cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
882
Views
0
Helpful
13
Replies

Dumb questions about ATM interface :)

OK, since i am fairly new to networking, here is my questions...

If i had a router, and the one interface was pointing to the ATM interface, what would be considered the ATM? Would it be the switch from the service provider or what?

1 Accepted Solution

Accepted Solutions

If you have the option between an ATM and ethernet uplink for a simple CPE, the only reason to choose ATM is if you are supporting something fancy like interleaved VBR priority traffic, VoATM, or multiple VCs to different edge routers or devices. Odds are you don't need to do this, so connect to the ethernet port.

If you have no CPE and must choose between ATM and frame or IP T1 lines, then you might want to consider ATM for the low latency cell-based multiplexing to scale past 1xT1, but only if you can find the IMA interfaces cheap and the provider actually supports IMA.

ATM would have been better accepted if the interfaces had not been so expensive (it has an awesome feature set), but these days it is a more or less a dead technology outside the provider backbone where there are people who actually value its features enough to learn how to use it.

View solution in original post

13 Replies 13

marikakis
Level 7
Level 7

Lots of entities create "the ATM" :-) I will give you an example:

--------

We have 2 routers on the left and right. Routers have ATM interfaces. They connect to an ATM cloud that has ATM switches via physical links. This is typical ATM setup. Routers at endpoints typically put IP packets into ATM cells and recover IP packets from cells. Within the ATM cloud generally only cells travel. ATM is generally a layer 2 technology.

ok so what is the difference between these....

1. Ethernet interface for PPPoE

Vs.

2. ATM interface for PPPoE

Can anyone help?????

1. An Ethernet interface is used if the router connects to a DSL modem.

2. an ATM interface is used if the router connects directly to the DSL network.

Can someone give me an example of #2????

HELP!!!!!!

Hello,

Please be patient. Answers in this forum are posted by volunteers who get as a reward some points in the rating system, and in many cases not even that. People will post answers as soon as they have time. I am working on giving you an answer right now.

Kind Regards,

M.

I am sorry, I didnt mean anything by that. I appreciate any help you can give me.

Hello,

People get interest in ATM usually because they work in provider environments, because they are taking an exam, but by far the most common reason is DSL setups. I should have figured you were going to talk about DSL eventually :-)

A PC connects typically via ethernet port to some DSL CPE's ethernet port. CPE connects to telephone line. You have a splitter there that separates phone calls (low frequencies) from data (higher frequencies). The DSL physical layer is the plain old phone cable. To speed up DSL deployments (because telcos had some ATM clouds in place), ATM was specified as the layer 2 above DSL. So, DSL CPE connects via ATM/DSL and through the telco cables the DSL connections reach a device called a DSLAM. DSLAM terminates the DSL physical connection. You can think this as a cable between your DSL CPE and the DSLAM DSL ports.

The early DSLAMs had an ATM uplink (e.g. STM-1 or 155Mbps speed) that would collect all ATM/DSL user traffic from all its DSL ports and forward it further inside the cloud. Later, to support higher collective user bandwidth, the uplink typically moves to GigabitEthernet.

This will become a long story, so I will give you something to keep you busy, until I have the time to complete the picture :-)

Kind Regards,

M.

We were left to the point were we managed to terminate DSL physical connection to the DSLAM. But, we need to terminate the connection logically as well. This typically means use of some PPP flavor. PPP was doing that in earlier dial-up scenarios. PPP is used so that the DSL client can get an IP address and other parameters from a provider that can be the same as the provider that terminates physical layer or it can be some other provider to whom the telco forwards clients.

Read those 2 paragraphs as well (they are quite good):

http://en.wikipedia.org/wiki/Point-to-Point_Protocol_over_Ethernet#Why_PPPoE

http://en.wikipedia.org/wiki/Point-to-Point_Protocol_over_Ethernet#How_PPPoE_fits_in

I have to go now. If you still have questions people will post as soon as they can.

I learn quite a bit form your posts, thanks and keep it up. Later

Since I am not in a provider environment and just getting a T1 DSL line, i should use the Ethernet Interface for PPPoE, rather than ATM interface?

Also, rather than getting a router to connect to the DSL modem, could I use a 3750 and just enable IP routing?

If you have the option between an ATM and ethernet uplink for a simple CPE, the only reason to choose ATM is if you are supporting something fancy like interleaved VBR priority traffic, VoATM, or multiple VCs to different edge routers or devices. Odds are you don't need to do this, so connect to the ethernet port.

If you have no CPE and must choose between ATM and frame or IP T1 lines, then you might want to consider ATM for the low latency cell-based multiplexing to scale past 1xT1, but only if you can find the IMA interfaces cheap and the provider actually supports IMA.

ATM would have been better accepted if the interfaces had not been so expensive (it has an awesome feature set), but these days it is a more or less a dead technology outside the provider backbone where there are people who actually value its features enough to learn how to use it.

First of all, I am glad that Brian resolved your issue. ATM does have features and interfaces can be expensive as Brian said. It can also be expensive in terms of bandwidth usage. Various types of overhead can cause up to 35% or more of the link to be just overhead depending on how small the IP packets transferred are.

One thing I find cool about using ATM switches in a network (that is usually not mentioned) is the fault isolation capabilities those switches bring to the network. We typically consider the case of normal operation when comparing technologies, but failure situations are also important. Downtime costs, so the better fault isolation capabilities you have, the faster issues are resolved and you reduce downtime cost.

Imagine this scenario (which is typical provider setup):

--------

The telco line connects to the switches instead of the routers. If you see in one of the routers the ATM subinterface going down and your routing protocol going down, you do not know what happened: is it this router's fault? is it the remote router? is it some ATM switch? is it the line from telco? This problem is typical with router endpoints (and endpoints in general). You just do not know what happened and try to explore all possibilities. If you log in into the ATM switch, the switch will tell you if its the line. The line is between switches and you see alarm on telco line ports instead of the router ports. So, you know right away what you have to do: Pick up the phone and call the telco. Telco line failure is a very common type of failure, so it pays to diagnose it quickly and say to the telco: we know its your fault and not ours, so stop blaming our routers and fix this right away.

ATM switches are also primarily L2 devices. So, if you see error in switch port that connects to router, you check that your router is up and router side cables are good, before blaming L3 routing protocols for the failure. You don't ping and trace hopelessly. Switch gives you the answer.

If you are interested in the DSL end-to-end setup I will keep posting. It will be a good exercise for my memory. Only problem is that my professor says I have to concentrate more on other things now. As long as you respond, I will try to answer your questions, but I might be slow. Also note that knowledge of networking history helps to understand the reason for various alternative setups not only in DSL, but in general, so keep this in mind for every technology that puzzles you in the future. History is also easier to read than various configurations with mysterious commands that don't explain the why's.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: