Frame relay, ATM, T(or E) and Ethernet; where do they stand beside each other?

junii

In my book here on the network + and ccna it makes such points as “large multinationals often use connections such as T3 to connect regional offices” and “lower level ISPs often use T3’s to connect to upper tier ISP’s”
Are T3’s best for this because they run over infrastructure like eircoms (POTS), hence long distance? Is a T or E connection the quickest available over the POTS?

Regarding frame relay and atm, do these just encapsulate TCP/IP packets while it’s in the network and why use frame relay or atm rather than Ethernet the whole way through?
Perhaps this is happening now with 10G, is it?

Thanks

rogue-entity

T1, E1 etc are Carrier Trunk lines used by telecoms to link exchanges, local and foreign. America uses T carriers, and Europe uses E carriers. Companies that require large capacity internet access or many telephone lines will often have these types of connections since a single E1 line can provide 32 phone lines, or a single 2Meg up, 2Meg down internet connection.
Wiki Link

The fastest internet connection possible over a single POTS line is AFAIK 24Mbit DSL.

Dont know much about ATM or Frame Relay, but you could take a gawk at Wikipedia.

steve-hosting36

Modern fiber based networks are using ethernet the whole way through (layer2). Very few if any 'modern' carriers, isps or providers would use atm or frame relay.

telecomsman

ATM, which is a cell-switched protocol, was historically used by Telecoms providers (primarily over SONET) due to is proficiency at voice transfer.....do a WiKi for full details as its to early for me to go through it. As new technologies came along that provided higher speeds than that on ATM while also going a long way to solving the voice latency issues of before, ATM was soon replaced in new service deployments.

Stephen

I finished college 2 years ago, but at the time they still had a 24Mbit ATM link to HEANet. I wonder if that's still the case.

I understand ATM never really took off because it was so prohibitively expensive with little extra benefits compared to the alternatives.

degsie

In terms of using using current infrastructure some providers are providing ethernet over their existing SDH network. SDH is a transport technology which traditionally carried E/T tributaries and the likes of ATM & frame relay.
Ethernet over SDH is proving to be a popular choice with the larger providers and is cost effective with their customers.

junii

What is the actual difference between 2Mbit DSL and an E1 (2.048Mbps)? Both are digital and can run over two pairs of copper.

Ok so things are all going towards Ethernet; fair enough. Is this because Ethernet is better or cheaper?

I still need to understand how ATM etc works. I have been doing a fair bit of reading around the place but I still have some remaining unanswered questions.

Just as an example: Lets say I’m working for a small company with a DSL connection providing internet access for the one Ethernet segment we have. Imagine I want to access a webserver in America which is running in a data centre with Ethernet infrastructure. This data centre is also connected to its ISP through a 10G Ethernet fibre connection.
Now, lets say I sent a http request for the index page on that American webserver and it goes into my ISP as a TCP/IP packet. My ISP is connected to its upper tier ISP with ATM and that ISP connects into another ATM backbone. Now how does ATM know where to set-up the PVC (permanent virtual connection…ATM being the circuit switching technology) to if there is no direct ATM link to the American data centre. After all, hasn’t it only got the IP address of the webserver to work with.

Stephen

FFS, i had a big long reply written here but boards ate it. Anyways, the important difference between an E1 and a 2Mb DSL connection is that an E1 will be uncontended. You also may want to link two geographically seperate sites with an E1, rather than just connect them to the internet. Also, DSL is typically asymmetrical unless you have something like SDSL, an E1 is always symmetric.

That said, some (many? all?) telcos provide lower bandwidth WAN links (say 2Mb MPLS and the likes) over DSL a lot now, but again this isn't via the internet.

ATM is a cell switching technology, as opposed to circuit switching. ATM provides very circuit switched-like service but its not really a pvc. Cells are like tiny fixed-size packets.

