Lord of the rings


A look into the mysterious voids between LANs and Networks of Core fortells the end of the battle between Gandalf and Aladdin - and Darwin's evolution. Spooky!

Today's Metro market space has two key characteristics: heat and confusion. Heat comes from the inexorable logic of linking faster enterprise networks to cheap, abundant core backbone networks. In the enterprise, there is almost no debate about protocols and architectures; Ethernet has easily displaced all rivals. Confusion reigns in the service provider networks due to the competing technologies (Circuit, Cell, Packet, Frame), transport protocols (SONET, ATM, and FR), and service providers' numerous network topologies.

At the risk of over-simplifying, there are three basic approaches towards resolution:

Enterprise technologies move from the LAN to WAN
These focus on packet-only solutions which are low-cost and immature ways to manage and protect the network. MPLS leads the thinking in providing a longer-term replacement to the complexities of engineering the public network. The RPR (Resilient Packet Ring) alliance was born from this general background. This should provide carrier-class protection in packet-based rings without the overheads of standard SONET. Standardisation squabbles blur the picture however; the relative merits of "Gandalf" and "Aladdin" standards are debated within the vendor community.

Carrier TDM technologies move Metro Core to Metro Edge
These are characterised by a combination of virtually concatenated SDH/SONET and DWDM technologies, referred to generally as Next Gen SDH/SONET. Typically the Metro Core network is enhanced optically with metro-optimised WDM technologies and newer packet-over-SDH/SONET technologies. Protection here uses established SDH/SONET BLSR, UPSR 1+1 resilience. The strength in this approach is the familiarity of the technology for the carriers; the weakness is the cost and complexity of using technologies never designed for data. As the dominant payload in the network becomes packet, TDM multiplexing alone will never provide an optimised solution.

Hybrid "Layer 1.5" solutions
These provide multi-protocol interfaces to the enterprise, and SDH/SONET-like interfaces to the core network, providing an intelligent middle layer to optimise multi-protocol packet traffic aggregation, bandwidth management, ring load balancing and service provision within the fibre. They leverage existing SDH/SONET protection without the traditional overhead costs, and with MS Spring eliminate the final 50% penalty normally associated with established ring and 1+1 protection. Combining layers 1 and 2 functionality (and in some cases such as Native's EtherMux replacing some layer 3 functionality such as VPN) they simplify the network architecture.

Among carriers, RPR was a regular topic of conversation six months ago, but some doubted that it would ever emerge. However, a version of the RPR standard will develop and consensus will be restored. At Native Networks, we track the progress with interest; we have an "RPR plus" architecture that offers higher security through the use of MPLS sublayer. The lessons from ODSI clearly showed the penalties in terms of delay, wasted development and wasted marketing dollars that accrue from splitting a standard for the sake of vendor differentiation. The protagonists of the alternate RPR flavours should learn these lessons, and market logic should eventually prevail.

A few days after completing this article, and as predicted here, the two divergent RPR groups got together, and resubmitted their standards proposals in a single joint document now named "Darwin". Market logic - today's natural selection - once again has prevailed.

More in Packet Transport