MPLS

Nortel cut back Carrier Ethernet; focuses on optical

Nortel will be cutting back on its Carrier Ethernet investment and focusing more on optical, according to a leaked memo later largely confirmed by the company. The move is an acknowledgement of the company’s Ethernet woes, which buried its early hopes for the success of its PBT technology.

The problem we found with Nortel’s positioning was that the company was unable to make a strong case for PBT in a unified national network like BT’s, and in many metro networks, there’s not enough traffic engineering need yet to justify it.

But the big problem according to both our research with providers and our TCO modeling was the lack of effective operations tools for PBT networks, which by inference means Carrier Ethernet in general. It appears that TCO would be lower for a hybrid MPLS/Ethernet network or an MPLS-TP network because of better operations tools, but there is not yet sufficient information available to model this conclusion.

Nortel also had a major problem explaining its own approach, according to a number of operators we spoke with. The decision to move away from higher-layer protocols to lower-layer ones will be a difficult one to reconcile with long-term profit goals, though, because optical margins are often thinner.

Ericsson develops packet-optical gear for metro surge

Ericsson is fielding a line of packet optical gear designed to address the expected surge in metro networking. The new products will support a T-MPLS control plane, but Ericsson plans to upgrade to the more modern MPLS-TP and is also considering PBT (PBB-TE).

Metro capacity may be driven by a host of factors in 2009 and 2010 and the operators may be more interested than usual in the optical layer. This has implications on the Carrier Ethernet and IP MPLS wars since optical spending tends to encourage operators to deploy some control plane architecture, and that could then pull through either switches or routers.

Carrier Ethernet/MPLS TCO comparison brings surprises

CIMI Corporation has completed an extensive survey-and-model process on the economics of carrier infrastructure, focusing on the TCO differences between Carrier Ethernet and IP/MPLS. The results of this study will be published in the July issue of our newsletter Netwatcher, and we will also release a separate report in the fall.

The results of the study, which surprised us considerably, were that while Ethernet always generated lower capex costs, it was also unusually vulnerable to issues in service management efficiency, especially in the WAN versus metro. Those issues could swing the TCO in IP’s favor, providing that IP service management was strong. In fact, the effectiveness of service management tools on controlling operations costs and support incidents had more effect on TCO than a swing in capital cost of over 40%.

Vendors showing up for Carrier Ethernet

NXTcomm 08 was an interesting show, something a lot better than some of the disasters that followed the breakup of the SuperComm partnership, but far less than SuperComm in its prime.

<p>We believe this is due not as much to the show as to the industry; infrastructure doesn’t have buzz any more. In the heyday of the older show, the bubble was in bloom and there was a lot of trade press action around it. Today, publication coverage of infrastructure issues is down because the buyers are all big telcos who don’t do things interesting enough to make the press happy. The big news in the show was the substantial vendor presence in Carrier Ethernet.

 <p>While most of the companies that showed products were objectively doing less than half of what the market would require in terms of features, there was enough support to make it clear that despite the PBT announcement by BT, this technology isn’t going away. We believe, in fact, that the major deployments will begin to roll in 2H09 and that most of the opponents of PBT will end up quietly supporting it by then.

<p>We had 10 interviews with vendors and carriers on our ExperiaSphere initiative and we were thus able to exceed our own objectives for the show. Service management issues and their relationship to standards and to network resources are a key part of the Ethernet picture, and also key for IP/MPLS in any form. In fact, a report we are publishing in the July issue of Newatcher, our newsletter, shows that service and operations management issues with both Ethernet and IP/MPLS result in more swing in total cost of ownership than technology issues do.

ECI enters Carrier Ethernet fray

ECI is entering the Carrier Ethernet market, leveraging in part the MPLS tools that the company obtained from Laurel Networks when it acquired that company. This space is a critical one because it is the focus of some fundamental debates—is Ethernet a service framework or a metro architecture, and can it displace some or all of MPLS. We believe that Carrier Ethernet can be a service and metro architecture and can displace an important chunk of the MPLS opportunity, but only if providers understand how to build metro infrastructure from Ethernet and not just offer Ethernet services. ECI seems to be moving in the correct direction, but the details of its products are still too sparse to be sure.

Hammerhead Systems’ service-based PBT strategy

Hammerhead Systems has announced a service-based PBT (Provider Backbone Transport) strategy that adds the MEF E-LAN and E-Tree connection models to the traditional PBT E-Line model. The move is the first in the industry to aim PBT directly at creating retail services rather than at managing traffic and failure modes. Hammerhead also announced a partnership with Soapstone Networks, a unit of Avici, to provide flexible control plane support for the Hammerhead products. We believe this is an important step in evolving PBT to serve a key mission in both metro infrastructure and service infrastructure, and for the first time it gives some credence to the PBT-versus-MPLS battle that has raged in the media.