When and how to migrate to a converged MPLS core

Today, most service providers maintain an ATM network to deliver ATM, Frame Relay and private line services along with a separate IP network to provide Internet access and voice over IP. Many would like to consolidate these services onto a converged Multiprotocol Label Switching (MPLS) core, but lack the solid economic data they need to plan a cost-effective migration. 

A recent Bell Laboratories study, "Quantifying the Value Proposition of MPLS Evolution: Why and When to Migrate to a Converged MPLS Core," reveals that carriers can reduce their total cost of ownership by more than 20% when they evolve their separate ATM and IP networks to a converged MPLS architecture. A follow-on study by the same team focuses more narrowly on the impact of migrating MPLS to the core network. This more focused look reveals that when and how providers migrate to MPLS can ultimately determine what they gain from that migration. Analyzing the network's total traffic, current service mix and IP growth rates can help pinpoint when migrating to MPLS is a smart move--and when it is merely an expensive one. Network management system and MPLS core equipment decisions are also critical. This article describes how providers can manage both timing and technology to derive maximum benefit from MPLS. 

As Figure 1 below shows, service providers with dual ATM and IP networks commonly scale those networks to meet demand in one of three ways:

• Growing the existing ATM and IP networks separately to meet increasing service demands (without implementing MPLS)

• Implementing an overlay MPLS solution that preserves (and scales) a separate ATM network while evolving the existing IP network to a higher-capacity MPLS core

• Converging both the ATM and IP networks onto a common MPLS core capable of supporting all current and future services

 

Figure 1: Common growth strategies for service providers who maintain separate ATM and IP service networks.

The Bell Laboratories research team evaluated the relative economic costs associated with each growth strategy. Their work confirms that converging ATM and IP service networks onto a common MPLS core can produce significant cost-savings over both the status quo and overlay MPLS strategies. In fact, service providers who implemented a converged MPLS core can improve the net present value of their operations by:

• 28% over a strategy that maintains separate ATM and IP networks (no MPLS)

• 21% over overlay MPLS core solutions

This net present value increase comes from two sources. First, implementing a converged MPLS solution can reduce the equipment, transport and operations costs in the access and core networks. Second, it enables providers to generate new revenues by introducing IP Virtual Private Network (VPN) and other IP services after the migration.

Of course, there are caveats. Migrating to MPLS did not reduce the cost of scaling network services in all situations. Also, the actual cost-savings depends to some extent on the network management system and MPLS equipment used to build the MPLS core.

Knowing when to migrate to MPLS can be just as important as deciding whether to migrate at all. Although the potential savings are compelling, migrating to MPLS may not make economic sense in every situation. As Figure 2 shows, the network's total traffic, mix of ATM and IP services, and IP growth rate intersect to create a Zone of Advantage within which the economic return justifies converging separate ATM and IP networks over an MPLS core. At one end of the Zone, service providers with an initial service mix of 70% ATM and 30% IP require an IP growth rate of 90% or more to reliably realize cost-savings from converging those networks over MPLS. At the other end, providers with an initial service mix of 10% ATM and 90% IP typically experience at least some cost benefit from an MPLS convergence solution. 

 

Figure 2: The interplay between the initial service mix, traffic levels, and IP growth rates creates specific scenarios in which it makes sense to migrate to MPLS.

In general, the more IP traffic the network already carries and the faster that IP traffic grows, the more likely a provider is to reduce costs by implementing a converged MPLS core. Consequently, service providers should move to a converged MPLS solution if they either:

• Already deliver comparable amounts of ATM and IP traffic 

• Have an IP growth rate that will generate roughly equivalent levels of IP and ATM traffic (within approximately one to two years)

Migrating to MPLS complicates network operations by introducing another technology across which operators must provision, monitor and maintain services. To converge services using MPLS, service providers must weave together multiple infrastructure layers, device types and protocol domains into a single coherent service platform. At the same time, to minimize the impact on existing customers, service providers must maintain established customer connections and meet the reliability and network performance commitments outlined in their existing service level agreements.  

Many MPLS solutions require operators to use a hodge-podge of single box element management systems, script sets and command line interfaces to manage their converged network. This approach generates provisioning, monitoring and fault management costs that can eclipse any equipment or transport savings realized in the core. To maintain their existing cost structure, service providers need an MPLS solution that uses an integrated network management system capable of managing services across all network layers, technology domains, and vendor platforms. These systems typically integrate device-specific EMS software with service provisioning and fault management software to automate the process of creating, activating and managing the quality of all services delivered over the converged network.

