MPLS: The icing on the networking cake?

Are we on the verge of a step-change in the quality of networking, a change that will put what we've come to expect from the likes of Asynchronous Transfer Mode (ATM), Frame Relay and all-IP to shame? Heather McLean has been investigating.

Multi protocol label switching, sexily tagged MPLS, is the juvenile standard being developed to give service providers better control over networks.

In the metropolitan and wide area networks (MAN and WAN) MPLS has a bright future. The technology is lauded as bringing greater efficiencies to all manner of networks.

Iain Stevenson, networking analyst at Ovum, put it like this: "MPLS is a mechanism for labelling traffic flows to define a path through a complex network. A network operator can use MPLS as a management tool, as capacity and routes can be seen."

To give you a mental picture MPLS could be described as a stiff wooden spoon bashing away at an internet protocol (IP) network that stubbornly refuses to move its thinking beyond simplistic.

Traditionally, a packet of data sent over an IP network goes straight from the original point to the receiving point following the shortest possible route, regardless of traffic volume at any particular time.

If all packets are sent using IP, the shortest points between popular destinations can become bottlenecks making the possibility of providing time-delay sensitive services - such as video or voice - difficult.

The delivery of value-added content services is key to the future of networking. Bert White, CEO of networking equipment company, net.com is optimistic about the technology.

"MPLS potentially solves a fundamental problem that has plagued IP networking. It's currently very difficult to put applications that are delay-sensitive over the net.

"MPLS also has the potential to allow service providers and operators to consolidate voice and IP networks," he said.

The next step for the future of networking means IP delivery faults need to be tweaked so service and content providers can churn out the advanced services already in demand.

It's all about control. A network spiked with a little MPLS can route data across, not necessarily the shortest, but the clearest route and provide dedicated bandwidth so consumers can chose which service level and type of agreement they require.

However, Martin van Schooten, product marketing manager EMEA at Extreme Networks, is happy to detract from the MPLS hype:

"MPLS is important but it's not earth moving. You can do high-speed services with guaranteed service levels and variable bandwidth without MPLS. The development of 10 Gigabit Ethernet is more impressive," he said.

There are three main functions of MPLS, quality of service, traffic engineering - which includes a unique ability to re-route traffic very quickly in case of network failure - and enabling virtual private networks (VPNs).

An MPLS network will run over whichever mix of networking technologies it faces, including Ethernet, ATM and Synchronous Digital Hierarchy (SDH).

MPLS is backed by heavyweights like Cisco, Lucent and Nortel who are involved in the core and newer companies including Juniper, Pluris and Torrent - a subsidiary of Ericsson - beavering away on new variations.

However, Cisco has been accused of dipping its fingers a little too deeply in the technology's development. The organisation has a vested interest in making MPLS popular because of the technology's strategic importance to the company's core disciplines.

Extreme's van Schooten commented: "This is a multi-vendor game even though Cisco has a big stake. Cisco is getting a fair amount of competition as it's strong in the core but not in the MAN area."

Although Cisco's involvement alone does not mean MPLS will definitely be on the market in five years time, it does increase the odds, which has made Ovum's Stevenson suspicious.

He said: "MPLS's impact on networking is significant but there's a lot of proprietary work going on around the edges."

There are a whopping 72 MPLS types in development carried out by different organisations. Protocols are tending to creep out of the tiering system that defines which other technologies they can be used with, so the MPLS sprawl has caused some IT professionals to criticise developers.

Van Schooten added: "There is some interoperability between vendor versions. However, the lack of maturity of MPLS means we should have a cautious reaction to claims of support. We have to see what they mean."

Stevenson agrees with van Schooten. "If I were a carrier I'd want to go over the claims of interoperability these vendors are making with a microscope," he warned.

Take MPLS VPN. Of the myriad of choices, two can provide much needed scalability as an alternative to ATM, Frame Relay or IP secure tunnels for service providers. These are RFC 2547 and Draft Martini.

Giles Herron, principle network architect and co-designer of Draft Martini, commented on Cisco's ownership of 2547:

"This is about locking customers in. Cisco is pushing 2547 into corporations. I have doubts about 2547 as it's more complex and therefore harder to support. The version gives the customer an outsourced relationship, where all the routing is done by Cisco."

Herron's bias has led him to favour Draft Martini.

"Martini is a twenty first century Frame Relay. It's simple and works in a similar fashion to ATM and Frame Relay," he said.

One problem with these slightly proprietary derivatives of MPLS is a lack of ratification through a standards body, yet this does not deter Herron.

He added: "For a standard to become popular you don't need perfection and navel gazing. You need a rough consensus and running code. If hundreds of service providers deploy it, it doesn't matter if it's been ratified or not."

The popularity of MPLS is picking up rapid speed. Although there are undoubtedly alternatives, MPLS is a technology ready to be used right now. It also has the financial and moral backing of key vendors and developers in the networking community.

But of course history has taught us these credentials don't necessarily mean it will definitely be a success. Let's just see what the future brings.