MPLS: King for a Day?
The project, called “KING” after the German for “components for next-generation Internet,” aims to demonstrate that connectionless protocols like Internet Protocol (IP) offer the best chance of giving carriers the three things they really care about on very large networks: control over quality of service, control over network reliability, and automation of network management.
The implication of this project is that MPLS can't cut it on these issues because it relies on setting up virtual connections across networks -- in the same way as frame relay and Asynchronous Transfer Mode (ATM) do. As the number of virtual connections increases, the task of keeping control of everything gets tougher and tougher.
MPLS "is something like ATM in a different color," says Karl Schrodis, head of technical planning for the KING project.
The difference between connection-oriented and connectionless protocols is similar to the difference between rail and road transport. Railway companies have to keep a close eye on where trains are to avoid collisions on their connection-oriented tracks, and doing that gets increasingly hairy the more tracks and trains they have [ed. note: especially if your trains are in the U.K. and being run on rails overseen by Railtrack].
In contrast, no central control is needed for folk driving automobiles. They may set out with a particular route in mind but can choose to divert around any congestion they encounter at almost any junction.
The key point is that connection-oriented protocols avoid congestion while connectionless ones find a way around congestion, if and when it happens.
This means that connectionless protocols make better use of infrastructure when handling volatile Internet traffic. They're also potentially much more scaleable because there's only a general need to make sure there's enough overall network capacity to carry the traffic. As a result, they're easier to manage and their resilience increases with size, because there are more options for avoiding congestion.
All of this has been proven on the Internet, based on the connectionless IP, which has scaled so well and amazed everybody by its resilience.
Originally, MPLS was invented to speed up packet forwarding in IP backbones, according to Schrodis. However, advances in hardware technology by companies like Juniper Networks Inc. (Nasdaq: JNPR) have rendered this unnecessary. "The reason for having MPLS has gone," he says.
"We don't want to say MPLS is generally bad," Schrodis hedges, probably because it would be politically inept. Siemens, like plenty of other vendors, has climbed on the MPLS marketing bandwagon.
Schrodis and his colleagues in the KING project are aiming to add some bells and whistles to existing connectionless IP protocols to improve their performance. The multimillion-Euro project started last October and is scheduled to continue for three years. This will include "some sort of field trial at some point in time," says Schrodis.
Three fundamental additions to IP are being considered in the KING project.
The first ensures that every node in the network can provide at least two alternate routes for outgoing traffic. It also provides a balancing mechanism so that outgoing traffic can be split among multiple ports.
The second establishes a mechanism for controlling quality of service that balances traffic load and network capacity in a more general way than the admission control that's applied to individual links in ATM networks.
The third development introduces mechanisms for dealing automatically with common network problems. This aims to improve resilience while minimizing the need for manual intervention by a carrier's operations staff.
Right now, Schrodis is reluctant to divulge further details of what could end up being an MPLS killer. Time will tell whether the concept will go the way of SMDS (switched multimegabit digital service), a project that also trumpeted the advantages of connectionless protocols but never gained serious traction.
Still, the folk backing the KING project deserve to be taken seriously. In addition to Siemens, they include famous technical academic institutions such as the Fraunhofer Institute as well as universities in Essen, Karlsruhe, Munich, Stuttgart, and Würzburg.
Siemens et al aren't the only ones looking to retrofit IP with new capabilities. Dr Lawrence Roberts, one of the Internet pioneers, is also said to be working on similar schemes at his latest venture: Caspian Networks (see Dr. Lawrence Roberts and Caspian Study: The Internet's Exploding).
— Peter Heywood, Founding Editor, Light Reading
http://www.lightreading.com Want to know more? The big cheeses of the optical networking industry will be discussing this very topic at Opticon 2002, Light Reading’s annual conference, being held in San Jose, California, August 19-22. Check it out at Opticon 2002.
Register now and save $500 off the registration fee. Just use the VIP Code C2PT1LHT on your registration form, and deduct $500 from the published conference fee. It's that simple!