The successor to Lucent's discontinued next-generation multiservice core switch will be a 160G bit/sec MPLS switch with a gateway to migrate Lucent's installed ATM base to the IP technology, Network World has learned.
The TMX-880 Multiservice Xchange Switch is in "early" testing now and will be available in controlled release in late January 2002, according to an internal Lucent memo obtained by Network World. The memo, authored by Ken Packert, Lucent president of multiservice switching, states that the TMX-880 supports ATM termination, frame relay and IP in a single platform, and will "seamlessly" integrate into existing Lucent frame relay and ATM networks via V-MPLS, a VNN-to-MPLS gateway.
VNN is Lucent's proprietary technique for setting up and tearing down virtual circuits, a technique employed in Lucent's GX 550 and CBX 500 ATM switches.
Lucent would not comment on the memo, but did not dispute its authenticity.
"We will detail our new multiservice solution at a later date," a Lucent spokesman said.
As a result, details on the architecture and the technological origin of the TMX-880 could not be learned by press time. The Packert memo stated that the switch will be managed by Lucent's NavisCore management system and "provides customers support of existing networks and bridges them to the new technologies many are planning and implementing now."
Observers speculate that the TMX-880 may be based on the Nexabit NX64000 IP router Lucent obtained two years ago via its acquisition of Nexabit Networks. Other believe it will be a hybrid of the NX64000 and the discontinued MSC 25000, or that it may even be an ATM core switch with IP/MPLS processing capabilities -- much like the MSC 25000 was designed to do but in a much smaller footprint.
"The 25000, in my view, was too IP-ish," says Tom Nolle, president of CIMI, Voorhees, N.J. "What Lucent is going to do is develop an MPLS strategy more out of evolving their ATM switches. Their customers are not looking to deploy a new IP/MPLS core; they're looking to deploy an ATM core that can be evolved to MPLS."
"The development of the ATM product is going to probably end up being the foundation of the next-generation product for the IP world," says Frank Dzubeck, president of Communications Network Architects in Washington, D.C.
Meanwhile, the Packert memo went into detail on the rationale for killing the MSC 25000. Entitled "MSC25K Cancellation and Re-positioning Core roadmaps," the memo stated that the decision to end the product did not come easily.
"The main drivers were market demand and product competitiveness," the memo stated. "The market is not firming up around the MSC. Initial '02 revenue targets were $75 million. The latest numbers were between $5 (million) and $20 million."
The memo also states that demand is down because some customers have overbuilt their core and are not looking for expansion except in "critical hotspots." It says that the longevity of ATM is being questioned daily.
"Overlay IP networks are a real threat," Packert states in his memo. "Adding more core capacity just increases (service provider) capital expenditures and does nothing to revenue."
Competitively, the MSC as a platform "missed the window," the memo states. "Building a system that takes up ~8 racks to deliver 320G bit/sec of bandwidth is just not competitive - especially in the '03 timeframe. Positioning it as a pure ATM switch has come under fire because we cannot promise the escape valve that the customers demand - a path to MPLS."
The MSC MPLS development would cost another $25 million annually until it shipped in mid-2003.
"If it were here today we would use it, but it isn't," Packert states in his memo.
While the TMX-880 is readied for the core, Lucent will apparently now focus ATM and frame relay activities on the edge, according to the memo.
"While building the MSC, the edge ATM/FR services have been neglected, starving the programs that allow our customers to generate more revenue," the memo states. "We need to refocus our efforts on the network edge, helping to drive service provider revenues."
Those efforts will include increasing the densities of customer-facing interfaces, supporting circuit emulation for packet-based transport of traditional lease-line services, developing inverse multiplexing for ATM and DS-3 subrate cards, and more IP.
"We will be improving the IP capabilities on the existing MultiService Switch platforms to enable basic IP services on existing networks at a lower cost," the memo states. "And for more advanced service enabled networks, we will provide seamless integration of the Springtide platforms, giving service providers even more options at the edge of the network."