Convergence is telecommunication's Holy Grail – a single network infrastructure delivering all services. Through the 1990s, ATM was touted as the technology to achieve this goal. While it had significant success within carrier networks, we're still waiting for 'ATM to the desktop'. Today, attention has turned to IP/MPLS networks. IP is already at the desktop, but can it deliver on the promise of convergence?
According to 'boom years' hype, IP was taking over. Telephone traffic was moving to VoIP (Voice over Internet Protocol). The network was developing in 'internet time', meaning equipment that took decades to achieve acceptance would be replaced within years.
In hindsight, customers were never willing to pay for this frantic growth and, in today's post-crash telecommunications world, everything has slowed to a more reasonable pace. We recognise convergence is built on evolution, not the revolution once so boldly predicted.
However, the promise of convergence still looms and internet time is still ticking. A recent industry report found the six major telecommunication equipment vendors (Alcatel, Cisco, Lucent, Marconi, Nortel and Siemens) all agree that the future belongs to converged, multiservice networks based on IP/MPLS. A similar survey indicates that all major service providers intend to migrate to a converged IP/MPLS infrastructure.
The modern vision of convergence focuses on the 'multiservice edge', where disparate backbone technology is unified under a common IP/MPLS layer and managed as a single core network. At the periphery, intelligent multiservice edge devices admit traffic from a variety of end points, each possibly using a different access technology.
'Pseudo-wire' technology makes this possible, with existing services transported across the core to a corresponding device on the opposite edge, where the original service is recreated. The core network can be upgraded or replaced, with no impact on the end points or revenue-generating services.
Convergence in the metropolitan and access networks offers different challenges. Though TDM remains the dominant access technology, bandwidth scalability is an issue. Service upgrades require the costly and time-consuming physical provisioning of a new circuit. Ethernet allows customers to purchase bandwidth incrementally or on demand, and is fast becoming the access technology of choice for high data-rate connections.
Most enterprises use Ethernet within their internal networks, making it an easier 'sell' since they understand its capabilities and benefits. Services such as VPLS (Virtual Private LAN Service) or E-LAN (Ethernet LAN service) allow carriers to offer a 'virtual LAN experience', seamlessly extending a company's LAN between sites.
As with IP/MPLS in the core, Ethernet may be considered as a unifying layer over a disparate array of physical access technologies. The ability to offer a familiar and easy-to-manage service over whatever access technology is present is influencing the move towards Ethernet.
Key to a truly converged network infrastructure is the ability to transport voice efficiently and reliably. Incumbent carriers have traditionally been cool to VoIP technology. Voice still represents significant revenue, and carriers were reluctant to entrust their bottom line to a technology that frankly didn't work very well. However, this is changing. Cable and broadband companies are rolling out the 'triple play' – combining voice, video and data services over one connection. Triple play is now impacting incumbent carriers, who are being forced into offering broadband services – including VoIP – to maintain their connection with the consumer.
While the pace has changed, convergence is definitely still on the agenda. Rather than a revolutionary network replacement, existing physical networks are being combined into a common layer, with intelligent edge devices interfacing to the various customer services.
Transitional strategies, such as pseudo-wires, maintain existing services and revenues until network infrastructure is upgraded or replaced as budget allows. This allows evolution towards a converged network to be pursued in a more sustainable manner.