frame Relay frames(because it can't be carried just so, without encapsulating it into frames of some kind), and Frame Relay happens to be a technology using switching based on DLCIs.

Best regards,

Peter

1306
Views
15
Helpful
3
Replies