MPLS ready for China's core challenge; Chinese carriers have not yet traded in their ATM networks for Layer 2 MPLS for a very simple reason: ATM works just fine as is. However, that's about to change. China Unicom was the first to shift, and if that doesn't convince rivals, growing demand for bandwidth and service flexibility will

Telecom Asia, April, 2004 by John C. Tanner

The debate over the role of MPLS in the packet-switched telecoms network has been more or less settled for some time now. While MPLS has not been without its critics and holdouts, it is nevertheless out there, somewhere, right now. Pick a network, any network--national, regional or global--and you'll probably find MPLS serving at the very least as the key ingredient of IP-VPN services, and increasingly as a key element of a live next-generation network (NGN) platform.

You'll even find it in China--but not nearly as much of it. In fact, this time last year, you'd be lucky to find MPLS in any network for any reason apart from a field trial or, these days, a VPN service. Layer 3 IP-VPNs based on MPLS popped up all over China in 2003 in provincial and metro networks via local players like China Telecom, China Netcom and CPCNet, as well as international partners like Asia Netcom, AT&T, BT, Equant and even Hong Kong-based telcos like New World Telecom.

However, Chinese carriers have generally been playing it cool on MPLS as a Layer 2 core backbone technology, preferring to stick with its existing ATM platforms. Frank Wu, director of wireline networks at Nortel Networks China, cites two main reasons for this.

"One thing is that they are concerned about the maturity of MPLS's OA&M capabilities, which aren't quite as mature as ATM's and are still progressing," Wu says. "The other thing is that their ATM networks are running well for the time being. They use them mainly for services like frame relay and point-to-point Ethernet, and these are doing well, so they have no near-term motivation to change."

Andrew Coward, Asia-Pacific VP of technical operations at Juniper Networks, agrees, but adds that this is now starting to change.

"In China there's this gap between high-end and basic connectivity, and they've been debating how to address that--should it be MPLS, or should we extend the ATM network, or should we do tunneling? They've also been watching how MPLS performs in other networks outside of China, and now they feel there's enough momentum with MPLS elsewhere for them to follow that model."

Agatha Pooh, Asia Pacific telecommunication strategies analyst with Yankee Group, concurs. "Chinese operators are well aware of the current technology trends and developments," she says. "Given that the performance of MPLS has been improved over the router platform, they understand that the core network will be increasingly based on an IP-based platform."

Coward adds that it's part of a new shift in focus for Chinese carriers as they look beyond simply building out infrastructure and face the reality of a market that is primed for more expansion on multiple levels, from economic growth and competition to bandwidth demand and network convergence.

"The priority has always been to get China connected. You know, Phase 1 was building out dial-up, and Phase 2 is rolling out broadband. Now they're starting to turn their attention to adding value, creating new services and generally getting more revenue out of their networks," he says.

If it ain't broke ...

As mentioned earlier, one of the chief obstacles to Layer 2 MPLS in the Chinese market is the reluctance of Chinese carriers to replace or otherwise fool around with ATM networks that work just fine.

Part of the reason for this, says Wu of Nortel, is that frame-relay/ATM services are popular among enterprise customers in China, and no carrier wants to muck around with a service customers like.

"The ATM/frame relay services are targeted at high-end enterprise users, which is a relatively small number of users, compared to the consumer/residential market, but the revenue for carriers is still high-margin," Wu explains.

Also, one of the key attractions of frame/ATM for corporates is the ability to guarantee QoS, says Hou Lianchang, product marketing and business strategy director for the IP division of Alcatel Shanghai Bell.

"The enterprise users still prefer frame relay/ATM because it provides higher quality," Hou says. "MPLS hasn't been able to provide the same level of QoS."

Meanwhile, other more basic concerns about MPLS include router reliability, vendor interoperability and the cost of deployment. The latter is a particular sore spot because upgrading an entire network--especially one in a country the size of China--is a major commitment, says Hou.

"You really need to upgrade the whole network to support MPLS, which becomes a big cost issue for the carrier," he says. "If you only have to upgrade one or two nodes, maybe it's okay, but if you have to do the whole network, that's a problem."

At the same time, says Coward of Juniper, ATM is an even more expensive upgrade in terms of scalability. "If you want the core to scale from 10 Gbps and beyond, it's very hard and expensive to do with ATM," he says.

Unicom takes the plunge

In fact, according to Coward, it was this issue that convinced China Unicom to take the plunge and finally deploy a Layer 2 MPLS core network.