Ethernet FTTH Triple-Play Services

Rapid advances in technology are opening the door to an expanding array of residential services under the Triple-Play banner – which is boosting competition as incumbents, cable MSOs, competitive carriers, and ISPs trespass onto each others’ turf and battle over subscribers.

Intense competition will require all these players to develop business processes and more intelligent networks that make them highly adaptable to change and more responsive to customer demands. And that means Ethernet – the base on which triple-play will be built – will have to raise its game.

To be successful, service providers will require:

Fortunately, new, higher-performance Ethernet access, aggregation, and service management platforms are becoming available that enable operators to deliver FTTH (fiber to the home) triple-play services successfully. This report, which meshes with the earlier reports such as PON & FTTx Update and Handling Broadcast IPTV Content, looks at key requirements and how Ethernet systems can handle them.

Here’s a hyperlinked contents list:

Webinar

This report is based on a Webinar, Using Metro Ethernet to Deliver FTTH Triple-Play Services, moderated by Stan Hubbard, Senior Analyst, Heavy Reading, and sponsored by Brix Networks Inc. , Extreme Networks Inc. (Nasdaq: EXTR), PacketFront AB , and Riverstone Networks – now part of Lucent Technologies Inc. (NYSE: LU). An archive of the Webinar may be viewed free of charge by clicking here.

Related Webinar archives:

— Tim Hills is a freelance telecommunications writer and journalist. He's a regular author of Light Reading reports.

Next Page: FTTH Triple-Play Requirements

Figure 1 highlights key FTTH triple-play requirements from the standpoint of the subscriber and the service provider.

On the subscriber side, consumers will demand a quality experience. According to Gary Holland, Senior Manager, Carrier Ethernet Marketing for Lucent Technologies Inc. (NYSE: LU), in triple-play every user is a network monitor.

“If a network glitch causes slight pixelization during a football game, a subscriber might find it a bit annoying,” he says. “But if a network outage or excessive pixelization causes our fan to miss the winning goal, then phones start ringing at the service provider’s customer service center, and the possibility of subscriber churn becomes a reality.”

So networks have to run flawlessly, with hiccups corrected in a timely manner. Consumers also are looking for triple-play service availability wherever they live, and they want to have a multitude of service and pricing options.

On the service provider side, everyone is focusing on ways to increase ARPU (average revenue per user) while controlling capital and operational costs as their subscriber bases grow. Key requirements here break down into four general areas:

Underlying all these requirements, of course, is the availability of lots of subscriber bandwidth. But just how much is "lots"? The question is important to triple-play, because different technologies have different bandwidth limitations, and the business/technology tradeoffs become complex and difficult.

Unfortunately, no one knows yet, and discussions can easily spin off into how many gigabits per second would be needed to support 3D life-size holograms for full-immersion interactive gaming and the like. More modestly, something like the following might be plausible during the next decade for a heavy-use home with multiple users/channels active simultaneously:

While much of this will be downstream bandwidth, the upstream bandwidth will still be substantial in applications like video gaming, peer-to-peer uploading, and so on.

Next Page: Active & Passive FTTH Ethernet

For Ethernet FTTH-based triple-play the basic access technology choice for service providers seems to be between active Ethernet and one of the various forms of passive optical network (PON). Although PONs began years ago as ATM-based systems, the more-recent standards support Ethernet. (See PON & FTTx Update for further details of PON standards and technologies.) Figure 2 illustrates the general picture of these two approaches to FTTH Ethernet.

Active Ethernet has dedicated subscriber bandwidth and active outside plant using low-cost Ethernet components. The subscriber connects typically over a 100-Mbit/s Ethernet connection to an Ethernet aggregation switch, which is connected in turn over Gigabit Ethernet to an Ethernet router.

In contrast, a PON uses non-powered optical splitters/combiners that allow a single fiber run to serve multiple subscribers – typically up to 32, but sometimes up to 64. This basically broadcast-like mode of operation means that the available bandwidth is shared among subscribers, giving useful bandwidths typically between 19 and 38 Mbit/s per subscriber. The fiber is typically terminated in the subscriber’s home by an optical network unit (ONU), connected by an optical splitter to the optical line terminator (OLT). The OLT is connected to the switch (ATM or Ethernet), and the switch connects into the service provider’s active network.

Both approaches have pros and cons. For example, active Ethernet needs powered outside plant, such as local aggregation switches located in street cabinets, and the local aggregation switches also need to be managed. However, it does act essentially as carrier-grade Ethernet, with the usual attributes of bandwidth flexibility, long range, fault isolation, security, and the like. Vendors are working to reduce power requirements in an effort to lessen that issue – for example, by eliminating equipment cooling fans.

PONs have big the advantage of using passive outside plant, but subscriber isolation is reduced, and there are range restrictions and security issues. It is also harder to perform fault isolation and management on the passive components.

Active Ethernet remains a rare bird in the U.S., though it's being used by the Utah Telecommunication Open Infrastructure Agency (UTOPIA) . But it is being deployed in various European countries – Germany, Finland, and the U.K., for example. PONs remain the firm favorite in the U.S. for incumbents.

