Regarding the frame relay technology, could you confirm to me that traffic using permanent virtual circuit use the same ISP routers in the path for all packets sent ? I'm a little bit confused.

Frame relay is considered as packet switching technology, but it seems to have the circuit-switching characteristics: Circuit Switching vs Packet Switching (Source: www.apposite-tech.com)

Could we say that the virtual packets switching technologies (like frame relay), are near the circuit switching technologies ?

Also regarding an additional point regarding frame relay. I know that the customers are identified with the DLCI identifier, locally significant, but then in the ISP infrastructure, how do the customer A traffic and customer B traffic are identified ?

  • 1
    Frame relay uses frame switches, not routers. The frames are switched at the data-link layer (the DL in DLCI stands for Data Link) based on the DLCI, not routed by the addresses on the network packets. – Ron Maupin May 21 '18 at 10:49
  • Ok, but DLCI are from 16 to 1007. So an ISP could have only a maximum of about 991 DLCI so only 991 customers ? (maybe less if the customers have multiple DLCI) It seemed to me it's a small range. It's like we have an addressing model based on only 1000 addresses. – bdes31 May 21 '18 at 11:15
  • 1
    As you said, DLCIs are only locally significant, and the DLCI numbers actually change during the transit from one endpoint to another. DLCI 100 for one frame switch is not the same as DLCI 100 for another frame switch. – Ron Maupin May 21 '18 at 11:18
  • I'm not sure to understand how the ISP could identify in the whole path the frame of company A, or company B, if the DLCI change all along the path. – bdes31 May 21 '18 at 11:46
  • 2
    The frame switches maintain tables, and they switch the DLCI numbers on the frames as needed, per the tables, to reach the destination. The tables will hold the incoming DLCI, the next interface, and the outgoing DLCI. The switch will change the DLCI accordingly. The carrier sets that up ahead of time. (Of course, none of the carriers offer frame switching anymore, but many still offer frame relay as a layer-2 option on a point-to-point link.) – Ron Maupin May 21 '18 at 12:04

With a circuit switched network, the two endpoints are connected with a dedicated circuit, meaning that for the duration of the connection, no other endpoints can use the circuit.

With frame relay, the links between the frame switches are not dedicated to any one connection, but can be used by multiple connections that may need to send traffic between two switches. The circuit is virtual (the V in PVC and SVC stands for virtual) because it looks like a circuit to the endpoints, but the physical path can be shared among many endpoints.

Contrast that with a true circuit-switched network, like standard POTS, where a circuit is set up directly from one endpoint to the other endpoint, and the entire physical path is dedicated to the two endpoints for the duration of the connection. The POTS fast busy signal tells you that all the circuits are busy, so your call cannot be completed, but that doesn't happen with virtual circuits like frame relay because the virtual circuits can use links that other virtual circuits are using.

A link between two frame switches can carry traffic for multiple virtual circuits, but you would need as many links between the switches are there are connections for real circuit switching, otherwise you could end up like POTS, where all the circuits could be busy, and you could not get a connection.

  • Thanks for your reply. You mentioned "because it looks like a circuit to the endpoints, but the physical path can be shared among many endpoints.", but does a physical path is the same for all packets for a PVC ? If I have well understood, it's the case. But regarding the SVC, it could be a different path for each new circuit ? And finally, regarding SVC, how does frame relay knows that the "session" is finished, and the virtual circuit could be closed ? – bdes31 May 21 '18 at 11:54
  • 1
    Parts of the physical path taken by a PVC can be used by other PVCs, but that is not the case with a real circuit. Many PVCs could use a single link between two switches, even though the entire path may not be the same, but you cannot do that with real circuits because each circuit is dedicated to the connection, end-to-end. You are getting far afield of the original question, and you should really ask in other questions. – Ron Maupin May 21 '18 at 12:08
  • 1
    @bdes31, this is a Q&A forum, not a discussion forum. We have Network Engineering Chat for discussions. You should ask specific questions to get specific answers here. If you want to discuss something, even if it is off-topic here, you can discuss it on Network Engineering Chat, otherwise you can ask multiple questions to get multiple answers. – Ron Maupin May 21 '18 at 12:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.