ATM vs Metro Ethernet

After the import of broadband access, there is a migration from ATM to Ethernet at the backbone network.
ATM provides a significant number of advantages, like it allows multiple services(voice, data) to coexist on the same network and mainly it offres a true QoS (CBR, VBR). So we have a reliable technology, but many large carriers transform their networks infrastructe from their legacy ATM swithes to the modern metro ethernet. This is also the main step for the next generation of ADSL, from ATM based DSLAM to IP based DSLAM.
Making a short technical survey about this migration, i found that the existing atm based DSLAM connect to the carrier's ATM aggregation network at speed of 155 Mbps (STM-1) or below (E3), via PPPoATM Tunnel(CPE-DSLAM-ATM-BRAS). Instead of the low-bandwidth ATM uplink, the new IP Dslam equipped with GE (about 6,4 times greater from STM-1 uplink) to the telcos metro ethernet aggregation network.
Also a very important parameter (but not technical) is the very low equipment cost between a ethernet and atm port.
But i think it is imature to ignore the greate success of ATM and just to adopt the Ethernet to the backbone. The ethernet is the leader at the local area networking but sure it not the same as the lagre of carriers metropolitan networks.
I have read a few technical papres about ip dlsam & metro ethernet for service provisioning triple play but i didnt realize which is the real big advantage for making the painful migration from ATM to metro Ethernet.
If somenone has a good picture of how is working the the new ip dslam with metro (vlan, 802.1q, qos) and the main advantages towards ATM technology, it will help us to understand the necessity of this big change.

Hi Koslyr,

Great question which in my experiance comes down to a couple of points.

Firstly cost. No way around it. Ethernet is cheap. ATM is expensive. And while ATM offers superior QoS mechanisms and therefore high quality Bandwidth this isn't enough when set off against the cheaper costs of ethernet, including cheaper equipment costs (CAPEX) and support costs (OPEX). Also consider the relative cost per MB of ethernet to ATM.

Secondly Services. It is easier to deploy broadband back haul combined with voice over ip and IPbased video services over ethernet then ATM. For example ATM & Multicast do not play nicely together. Ethernet being the natural partner to IP just works better for these IP based services.

Most telco operators are now seriously [palnning] migrating their legacy access networks from PDH to Ethernet. This means that IP & Ethernet awareness is needed within the DSLAM to interact with Ethernet. Telcos will not be putting in seperate ATM networks for Broadband only. In addition to this nearly all commercial ATM DSLAMs are being phased out over the next 18 months.

Try reading TR-059 from www.dslforum.com It explains the need for IP aware DSLAMs and the architecture required.

R

Hello Koslyr & Ranger,
I am a WAN Instructor for one of the Telcos and am in the thick of this. Optical Ethernet may indeed replace ATM but not just yet. We are into the second phase of our MOE deployment and do see our existing DSL service cut from our ATM network to our Ethernet network in time. The priority just now is the triple-play, of course. ATM as a backbone for DSL works for now so cutting all vanilla DSL off ATM is lower on the old TODO list. We upgraded our ATM to a Multi-protocol network and will develop that network on a different avenue, for certain to support our old Frame Relay network. Frame sales are stagnent, as you would guess, but we still have a rather large customer-base to service there. Alot will depend on interworking standards for VLANs. But enough of the crystal ball stuff! I have to get back to the article that drew me to this site about VTP. I am researching the use of extended range VLANs on a Cisco platform. Nice to meet you! MV

holy cow - does this topic classify as a 'basic concept' as the forum title suggests? If so, I'm in some SERIOUS need of some new study material!

Another important issue is that the growing demand for Ethernet at MAN network has led to a rebirth of SDH. More specific the Next Generation SDH (NG SDH) is the primary solution for transmitting Ethernet traffic over SDH. The basis technologies of NG-SDH are the Generic Framing Procedure (GFP), Virtual Concatenation (VCAT) and Link Capacity Adjustment Scheme (LCAS). So there is the next generation of ADM multiplexers which are supporting GFP, VCAT and LCAS and are known as Multiservice Provisioning Platforms nodes(MSPP).
One of the main challenge of legacy SDH is the Ethernet-based traffic is burtsy and periodic. This is a serious problem as the SDH is a voice-optimized TDM technology and allocates a fixed bandwidth between a pair of endpoints. This is inefficient of SDH when handling Ethernet traffic, where there is a bandwidth on demand.
But I think the same problem is also appears at ATM, so my question is if the adoption of Ethernet Over SDH will go to inherit the same basis as ATM over SDH in order to address the problem of TDM multiplexing( fixed bandwidth vs bandwidth on demand).