Have a look at this page: http://www.cisco.com/univercd/cc/td/doc/cisintwk/ito_doc/atm.htm

rogue-entity

junii wrote:

What is the actual difference between 2Mbit DSL and an E1 (2.048Mbps)? Both are digital and can run over two pairs of copper.

Ok so things are all going towards Ethernet; fair enough. Is this because Ethernet is better or cheaper?

I still need to understand how ATM etc works. I have been doing a fair bit of reading around the place but I still have some remaining unanswered questions.

Just as an example: Lets say I’m working for a small company with a DSL connection providing internet access for the one Ethernet segment we have. Imagine I want to access a webserver in America which is running in a data centre with Ethernet infrastructure. This data centre is also connected to its ISP through a 10G Ethernet fibre connection.
Now, lets say I sent a http request for the index page on that American webserver and it goes into my ISP as a TCP/IP packet. My ISP is connected to its upper tier ISP with ATM and that ISP connects into another ATM backbone. Now how does ATM know where to set-up the PVC (permanent virtual connection…ATM being the circuit switching technology) to if there is no direct ATM link to the American data centre. After all, hasn’t it only got the IP address of the webserver to work with.

E1 will firstly provide a symmetric link, 2Mbit up, 2Mbit down, and as Steven said, it is uncontended, you get all 2Mbits of that bandwidth to yourself, DSL 24Mbit will have to share that bandwidth with everyone on the same DSLAM (in the exchange). And its not symmetric, it will probably be 24Meg down and 8Meg up.

Wikipedia will give you a good understanding of ATM, Ethernet protocols, Ethernet's physical standards amonst other things.

No direct circuit would be established between your ISP's upstream ISP and the ISP serving the datacenter, seperate connections would be established internally between routers, and these routers would bridge the ATM links with regular packet-switched ones I suspect, so TCP/IP would still be used on the upper layers.

telecomsman

junii wrote:

I still need to understand how ATM etc works. I have been doing a fair bit of reading around the place but I still have some remaining unanswered questions.

Just as an example: Lets say I’m working for a small company with a DSL connection providing internet access for the one Ethernet segment we have. Imagine I want to access a webserver in America which is running in a data centre with Ethernet infrastructure. This data centre is also connected to its ISP through a 10G Ethernet fibre connection.
Now, lets say I sent a http request for the index page on that American webserver and it goes into my ISP as a TCP/IP packet. My ISP is connected to its upper tier ISP with ATM and that ISP connects into another ATM backbone. Now how does ATM know where to set-up the PVC (permanent virtual connection…ATM being the circuit switching technology) to if there is no direct ATM link to the American data centre. After all, hasn’t it only got the IP address of the webserver to work with.

As RogueEntity said: if your ISP is connected to its ISP via ATM both would have an ATM router (or something like it) on either end of the link. This router would have an ATM connection for terminating the incoming ISP and an Ethernet connection for the packets to route on into its network and ultimately to the American ISP and then the data centre. This is much simplified so apologies if it comes across as condescending.

What you need to understand is that when you send the HTTP request from your Web Browser, as it goes down the different OSI layers it is encapsulated within the different protocols. Think of it as your HTTP request as an letter and each protocol puts the request into its own envelope, your first envelope will have the IP address on the outside as the destination. Eventually what you have is a large datagram/frame containing your initial request within a number of envelopes. When the frame reaches the next point along its trip,ATM router in this case, each of the envelopes is taken off until you come to your initial IP packet. The router will then send the packet out the relevant route to the next hop on the journey. The packet will then be put into the envelopes again as it passes down the OSI stack again.

Basically what Im trying to explain is that, IP addressing only matters when you are in an IP world (OSI Layer 3).....ATM lives in a Layer 2 world.

ZygOte

Stephen wrote:

I finished college 2 years ago, but at the time they still had a 24Mbit ATM link to HEANet. I wonder if that's still the case.

I understand ATM never really took off because it was so prohibitively expensive with little extra benefits compared to the alternatives.

xDSL is run over ATM Circuits. afaik

BTW: this is quite nice, i do like the heanet

http://www.hea.net/