By Mark Showalter
The Metro Ethernet Forum’s new MEF 22 implementation agreement paves the way for Ethernet mobile backhaul services–but doesn’t spell out which metro packet network architecture to use. The three primary options are Ethernet transport, MPLS, and VPLS.
The number of mobile subscribers worldwide in 2008 reached 3.9 million, according to Infonetics Research (www.infonetics.com), representing 17% year-to-year growth from 2007. This number is expected to top 6 billion by 2013. This surge in consumer demand for wireless services is driven by a combination of convenience and affordability.
A significant part of the demand is reported to come from Brazil, China, India, and Russia, with China and India adding more than 150 million new mobile subscribers every year. In these countries, demand is driven by the need for basic voice services. In developed countries, the demand is driven by bandwidth-intensive applications such as streaming video. Across the globe, consumers may choose from a variety of sophisticated end-user devices such as iPhones, Blackberries, and other smart phones. These smart phones drive significantly more traffic across the mobile backhaul network. Altogether, both the increasing number of subscribers and the increased demand for data-intensive services are accelerating the demand for increased mobile backhaul capacity.
Historically, mobile backhaul services relied upon T1 circuits delivered over a copper-based access network to connect radio towers in the field to base station controllers. T1 services inherently provide end-to-end clock synchronization, which is important when handing off a call from one radio tower to another as the end user roams. There are several drawbacks with T1 services for mobile backhaul, including cost, capacity, and susceptibility to lightning strikes.
While competition has generally reduced T1 pricing globally, T1 circuits to rural areas are relatively expensive. Since each T1 line delivers only 1.5 Mbps of bandwidth, it doesn’t take long before a wireless operator must provision multiple T1s to a radio tower, each adding significant fixed cost for relatively small amounts of bandwidth. The need to transition from legacy T1 becomes more urgent with the shift toward LTE and WiMAX for next-generation wireless networks.
T1 mobile backhaul is copper based and more susceptible to lightning strikes than fiber-based Ethernet mobile backhaul. A lighting strike that causes a 2-to-3-s disruption in T1 mobile backhaul communications will cause the radio base station to drop all calls in progress and start a resynchronization process with the controller that can last 45 min in some cases.
These trends in mobile networking are driving wireless network operators to seek lower-cost, higher-capacity alternatives to T1 for mobile backhaul. To address this demand, facilities-based metro carriers are deploying new mobile backhaul services that replace traditional copper-based T1 circuits with packet-based networks over fiber-optic facilities.
The Metro Ethernet Forum (MEF; www.metroethernetforum.org) is an industry consortium that standardizes Ethernet service definitions and provides a process for carriers and equipment providers to achieve certification of compliance to these standards. The migration to Ethernet mobile backhaul services (EMBS) from legacy T1 services is facilitated through service definitions the MEF recently created. In February 2009, the forum announced the MEF 22 Implementation Agreement (MEF 22). MEF 22 provides a foundation for the implementation of EMBS using Carrier Ethernet between the radio network controllers (RNCs) and the radio access network base stations (RAN BSs) as illustrated in Fig. 1.
While MEF 22 addresses EMBS requirements for synchronization, legacy mobile backhaul migration, traffic separation, and Ethernet operations, administration, and maintenance (OA&M), it does not define how to architect the underlying Carrier Ethernet transport network. Several alternative metro packet architectures are available to deliver EMBS, and these alternatives share several common elements.
Synchronization is a key EMBS requirement. To address the timing requirements for EMBS, standardization work from both the Institute of Electrical and Electronics Engineers (IEEE; www.ieee.org) and the International Telecommunications Union (ITU; www.itu.int/net) imbue packet networks with the synchronization required for the most time-sensitive applications such as mobile digital video.
The IEEE 1588 Precision Time Protocol achieved standardization in 2002 and brings submicrosecond synchronization between Ethernet attached devices. IEEE 1588 as applied to mobile backhaul provides the advantage of not needing to be deployed on every intermediate node in the network; the tradeoff is that variable packet delays across the network may adversely affect synchronization performance.
The ITU’s G.8261/G.8262/G.8264 is referred to as Synchronous Ethernet (SyncE). SyncE defines a physically connected timing infrastructure similar to that found in SONET/SDH deployments. Because SyncE is not a packet technology, synch performance is not affected by traffic load on the packet network. The tradeoff here is that every intermediate node in the backhaul path needs to support SyncE.
Legacy mobile backhaul migration addresses the consolidation of multiple generations of mobile backhaul including 2G, 3G, and the most current 4G networks into a single metro packet network. Guidelines for supporting varying wireless technologies such as LTE and WiMAX are also provided by MEF 22. When migrating to EMBS from legacy T1 services, carriers can choose between immediately moving to an all-Ethernet backhaul approach and transitioning gradually, which maintains legacy T1 connections for voice traffic while supporting EMBS connections for data traffic. The MEF 8 Implementation Agreement for the Emulation of PDH Circuits over Metro Ethernet Networks defines the requirements that enable metro carriers to offer T1 services while taking advantage of the cost-effectiveness of Ethernet mobile backhaul.
