Is Asynchronous Transfer Mode a wide area network? Or is it just a framing protocol?

What I mean by wide area network is that, if there is such a thing called "ATM provider", does it mean that they deploy multiple ATM switches spread over diffferent geographic points and these switches are interconnected?

And if I am going to avail their WAN service, for example, I want to connect my two sites at City A and City B, are they going to connect my router at City A to their nearest ATM switch in City A and connect my router at City B to their nearest ATM switch in City B?

Now if the short answer is yes, I have a follow up question. How does the ATM provider interconnect these switches? Are they using SONET/SDH?

If this is also yes, then can we say the following:

  • End Users use the service of ISP
  • ISP use the service of ATM Provider
  • ATM Provider use the service of SONET/SDH Provider?
  • ATM is a protocol. Protocols can't be wide-area networks because they are protocols, not networks. Only networks can be wide-area networks. Commented Feb 15, 2021 at 14:54
  • Did any answer help you? if so, you should accept the answer so that the question does not keep popping up forever, looking for an answer. Alternatively, you could post and accept your own answer.
    – Ron Maupin
    Commented Dec 23, 2021 at 16:32

1 Answer 1


ATM is a layer-2 protocol. Yes, it is (was?) mostly used in WAN deployments. However, there were PCI network cards to add ATM links to a PC/server. (3com 3c975, comes to mind)

When you purchase a point-to-point link between sites, the provider will set it up however they see fit. (It may not even be on their infrastructure.) In general, yes, each site will be connected to the nearest local POP, but that's not always the case. How the SP links their POPs is totally up to them, and as a customer isn't anything you have any control over. How the SP gets traffic between your sites isn't your problem, that's why you're paying them. (assuming they meet the term negotiated in your contract)

The example I like to make is somewhat dated, but still valid. Take the case of a point-to-point T1. It hasn't been a true p-t-p circuit for decades. In fact, it isn't even a T1 for more than a few feet. Between the smartjack and CPE, it's a 4-wire T1. From the smartjack to the CO, it's an HDSL circuit. Between CO's it's anyone's guess -- it's whatever circuit emulation service the carrier is using, over whatever infrastructure they have. From the customer's perspective, they have a T1 between sites, how the SP achieves this isn't their problem.

  • There are L1 components as well, as required for a NIC - but they're long obsolete.
    – Zac67
    Commented Feb 12, 2021 at 7:21
  • I don't recall what the layer-1 was for those cards; they were RJ45. Everything in actual use... T1, T3, OC3/12/48. (and HSSI to a T3 CSU/DSU because the switch was too far away for HSSI-HSSI)
    – Ricky
    Commented Feb 12, 2021 at 7:27
  • Well, those NICs had to use some layer 1. I remember something about UNI and NNI. It was little more than a serial line with differential signaling but nonetheless a physical layer.
    – Zac67
    Commented Feb 12, 2021 at 7:46
  • Yes, we experimented with the desktop ATM LAN interfaces because it offered 25 Mbps vs. the 10 Mbps ethernet, and both were pretty expensive, but then ethernet was easier and the prices started dropping for ethernet, but not ATM.
    – Ron Maupin
    Commented Feb 12, 2021 at 13:38
  • Fun fact: my employer makes a device that uses ATM on the backplane of a device with modular interface cards. The maximum distance an ATM cell can ever travel is 19", from one end of the frame to the other. Commented Feb 13, 2021 at 11:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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