Adding intelligence to Ethernet access
By MATTHEW SQUIRE, Hatteras Networks--Upgrading Ethernet to carrier class requires a look at its ability to shed light on last-mile network traffic.
Upgrading Ethernet to carrier class requires a look at its ability to shed light on last-mile network traffic.
By MATTHEW SQUIRE
A costly limitation of Ethernet access platforms is their lack of intelligence, particularly with regard to carriers' visibility into the final traffic handoff from their services network to the enterprise network. When questions arise, carriers must send Craft personnel to the customer premises to isolate and troubleshoot problems, even though the fault may lie with equipment or circumstances outside the carriers' realm of responsibility. Enabling carriers to extend visibility to the point in the network where their traffic originates or terminates allows providers to clearly determine whether network issues stem from problems on their own infrastructure or that of the customer's.
With carriers increasingly focused on improving top-line revenues and bottom line expenses, providing technology that can address these issues is critical to improving margins and decreasing network management costs. The disparate nature of the access network, with facilities based in central offices, remote terminals, and controlled environmental vaults, further compounds the problem. The management solution requires a complete set of monitoring, diagnostic, and troubleshooting capabilities spanning all locations. In addition, it must operate in a seamless fashion with existing networking equipment, and within regulatory restrictions.
Perhaps, the biggest challenge for the vendor community is to provide carriers with an economic model for deploying last-mile Ethernet services over their existing networks. An important step in this process is the development of intelligent network demarcation points that can extend the carrier's knowledge of their network traffic to the edge of customers' networks. The cost of this demarcation or network interface device must be tightly managed, which means designers are faced with the dilemma of trading off intelligence to lower costs. Finding the optimal balance between the two is paramount for wide-scale deployment of Ethernet access services.
Delivering a comprehensive set of Ethernet services with requisite service level agreements, requires an intelligent network interface device or unit (NIU) to serve as the service demarcation point at the edge of the customer's network. A critical component of the carrier-class Ethernet access system, the NIU also defines the edge of the carrier's management domain. Once in place, the NIU serves an important function by gathering local performance and error data on both the access loop and the customer's network. This performance data is made available to the network operator through the Ethernet access system's element management system (EMS). Ideally, the NIU serves as an extension to the head-end device. Data from the NIU is continuously correlated with data from other network elements so that operators can predict and diagnose faults within the access network.
Integrating any new technology into an existing network infrastructure can be costly and time consuming. To ensure a smooth transition, carriers require technologies that do not mandate significant changes to their existing management and operational procedures. One of the gating factors inhibiting deployment by carriers is Ethernet's lack of carrier-class OAM (operation, administration, and maintenance). Most agree that Ethernet will not be widely deployed until it offers similar OAM capabilities to other access service technologies (ATM and TDM) therefore upgrading Ethernet to include this functionality is a top industry priority.
The Ethernet in the First Mile (EFM) taskforce is working to develop IEEE 802.3ah, an EFM standard, which includes a set of Ethernet OAM capabilities to address the needs of large and small carriers. These capabilities provide the basic tools for carriers to monitor, diagnose, and troubleshoot first-mile Ethernet access links.
Welcome to the real world
The work of IEEE 802.3ah is only the first step, however. In reality, networks are much more complex than individual last-mile network links, and require functionality that is over and above the scope of the work within IEEE 802.3ah. Other standards bodies, such as the Metro Ethernet Forum (MEF), Internet Engineering Task Force (IETF), and International Telecommunications Union (ITU) are pursuing additional standardization efforts to address network-wide OAM issues, and higher-layer problems such as service provisioning.
Network-wide Ethernet OAM functionality will provide visibility into all aspects of the network's performance and operation, from remote centralized locations. Any proposed approach must support industry-standard interfaces already widely used by carriers-transaction language-1 (TL-1), remote monitoring (RMON) and simple network management protocol (SNMP).
With legacy access networks, truck rolls are often required whenever the customer elects to change any aspect of their services, from logical connectivity and service type to bandwidth. This type of provisioning not only lengthens time-to-service and erodes customer confidence because changes are slow, it delays time-to-revenue.
The plug-and-play capability of Ethernet enables carrier-focused equipment vendors to build systems that can auto-sense and provision new services, as well as re-route existing services in the event of a network failure. This allows carriers to remove one of the most time-consuming and costly steps in provisioning services to their business customers.
Additionally, by enabling real plug-and-play service when connected to any access port in the network, carriers can reduce the time it takes for Craft personnel to turn up services at the customer site. Ethernet also provides true service mobility for the carrier - if the assigned services need to be allocated to a different port, the Ethernet access service platform automatically recognizes the service and re-provisions the existing service correctly.
Given the cost and service delays associated with troubleshooting problems within the last mile, carriers want to leverage existing methods of providing early warning of failure for a system that may be degrading. SONET includes a complete suite of reporting mechanisms under the OAM umbrella. The IEEE is currently working on methods to compute bit-error-rate metrics for Ethernet links, including measurement of low-level errors to give an indication of the overall health of a physical or logical link. By addressing problems as they begin to occur, and by detecting link degradation instead of link failure, carriers can better maintain the overall health of their last-mile networks and ensure service integrity. This is a win-win situation for carriers and their customers.
Revenue protection and migration
When developing new platforms for the last mile, equipment vendors must consider a fundamental challenge facing carriers today--how to protect existing high-margin service revenues while responding to the demand for Ethernet services. According to Vertical Systems, the market opportunity for existing private-line services is expected to reach $14 billion by 2004; carriers are unlikely to deploy a technology that forces them to cannibalize this substantial revenue opportunity. Providing T1/E1 services alongside new Ethernet services allows carriers to maintain their current revenue base while expanding their service capabilities to attract new customers and up-sell new services on the same network infrastructure. This ability provides an attractive value proposition for the carrier in any market, but even more so today given current dynamics.
Emerging Ethernet access systems enable carriers to deliver Ethernet and T1 services along the same light path. Carriers can then migrate T1 services to a fiber infrastructure, as it becomes available, in a manner that is transparent to the end customer. This protects the user's investment in PBX technology and the carrier's investment in its voice network, while enabling an access infrastructure that is optimized for the still-growing data access market.
Finally, network intelligence throughout the last-mile will enable carriers to reduce the costs and improve the serviceability of their access infrastructures, which is paramount in today's market given the renewed focus on the bottom line. Overcoming the lack of intelligence into the Ethernet service traffic as it terminates off of the carrier's network is critical to a carrier's ability to reduce the operational burden of last-mile access networks.
Dr. Matthew Squire is chief technology officer at Hatteras Networks (Research Triangle Park, NC).