Perhaps not surprisingly, vendors of Layer 2 ATM switches say carrier requests for proposals show more interest in Multi-protocol Label Switching than demand.
Among incumbent local exchange carriers and regional Bell operating companies, which have separate ATM, TDM and IP backbones, MPLS is deployment is still off by about two or three years, vendors say. The technology is sought after to consolidate those three separate networks and support multiple services - voice, data and video.
Yet concerns over capital and operational expenditures, the immaturity of MPLS, and the familiarity and revenue-producing reliability of current infrastructures tempers the urgency of implementation.
ISPs, meanwhile, are more bullish on MPLS because they have a single IP backbone that can readily benefit from the traffic engineering aspects of the technology and as an enabler of IP VPN services - they don't face the mutliservice support requirements of their ATM counterparts. With little or no investment in ATM or TDM, ISPs can be a bit more liberal in injecting their networks with MPLS, vendors say.
"The timeline for MPLS as a hard capital budget requirement is moving out" among incumbent carriers, says Dennis Rainville, president and CEO of multiservice switch start-up Equipe Communications Corp. "ISPs want MPLS for traffic engineering and VPNs, but incumbent local exchange carriers have completely different requirements. They have ATM and frame relay to bring along."
Core data switching RFPs from incumbents seek increased scale in bandwidth, ATM virtual circuit density and DSL aggregation, incremental ATM and frame relay service differentiation, voice transport, and 2.5G and 3G wireless support, Rainville says. Multiple RFPs have specified faster ATM call processing, scalability on the order of 1 million or more VCs, and 10G bit/sec OC-192c ATM interfaces for near-term production deployment, he says.
MPLS, meanwhile, is requested for trial purposes with initial deployments not expected until 2004 or 2005, Rainville says.
"Next year, there will probably be a lot of (MPLS) trials," Rainville says. "The seven (largest incumbent carriers) are only dabbling right now. In 2004, 2005 it will ramp up and then there will be a seven-year bell curve behind that."
According to an "unofficial" tally on the Web site of Cellstream, a telecommunications consultancy, 76 service providers worldwide are deploying MPLS. The lost includes AT&T Corp., Sprint Corp. and Worldcom Inc., and only one U.S. regional Bell operating company - Qwest Communications International Inc.
Carriers are evaluating MPLS as a method for consolidating ATM and IP networks, and for "future proofing" current investments in ATM switches, according to Alcatel SA. But two key issues remain before MPLS is widely deployed: the common refrain of reliability; and the potential to use MPLS for more than traffic engineering and VPNs.
"Can we use it as a service?" asks Jim Guillet, assistant vice president for product marketing in the Edge Data Networks group of Alcatel's Carrier Internetworking Division. "It's now an issue of what is the motivator to evolve. There are some issues as to whether IP and MPLS are ready for business critical services. To make IP the primary communications medium, you've got to make it reliable."
Edge switch maker WaveSmith Networks Inc. agrees on both counts:
"MPLS services, apart from VPNs, are a ways off," says Chad Dunn, WaveSmith director of product management. "Transparent LAN services are happening but end-to-end MPLS is still a ways off." In an effort to prove that MPLS is stable enough to support services and generate revenue, the MPLS Forum recently sponsored the largest interoperability demonstration to date of the technology at the recent SuperComm 2002 trade show. Twenty-one vendors and 27 pieces of equipment displayed traffic engineering using explicit routes, Layer 3 MPLS VPN tunnels and Ethernet over MPLS tunnels.
Stability and service revenue issues aside, MPLS would probably be more widely implemented right now were it not for the telecom bubble bursting in early 2000. But burst it has, so carrier requirements are now back to areas where they know they can make money, according to Nortel. "Eighteen months ago, people were pushing harder on MPLS," says Errol Binda, senior manager for product marketing at Nortel. "The realities of the industry are sticking to what's proven. People are still asking for MPLS as an end-game, but it's not as critical."
People would be asking for it a lot more fervently if vendors offered the right solutions, says Sarbpreet Singh, vice president of product and program management in Lucent Technologies Inc.'s Internetworking Systems group. Naturally, that's where Lucent has the others licked, Singh believes.
Lucent boasts an ATM/MPLS control plane breakthrough with its TMX 880 MPLS core switch, which will be generally available in September. The switch's Fluid Signaling feature maps ATM VCs to MPLS label-switched paths, and proxies ATM OAM&P flows through the MPLS domain to maintain service monitoring and integrity.
"We are actually accelerating the (MPLS) market," Singh claims. "In 2003, you will start seeing deployments. If RFPs called for MPLS in 2004, they wouldn't be issued at this time. And if we didn't have this capability, deployments would be in 2004 or 2005."
Marconi Corp. PLC also believes 2003 will be the year of MPLS early adopters among the incumbents. The company's been invited to propose its BXR-48000 core switch in ATM and IP/MPLS bids, 70% of which have called for a packet switched multiservice core.
"There's been a lot of investment in MPLS capabilities but service providers are still being a little guarded," says Harry Ostaffe, director of product line marketing for broadband routing and switching at Marconi. "They want to make sure it's fully baked and fully interoperable."