Operations expense drives connection-oriented Ethernet design choices

Michael Kennedy, Network Strategy PartnersCarrier metro networks are migrating from SONET/SDH transport to connection-oriented Ethernet designs as new business models are emerging built upon residential triple play, Ethernet mobile backhaul, and enterprise Carrier Ethernet services.

Connection-oriented Ethernet represents yet a further evolution of Ethernet to transform it from its original form as Local Area Network technology to one suitable for service provider networks. The focus of a great deal of work by systems vendors, service providers and standards organizations is to make Ethernet into a scalable, manageable, and engineered transport technology. Key objectives of these efforts are to standardize operations, administration, and management (OAM) to give Ethernet some of the management capabilities of SONET/SDH and to introduce traffic engineering to effectively manage service transport and network resource allocation.

Adding transport technology characteristics to switched Ethernet provides the technology foundation to combine Ethernet switches and transport systems into systems with integrated functionality. Many capital and operations expense benefits derive from reducing the number of chassis and O-E-O (Optical-Electrical-Optical) interfaces as well as enabling control of QoS (Quality of Service) and deterministic traffic management.

Expenses are determined primarily by the working hours incurred by the "hands-on" technicians who make physical changes to the connection-oriented Ethernet systems

Several new standards and additions to existing standards including tag switching, MPLS-TP, T-MPLS, and PBB-TE are employed to implement connection-oriented Ethernet. While all the approaches deliver connection-oriented Ethernet functionality, their impact on operations expense differ. Important operations expense categories include network upgrades and patches, network care, training, network management systems, and OSS system integration. Expenses are determined primarily by the working hours incurred by the "hands-on" technicians who make physical changes to the connection-oriented Ethernet systems, and by Tier 1 and Tier 2 engineers located in the network operations center (NOC). There is some cost disadvantage for the MPLS-based approaches in those engineers and technicians with MPLS skills generally command higher labor rates than the more generic skills required to manage and maintain Ethernet approaches.

Network upgrades and patches are required as changes are made to systems software. The changes are made in the NOC and consequently involve Tier 1 and Tier 2 engineers. More mature standards such as tag switching and MPLS have an advantage over the newer standards because fewer changes are made each year. Newer technologies undergo a growth period where many alternative additions to standards are proposed and consequently upgrades and patches are more frequent. 

Network care includes configuration, fault and performance management. Tag switching enjoys a cost advantage in this expense category because it is a simple approach and NOCs generally already posses staff with the required skill set.

Test and certification activities include test case development, automation and execution. Test and certification expenses are lower when there is less novelty associated with the technology--this favors tag switching and MPLS-TP.

Training expense is highly dependent upon the service provider's work force and architectural approach to connection-oriented Ethernet.

Training is required for all skill groups and includes training when the technology is introduced to the network and training updates in subsequent years as the standards are updated and modified. Training expense is highly dependent upon the service provider's work force and architectural approach to connection-oriented Ethernet. For example, MPLS-TP approaches are more common when connection-oriented Ethernet is implemented in Ethernet Switch Routers. Service providers already committed to MPLS networks therefore will already have staff in place with strong MPLS backgrounds. Many smaller service providers are more likely to already use Ethernet switches employing tag switching in their networks so in those cases tag switching and PBB-TE approaches will be more familiar to the network operations staff. Connection-oriented Ethernet also is finding favor in service provider's transport organizations. Transport organizations will have some understanding of Ethernet technology but little or no understanding of MPLS which involves knowledge of routed networks.

Finally, larger service providers use large and complex OSS (Operations Support Systems) that integrate network operations, customer service, sales, and billing systems into full business process solutions. New technologies such as connection-oriented Ethernet must be integrated and tested for interworking with the OSS systems. Work activities include business process design and OSS software specification, coding, test and integration. This can be expensive as well as require long lead times because the necessary ITT resources usually face large backloads of systems integration projects. Technologies already fully or substantially integrated into the existing operation will enjoy a clear advantage over those that must make a cold start on this process.

The several approaches to connection-oriented Ethernet each address the basic requirements of providing Ethernet with some SONET/SDH-like OAM capabilities and the ability to apply traffic engineering principles to the service transport and network resource allocation. The choice of best approach therefore will come down to the skill-sets of the network operations organization and technology's impact on network operations expense.

Michael Kennedy is a regular FierceTelecom columnist and is the co-founder and Managing Partner of Network Strategy Partners, LLC (NSP) management consultants to the networking industry. He can be reached at mkennedy@nspllc.com.