Service providers still derive most of their profits from legacy voice, Frame Relay, and ATM. They are not making money from selling Internet-based services, yet most believe that the IP network will form the future of all communications. So why does the industry continue to invest in this technology and announce ever higher speed backbone and metro-access deployments?

The answer is simple: While IP traffic has not achieved the 80-100 percent growth per quarter that was anticipated by pundits and carriers alike, it is still experiencing an average growth rate of 80-100 percent per year. With this growth rate and solid drivers behind it, this is still in one of the fastest growing markets in the world. The technology also provides exciting new economics for operational efficiency and raw speed. We can provide converged services from a single network. We can create globally accessible networks that provide uniform services and costs independent of location.

So where is the money? All providers and vendors are struggling with this question today. Every day new services are announced--ranging from video streaming to corporate IP Virtual Private Networks (VPNs) — but these services are not taking revenue streams away from the more traditional leased-line, ATM and Frame Relay services on which companies still depend. There is a direct relationship between service availability and gross margins. Gross margins on legacy services are in the 60+ percent range which correlates to network availability of four-nines. Compare this to IP services, where 20 percent gross margins are directly tied to the lack of availability.

The legacy IP equipment currently deployed in carrier networks was designed to provide best-effort IP services and has never seriously addressed the requirements of carrier-class-grade services. Only when carriers can deploy solutions backed by the same stringent Service Level Agreements (SLAs) that traditional services can offer, will customers be tempted to take up this new technology and use it for business-critical applications. Until this occurs, carriers' business cases will continue to collapse under the availability of less expensive best-effort services from the "me too" new entrant carriers. Consolidation will certainly take care of a lot of these new entrants, but carriers will still need to be competitive with those left standing.

As IP networks evolve from supporting e-mail and web browsing to more demanding applications such as voice, video, and VPNs, the standards for IP network reliability must also evolve.

The current gap between IP and traditional network reliability directly impacts a service provider's bottom line. IP network equipment must not only support high packet delivery, but deliver packets with stringent latency parameters (less than 50 milliseconds). This latency prerequisite also requires fewer router failures that result in rerouting and excessive latency.

Carriers must begin to hold their IP networks to the same reliability standards of current legacy networks to support real-time services.

Causes of failure

Carriers are beginning to analyze the root causes of poor quality and to implement corrective measures. According to a recent study, router operation represents the single largest cause of downtime (36 percent) in core router networks. Router operations encompass router installations, chassis upgrades, router upgrades, line- card installations, and configuration changes.

Carriers seeking to reduce the network downtime resulting from router operations must deploy routing platforms that improve the operational stability of the network by reducing or eliminating disruptive operational activities such as forklift upgrades, topology changes, and link expansion.

Link failures represent the second largest cause of IP network downtime (32 percent). To overcome link failures, carriers are focusing on reengineering router diversity and seeking protection solutions in the backbone.

Router failure is the third largest cause of IP network downtime (23 percent). Traditionally, core routers lacked the redundancy, scalability and robust software that have been the hallmark of carrier-class voice and multiservice data switches. New carrier-class router platforms are built specifically to support high-availability services. Carrier-class redundancy, robust software and sophisticated backbone protection mechanisms such as composite trunking for fast failover and MPLS fast reroute local protection are the tools being used.

For example, Multi Protocol Label Switching (MPLS) delivers a number of key attributes to service provider networks. One of the most interesting is the fast reroute or local protection capability for the backbone network. Fast reroute provides Sonet-like restoration, supports interface types other than POS such as Gigabit Ethernet, and does not require the investment in idle protection fibers. Backing up fiber using Sonet is particularly expensive for high-speed OC-192 interfaces.

Fast reroute enables service providers to set up a backup network path with certain bandwidth stipulations. It also provides a way for traffic to be explicitly routed down that path in the event of an interface or link failure. An MPLS backbone running fast reroute can redirect traffic down a predetermined alternate path that has the service quality characteristics required for that particular traffic flow. This speeds up recovery times in the event of a failure, translating to considerably higher network availability and performance. With these improvements, carriers running MPLS fast reroute can offer premium services, including VPNs or Voice-over-IP (VoIP) to their customers.

While carriers investigate services that can generate revenue over the coming years, they are also being forced to cut the cost of building and operating the networks to carry those services. One of the most efficient ways to slash infrastructure costs is to reduce the amount of excess equipment. This can eliminate layers that complicate the network and reduce the number of networks that are used for data services.

For example carriers might have three or more data networks. One could be could be Frame-over-MPLS, another could be IP running across ATM transitioning to MPLS, and yet another could be Frame-over-ATM. It can also include deploying routers that are scalable in capacity. It is no longer about raw speed and ports, but rather how a router scales in service to support near and long-term requirements. Upgrading routers and spending money on interfaces to interconnect routers together is no longer feasible.

The network that is envisioned by many carriers will utilize MPLS-enabled scalable IP routers. Network efficiency through the elimination of redundant layers and equipment is increasingly viewed as a necessary migration scenario.

To meet the demanding needs of their customers and to remain competitive in this economic environment, a world-class IP service provider must demonstrate scalable capacity, advanced technology, and dependable service. This means providing consistently low latency supported by SLAs of no greater than 50 ms for a network-wide round trip; low packet loss; network availability of at least four nines with a credible roadmap to five-nines; multilayered security and physical security for all backbone nodes; support for VPNs, VoIP, and lastly global reach.

So carriers must deploy IP routers that were designed from the start to support carrier-class requirements. These routers support scaling from 5 Gbits/s to more than 5 Terabits/s of capacity, as their demand increases, lower the cost of building and operating IP networks and deliver the highest levels of network reliability to increase Internet services availability and profitability.