The importance of using an integrated network management platform to control costs can easily be seen in fault management operations. To effectively resolve network faults in a converged MPLS network, operators must correlate alarms from multiple network layers and suppress any redundant fault information generated by affected devices. Manually sorting the thousands of alarms generated by the affected devices, links and other underlying network facilities can make it virtually impossible to identify the root cause and restore services in a timely fashion. 

An integrated network management platform, however, typically includes cross-platform fault management software that can filter through alarm and trap notifications to quickly identify the root cause of any service failure. During the course of its research, the Bell Laboratories team found that this automated approach can speed time to dispatch by 40% to 66% and time to repair by 25% to 37% over manual fault management procedures. Automating the fault management process can also help providers better manage existing service contracts by quickly correlating problems to affected customers and prioritizing specific repairs. The labor savings from integrated fault management operations can be particularly important for operations in North American and Western European countries that have relatively high labor costs.

Using an integrated network management system also reduces the costs associated with introducing and scaling new services across the MPLS network. As Figure 3 shows, using an automated provisioning system can reduce the cost of designing, provisioning, and testing new IP VPN services by 72% when compared to a CLI/script-based provisioning system. Service providers can reduce their provisioning costs by as much as 98% by integrating the automated provisioning system with the other components of their OSS, such as their order/entry system.

These savings came primarily from simplified training, service design, creation, testing, and configuration procedures. With an integrated network management platform, operators can quickly create new edge connections and establish core paths to reliably transport all service traffic.

Figure 3: Automated service provisioning can significantly reduce the management costs associated with introducing new services.

When planning their MPLS migration, operators should keep in mind that their core routers' MPLS feature support, reliability and scalability will affect their potential economic return.

First, to maximize potential cost savings, providers should look for equipment that already supports critical MPLS reliability and service support features, such as MPLS Fast Reroute, Martini-based encapsulation, ATM-to-MPLS interworking, and MPLS traffic engineering. Waiting for a vendor to release hardware or software upgrades to add a critical feature can indefinitely postpone the entire migration. In addition, when the software or hardware upgrades finally are available, operators must allocate additional time and resources to retrofit, test, and integrate the new functionality with their existing equipment.

Second, service providers who have more than 30% IP traffic or high IP traffic growth rates in their existing network can significantly reduce long-term capital costs by deploying high-capacity routers for their converged MPLS core. Some providers try to minimize their capital costs by using multiple smaller routers to size their core network to meet current and future traffic demands. While this can work in the short-term, it is only cost-effective to a point. To create a cohesive MPLS core, service providers must maintain connections between each core router used in a given central office, so that these routers can function as a single network node. As providers add new routers to the CO, the overhead from maintaining these inter-router trunks will eventually reduce the efficiency of each CO to zero. The increased overhead will eventually force service providers to purchase more expensive, higher-capacity routing equipment anyway. The Bell Laboratories team found that implementing higher-capacity core routers from the outset could reduce long-term equipment costs by 40% to 65%. The higher the IP traffic growth rate and the higher the initial load of IP traffic, the greater the cost savings from implementing more scalable core routers.

Finally, providers who operate large core hubs should look for routers that efficiently use available central office space and power. Implementing a converged MPLS network using high-capacity core routers can significantly reduce space and power requirements when compared to strategies that maintain separate ATM and IP networks or that implement an overlay MPLS solution. Since this is primarily due to reduced equipment requirements in the core, providers can maximize their savings by deploying core routers that require less space and power to deliver the same service capacity. Figure 4 compares the space and power consumption required by an overlay MPLS core, a converged MPLS core solution built with a standard core router and a converged MPLS core that uses a high-capacity core router that requires 20% of the space and less than 50% of the power to deliver the same services.

Figure 4: Service providers can reduce costs by implementing core routers that make more efficient use of central office space and power supplies.

 

Clearly, taking MPLS feature support, scalability, space and power consumption into account when choosing MPLS core equipment can dramatically increase the economic gains service providers realize from migrating to MPLS.

Service providers who already deliver comparable levels of ATM and IP services or who are facing surging demand for their IP services can benefit from consolidating their Layer 2 and Layer 3 service networks onto a converged MPLS core. But to realize those benefits, providers must look for MPLS solutions that implement an integrated network management platform and high-capacity, efficient core routing equipment. Implementing MPLS using a piecemeal network management approach or multiple low-capacity core routers can be worse than not implementing MPLS at all.

Visit Lucent Technologies online.