Traffic separation as defined in MEF 22 sets guidelines for the number of service classes to use, a framework for bundling traffic types into a limited number of service class types, and the performance requirements for each of these classes. Traffic separation is important because of the wide range of services that are transported over the metro network, enforcing the traffic prioritization necessary to meet the synchronization requirements between the RAN BS and RAN NC.
Ethernet OAM standards address fault detection, service management, and performance monitoring. Pertinent existing standards include IEEE 802.1ag Connectivity Fault Management (CFM), which enables the service provider to determine whether a particular service is being delivered to a customer (a finer level of granularity than merely detecting a faulty link in the network), and ITU Y.1731 Performance Monitoring, which enables the service provider to analyze the packet loss or delay to validate service performance.
There are several alternative metro network architectures for delivering EMBS. Generally, the approaches may be categorized as metro virtual private LAN service (VPLS), metro Multiprotocol Label Switching (MPLS), and metro Ethernet transport. Each of these approaches enables service providers to offer EMBS as well as services such as E-FTTX to businesses and residential triple-play aggregation.
MPLS has been widely used at the core of IP networks to address complex routing issues. The metro MPLS approach extends this complex set of IP routing capabilities across the metro. VPLS was introduced to add a robust set of Ethernet private LAN services at the edge of MPLS core network services. The metro VPLS approach extends these services to the metro edge by deploying routers with VPLS-optimized silicon at the serving central offices and at the metro core. The metro Ethernet Transport approach primarily relies on Ethernet across the metro, including VLAN translation, IEEE 802.1ad Provider Bridging (PB), IEEE 802.1ah Provider Backbone Bridges (PBB), and IEEE 802.1Qay Provider Backbone Bridges - Traffic Engineering (PBB-TE). Regardless of the approach, the metro network connects to an MPLS carrier core network that extends EMBS beyond the metro for global wireless operators.
While each of these metro approaches can deliver EMBS and other services, selecting the right network architecture involves tradeoffs that affect the total cost of ownership (TCO) over the life of the network. Network Strategy Partners (NSP), management consultants to the telecommunications industry, published a white paper comparing the TCO of metro VPLS, metro MPLS, and metro Ethernet transport approaches for residential triple-play, business Ethernet, and wireless backhaul services. The paper shows the TCO for Ethernet transport is 61% and 63% lower than for the metro VPLS and metro MPLS alternatives, respectively.
As reported in the white paper:
“The VPLS and MPLS alternatives have much higher capital expense than the Ethernet Transport alternative. This cost difference is due to the VPLS and MPLS alternatives use of IP/MPLS embedded features in their line-cards as compared to the widely deployed Ethernet L2/L3 features of the Ethernet Transport line-cards. The volume of production of Ethernet line-cards is two orders of magnitude greater than that of the VPLS and MPLS line-cards. Consequently Ethernet Transport vendors can profitably support price levels that are less than half of those of the highly specialized (and relatively low volume) VPLS and MPLS line-cards.”
Additionally, NSP reports that VPLS and MPLS switches have lower 10-Gigabit Ethernet (10GbE) port densities, thus requiring more switches to satisfy the same amount of metro traffic.
The NSP report shows the top contributors to operational expenditures (opex) include service contracts, engineering/facilities/installation, network care (which includes the cost of provisioning, monitoring, and maintenance of the network), training, and testing/certification expenses. Service contracts are the biggest contributor to opex, typically costing 10 to 15% of the capital expenditures. Other opex savings accrue from lower power consumption and subsequent reduction in heating, ventilation, and air-conditioning costs.
Figure 2 depicts an EMBS reference architecture representing the metro Ethernet transport approach. It comprises an Ethernet distribution network that connects multiple ring aggregation points-of-presence (PoPs) to the metro core PoP using multiple 10GbE links. The Ethernet access network connects the RAN BSs to the ring aggregation PoPs with multiple 1-Gbps and/or 10-Gbps Ethernet access rings. Ethernet transport switches are placed at the RAN BS and provide the first level of aggregation in the metro network, while the next level of aggregation switches are deployed at the ring aggregation PoP, collecting traffic from multiple access switches. This architecture gracefully scales from hundreds to tens of thousands of Ethernet UNIs per metro core PoP.
This topology enables logical Ethernet access protection switching rings (as described in RFC-3619) to be configured from the metro core PoP through the ring aggregation PoPs to the Ethernet transport switches that serve the RAN BS. Such an approach supports 50-ms recovery from faults between the RAN BS and the metro Core PoP.
As the demand for wireless services increases, metro operators will see increasing opportunities for the deployment of EMBS. Next-generation metro network architectures enable metro carriers to deliver an array of packet-based services across a common metro packet infrastructure, maximizing return on investment.
Mark Showalteris director, service provider marketing, at Extreme Networks (www.extremenetworks.com).
Network Strategy Partners: Total Cost of Ownership Comparison: Ethernet Transport vs. VPLS and MPLS in the Metro Network
Lightwave Online: Verizon Opens up Fiber-Optic Network to Wireless Carriers
Lightwave:S. Korean Wireless Backhaul Evolution Provides Lessons