IPTV in Carrier Ethernet Networks

The increased service flexibility and economic advantages of Ethernet have driven carriers to steadily replace circuitswitched equipment throughout the network. Continued advancements in Ethernet technology through advanced quality of service (QoS) capabilities, scalability enhancements, and service resiliency now enable Ethernet networks to deliver bundled residential triple-play voice, video, and data services. Service providers can attract new customers with attractively priced service bundles, while relying on Ethernet to increase bandwidth and reduce network costs.

Hierarchical QoS

With triple-play services, QoS must be guaranteed not just at the subscriber level but also for the individual services being offered. Hierarchical QoS provides three tiers of QoS support — port, subscriber, and application — allowing service providers to support real time services over high-speed connections carrying tens of thousands of diverse flows. This eliminates the need for distinct overlay networks such as a residential DSL data network and a separate cable or satellite TV network.

Three tier hierarchical QoS enables service providers to set both a committed and peak rate for each outgoing port (Tier 1) with individual committed and peak rates for each subscriber on the port (Tier 2), as well as up to eight QoS levels per subscriber (Tier 3) to support individual triple-play services. In order to simplify provisioning, residential subscribers typically choose from a set of pre-defined service profiles. An example would be a residential subscriber selecting a triple play package of two IPTV channels, one of which supports HD, two VoIP lines, and 2Mbps peak download data service.

New hardware based hierarchical QoS capabilities are just now being deployed in service provider networks. These enable scaling up to 40 times larger than traditional implementations with the ability to set SLAs (Service Level Agreements) down to the individual flows. For example, at a larger central office a single switch may aggregate 20 independent 1GE ports each feeding a DSLAM that aggregates 300 and 500 subscribers, with two 10GE ports providing a redundant network uplink. Such a switch will have on the order of 6-10 thousand incoming subscribers. Without a scalable QoS implementation, switches must group traffic by service class, overprovision bandwidth, and hope for the best. A hardwarebased, three tier hierarchical QoS capability enables an elegant and deterministic triple play network deployment.

Scaling Through MAC-in-MAC

Existing technologies, such as the IEEE 802.1ad Q-in-Q standard, provide isolated tunneling of VLANs, enabling service providers to separate subscriber traffic while providing different levels of QoS. This is an effective solution for smaller deployments, but is insufficient for use in large Metro networks because of VLAN ID space limitations. The recent introduction of the IEEE 802.1ah standard (also known as MAC-in-MAC) builds upon existing Q-in-Q technology to overcome Ethernet scaling limitations. By inserting a provider MAC address into each frame that is transparent to customer data, MAC-in-MAC economically scales to millions of service VLANs, more than 4,000 times the number supported by traditional VLAN and Q-in- Q deployments. MAC-in-MAC also offers seamless interoperability with Qin- Q networks without the complexity or expense of a Layer 3 MPLS deployment. Additionally, further standards work is under way by the IEEE that will build upon MAC-in-MAC technology to further improve traffic engineering capabilities.

Service Resiliency

Subscribers like the idea of consolidating three bills — TV, phone, and Internet — into a single bill. Carriers like consolidation because subscribers are less likely to switch service providers due to the hassle involved in moving multiple services. Unfortunately, the same triple-play capabilities that attract subscribers can also make it easier to drive them away.

With data services, it’s not always clear when a connection is experiencing an outage. Even if a service disruption was a network problem all along, subscribers often perceive that they have fixed the problem by rebooting the computer or modem and conclude that somehow the problem had been on their end. IPTV services are much more likely to impact subscriber churn as it becomes quite clear when the network is experiencing an outage. Subscribers are able to more accurately evaluate the quality of the services they are receiving, making it crucial for service providers to maintain the overall resiliency of their network.

Resiliency for triple-play networks must be maintained at the hardware, software and network levels. Hardware resiliency achieved through equipment redundancy is a fairly well-understood issue. However, this is the level at which many service providers stop thinking about resiliency. Maintaining the highest quality of service and availability requires not just a focus on hardware resiliency but rather a perspective that includes service availability. Just because a connection is up and successfully rerouted, there is still a possibility that there is a disruption to the service being offered over the connection.

With the increased role of software in the network, software-based outages have become more prevalent. The problem is exacerbated by the increasing aggregation of links in high-speed networks since a single application failure can tie up limited processing and memory resources, bringing down other processes and spreading the outage across all aggregated connections.

Achieving software resiliency requires failover mechanisms that mirror hardware redundancy through modularity. The operating system must be broken into modules that can be operated independently. In this way, when a process does fail, the failure can be contained to prevent the outage from spreading to other processes. The operating system also needs to manage the separate processes so that the system can automatically terminate a failed application or process and start a new instance.

The third level of service resiliency is network resiliency. Traditional Ethernet networks do not have the inherent resiliency of circuit-switched networks. In order to increase Ethernet network resiliency, technologies such as Ethernet Automatic Protection Switching (EAPS) are now widely deployed. EAPS is a service-aware, ring-based protection protocol that utilizes a standard Ethernet MAC and provides a carrierclass failover response of less than 50ms. Ring connectivity is verified continuously through the exchange of “heartbeat” messages between neighboring switches.

EAPS offers a number of advanced resiliency features to networks, including primary/backup path designation on a per-VLAN basis, dual-attached and subtended rings, and support of multiple domains of protected VLANs on the same ring, enabling carriers to create highly available platforms that are able to continue service delivery even during near-catastrophic outage conditions. Additionally, because EAPS addresses failures at Layers 1 and 2, recovery often occurs before higher layer protection mechanisms need to kick in, giving the experience of uninterrupted service to subscribers.

The new Ethernet technologies such as hierarchical QoS, MAC-in-MAC and service resiliency bring carrier-class performance and scalability to service provider networks and allow network operators to provide reliable IPTV and other triple-play services while simplifying overall network design. When coupled with the economic benefits and complete control over priority for subscriber applications, it is clear why Ethernet is poised to dominate service provider networks as it has enterprise networks.

Peter Lunk is the Director of Service Provider Marketing for Extreme Networks. (news - alert) For more information, visit the company online at http://www.extremenetworks.com.