Umesh Kukreja, Atrica’s director of product marketing, said Atrica develops Ethernet technology to sell into carriers for their Metro aggregation layer, which it divides into four subsectors.

He said first are the Ethernet demarcation devices, the A-100 and A-200, which compete with boxes from companies like ADVA (the Covaro technology), Adtran, Metrobility (now part of Telco Systems), and Omnitron. These are CPE devices that are owned by the carrier, and used to demarcate the end of its network and the beginning of the customer’s. They also provide test and monitoring points on Ethernet tail circuits provided by other carriers.

Second are our Metro edge boxes: the A-2100, which is still a CPE device, the A-2140 in the carrier’s point of presence and the A-2160, which is an outdoor version of that box for wireless backhaul, said Kukreja. The competition here is Alcatel with the 7250 box from its TiMetra acquisition, as well as Cisco with the 2900 and 3750.

Next is the Metro aggregation layer per se, for which we offer the A-4100, with a 60Gb backplane, multiple GbE links in and 10Gb out, and finally we have our Metro core box, the A-8100, he said. This is a bigger version of the A-4100 with a 150Gb backplane. In both segments the competitors are the same: Cisco’s 7600 and 6500, Alcatel’s 7450, and some members of Extreme’s BlackDiamond range.

The new box, which will launch next month, will be for a different requirement: a corporate customer’s operations in another country to which a carrier might want to extend a WAN service. If it already has an MPLS PoP in that country, it is faced with taking an Ethernet tail from another carrier to reach the foreign subsidiary office there. The new box proposes a different approach whereby the other carrier provisions a Layer-2 tunnel across its network in a point-to-point topography. It will then bring the smarts of its Aspen service management platform to bear in order to deliver SLAs across that tunnel, back to the home country carrier’s network.

Because the network in the other country will probably not be running Atrica equipment, the new box will also have to guarantee the five key attributes of Carrier Ethernet: scalability, protection, hard SLAs, TDM integration, and service management.

Santa Clara, California-based Atrica is not a heavy hitter compared with Cisco or Alcatel, and it is not clear how many carriers will want to buy the CPE device. The attraction of such an approach, assuming the technology does what Atrica is promising, would be its obviation of the need to build out highly granular and expensive MPLS PoP networks in order to offer WAN services internationally. Even if a carrier does have some MPLS PoPs abroad, the tunnel could stretch a lot further across the country to reach one of them.

It is tempting to see Atrica’s move as another attempt to chip away at MPLS’s claim to be the natural way forward for delivering Carrier Ethernet services. Nortel, Siemens, and Extreme are all talking up the potential of Provider Backbone Transport for the aggregation layer, and presumably also for access, though their most high-profile supporter in the carrier world, the UK’s BT Group, is only looking at using it to connect its MSAN aggregation hubs back to the MPLS core of its next-generation network, 21CN.

In the MPLS camp, Alcatel has responded with T-MPLS, a proposal doing a subset of MPLS for aggregation and access, while another MPLS proponent, Cisco, dismisses the idea. It argues that normal, fully featured MPLS is fit for purpose in the outer reaches of carrier networks as is, using Pseudowire circuit emulation to deliver Ethernet services while retaining the control plane at Layer 3.

Kukreja was dismissive of both T-MPLS and PBT. While the first dumbs down MPLS to reduce cost, the latter dumbs down Ethernet to achieve determinism, he said. He said scalability is being delivered by the Provider Backbone Bridging on which it is built. Our service management enables us to pre-provision point-to-point and popint-to-multipoint links over standard MPLS, he said. Both [of the other approaches] will trip over when it comes to service management.