MPLS helps deliver premium services over Ethernet

Two of the strongest trends in networking today are the migration of Ethernet from local area networks (LANs) to metro area networks (MANs) and the enhancement of IP networks with Multiple Protocol Label Switching (MPLS) technology. Most service providers are familiar with either or both of these developments. But they may be less aware of the potential stemming from the combination of the two, particularly for offering premium service level agreements (SLAs), says Tim Wu

For service providers building out Ethernet-based networks, or thinking of doing so, the available return on investments is an important question. The kinds of SLAs a service provider can offer are crucial to these returns. First, SLA capabilities define the scope of potential customers, as certain customers simply will not purchase services below certain standards.

Second, the difference between profit and loss on individual deployments may depend on the ability to charge more for services, such as transparent LANs, backed by premium guarantees of network performance.

For Ethernet carriers, MPLS is the key to the connection-oriented capabilities that translate into premium service level capabilities. In the long run, MPLS will provide Ethernet service providers a chance to reach those customers once removed and improve overall profitability.

MPLS and Ethernet

Interestingly, MPLS and Ethernet technologies come from opposite ends of the network. Ethernet comes, of course, from the LAN, while MPLS has its origins in the IP core. Metro networks mark the first and only point of direct interaction between the two.

Ethernet is by far the most highly deployed network type on earth, with over 250 million Ethernet devices already connected. With its domination of local area networking secure, Ethernet has most recently taken a well-publicised leap into metro networking.

Metro service providers take the same simplicity and huge bandwidth capabilities (10 Gbps products are now available, 100 Gbps products are expected within two years) and add long distance capabilities (40-70 kilometres) and the high-performance switch-routing equipment suited to carrier environments. The result is a potent combination of speed, scalability and operational simplicity that has won many converts.

Leading the deployment of Ethernet metro networks are various forms of metro service providers (MSPs), including the EtherLECs and building LECs (BLECs) companies like Telseon, Intellispace, Yipes and others. But established carriers are now also looking to metro Ethernet, particularly as a data-focused supplement to existing Synchronous Optical Network (SONET) and Time Division Multiplexing (TDM) networks.

MPLS has its origins in the opposite end of networking, the network core, where its first deployments (by such industry heavyweights as Global Crossing and UUNET) are found. MPLS has thus far been deployed to address various backbone network problems, such as integrating IP and ATM networks, reducing IP router overhead and solving route propagation problems.

Yet, since MPLS generally adds connection-oriented, path-switching capabilities to IP networks, it is an extremely versatile technology. In the metro, its most obvious utility comes as an instrument for creating guaranteed and secure service capabilities.

How the two work together

In many ways, MPLS and Ethernet are an ideal match. Ethernet is well-known for its simplicity and comfort-level with IP data traffic. Ethernet is also known, however, to lack sophisticated quality-of-service (QoS) capabilities, which may be the key to its low cost, but is also a major shortcoming.

MPLS improves on Ethernet not by changing what is good about Ethernet, but by adding desired capabilities at a higher layer (what some call Layer 2 1/2). When MPLS and Ethernet are used together, Ethernet is called to focus on only what it does best point-to-point transport. MPLS, meanwhile, adds the connection-oriented capabilities that are tools for creating services-backed by SLAs.

Service level agreements

Service level agreements have come into the forefront with the increase of network outsourcing and the greater importance of inter-office networks. SLAs, from a customer’s perspective, are crucial to optimal performance of the enterprise network and mission-critical applications. From the service provider side, the ability to offer SLAs both sets the scope of possible customers and marks the critical point where network features translate into additional revenue.

The phrase ‘service level agreement’, refers to a contractual agreement between the service provider and customers, wherein the provider guarantees a certain level of service. Obviously, the guaranteed level of service can differ from customer to customer, as does the amount charged for such service levels, thus creating a differential pricing strategy for service providers. Traditionally, typical subjects of negotiation are:

• Network availability (minimum guaranteed)
• Path/PVC availability
• Average round-trip network delay
• Effective throughput (network performance, burst allowance)
• Service restoration time
• Billing methods (per usage or flat rate).

Over the last few years, SLAs have come to include more of a focus on application performance and other end-user metrics, such as:

• Application availability
• Application response time
• Performance in certain time windows.

Many customers simply will not buy services that do not come with certain levels of SLAs. If a transparent LAN or externally hosted application is mission critical, one can expect a service offering without a certain level of SLA to be a non-starter. SLA capabilities are, therefore, a direct determinant of a service provider’s potential customer base; hence the importance of premium SLA capabilities.

Using Ethernet to create transparent LAN services

Ethernet is used in native form to offer transparent LAN services (TLS) backed by a SLA, but the result is less than ideal. Most of the problems arise because Ethernet is a Layer 2 technology ideally suited to handle point-to-point connections or LAN broadcasts.

Although carriers can currently offer SLAs over Ethernet networks, these SLAs have significant shortcomings, including lack of bandwidth reservation, slow recovery from network failure and scalability limits. Transparent LAN services using 802.1q-based virtual local area networks (VLANs) illustrate these shortcomings.

First, it is difficult for the service provider to offer a guarantee of network throughput using native Ethernet. Second, native Ethernet provides less-than-ideal resiliency features. The gold standard in this area is SONET, with its guarantee of greater than 50 millisecond recovery time from fibre cuts. This feature accounts for the ability of service providers to achieve five-nines reliability and sell SONET-based services at a premium.

In contrast to SONET, Ethernet relies on spanning trees to provide network resiliency. Unfortunately, spanning trees are not a good answer for metro area networks. They can take over 30 seconds to recover from an outage, fibre cut or switch failure. Rapid spanning tree technology is an improvement, but still taps out at about a full second for reconfiguration.

Finally, the Ethernet model comes with profound geographic limitations. The TLS services will be limited to the single network in question. A service provider, ideally, would like to offer SLAs, not only within a single Ethernet network, but across wide area networks (WANs), as well. This ability allows a service provider to generate revenue from customers connecting offices across the country.

The path to long-term profitability

Service providers can create a long-term profit model by implementing MPLS technology. Consider a moderately successful mid-sized company earning an 8 percent operating margin on revenues of $250 million and a subscriber churn of 20-percent-per-year.

Adding MPLS-enabled customised services can quickly lead to substantial gains. The provider can offer premium-level TLS and other services, along with nationwide and worldwide connectivity, producing a 3 percent growth in new subscribers. In addition, targeting high-churn customer segments with customised MPLS-based services may reduce churn by 3 percent. And, migrating existing customers to premium SLA packages can reasonably be expected to produce a 2 percent improvement in average revenue per subscriber.

Collectively, these modest improvements will deliver an 8 percent boost in revenue. Conservatively assuming that the incremental services revenues carry a 50 percent operating margin, then the overall impact of this program will be a 50 percent improvement in operating profits.

The ‘Ethernet Revolution’ is well underway. But it will take MPLS and similar technologies to add to the enthusiasm and stronger profit models for service providers.

www.ispworld.com