Aggregation, VLANs & MPLS

If a service provider plumps for active Ethernet, an interesting question is whether to extend MPLS from the network core into the access network to some point near to (but not at) the service subscriber. In principle, this would provide a common QOS, traffic engineering, redundancy, and scaleability capability across as much of the service infrastructure as possible.

“Most service providers are making do with traditional VLAN technologies in their access networks, with all of the well known limitations that that implies,” says Lucent’s Holland. “The main reason is cost, but it is the perceived cost and complexity of MPLS that service providers are thinking about. In reality, using MPLS across the service provider infrastructure – including the access part of the network – makes much better sense.”

Table 1 gives a summary comparison on some VLAN and MPLS characteristics, while Figure 3 shows the basic idea of using MPLS to provide an end-to-end carrier-Ethernet service infrastructure. The notion is to combine a consistent end-to-end operational model and end-to-end management with flexibility in network design.

Table 1: VLAN & MPLS Compared for FTTH Triple-Play Services

However, it may not be necessary to extend MPLS (which does increase complexity) all the way to the access edge if a service provider implements an Ethernet aggregation architecture using the latest enhancements to Ethernet technology.

“A carrier-class Ethernet switch providing aggregation for both active Ethernet and PON access technologies, and providing a hierarchical QOS capability to each subscriber and to each of the subscriber’s applications – we have those capabilities in Ethernet switches today,” says Peter Lunk, Senior Product Manager at Extreme Networks Inc. (Nasdaq: EXTR). “The switch then passes the data on to the appropriate service network, using a low-cost Layer 2 approach that is much more economical than extending MPLS out to the edge.”

But much will depend on the provider’s particular circumstances. If, for example, it has already heavily invested in an MPLS core, extending MPLS into the edge may be an effective way of leveraging that investment.

However, for providers that have not made that investment, there is new technology called "MAC in MAC," which provides an alternative to MPLS with an Ethernet approach. This is the 802.1ah Provider Backbone Bridges standard, which allows Ethernet networks to scale to millions of VPNs while maintaining all the economic advantages of Ethernet. Figure 4 shows the basic principle.

For more on carrier-class Ethernet issues in an FTTH context, see the Light Reading report, Handling Broadcast IPTV Content.

Next Page: Handling FTTH Ethernet Services

Whether a service provider uses active or passive Ethernet approaches, FTTH Ethernet throws up some serious issues in actually handling the end services, such as IPTV, VOIP, and high-speed Internet. These include, in particular:

Complexity and Cost

Introducing multiple services over a single network often results in a great increase of complexity for service providers, and thus considerable increase in costs. And in an open environment or one in which the provider is offering wholesale services, introducing multiple service providers onto a network makes that complexity even greater.

All this doesn’t look good for cost efficiency, so networks are going to have to get smarter.

“Automation is key to cost efficiency – it is absolutely required,” says Stefan Gustafsson, PacketFront AB 's VP of Product Management. “For example, you can change the service profile in one place, and that will be replicated out to all the network devices, handling QOS and so on.”

He gives the following as key aspects of network and service automation in Ethernet FTTH:

Automated element control

Automated service control

Automated subscriber control

Service Assurance

Ethernet FTTH assumes a provider business model based on flexible, competitive service bundles. As pricing can go only so far in sustaining competitive advantage, providers must also compete on service quality. And this must be done for each element of the service bundle, or the provider risks losing the whole bundled-service subscription because of dissatisfaction with just one element. Service convergence is not necessarily a one-way bet for providers.

“Service assurance starts with gaining network and service performance visibility at the MPLS and IP core, and really extending that performance monitoring into the access network and even all the way to the customer or digital home,” says Bob Travis, Director of Product Marketing, Brix Networks Inc. “This really provides the network operator with a consistent and accurate view to measure and monitor all the relevant QOS metrics related to the converged service.”

According to Travis, an area of great interest to service providers is gaining deeper visibility of service performance into the digital or triple-play customer’s home. He points out that a self-service portal can also work for the service provider as a way to measure its customers’ service experience.

“Essentially, the customer would download a Java thin-client agent and collect information displayed as a service quality report per that customer,” he says. “Service providers could use this information to shorten support call lengths, for example. They could also use the information for service sign-up prequalification capabilities, and also provide some installation verification and some deeper intelligence into troubleshooting, if there are some problems with that customer.”

Network Troubleshooting

If there is a service problem caused by the network itself, the spotlight turns on Ethernet’s troubleshooting capabilities – and, in carrier terms, until fairly recently these have left much to be desired. However, carrier Ethernet is advancing, and the 802.1ag Configuration Fault Management standard offers Ethernet ping and traceroute capabilities to simplify the troubleshooting of large Layer 2 networks.

Figure 5 shows the different maintenance domains defined in the standard. This includes an end-to-end capability where the Ping and Traceroute can be used by the customer – as well as the provider. There is a provider domain that can span multiple operators, and there is also an operator domain for troubleshooting within those specific service provider networks.

“Adding these kinds of capability is really the result of lots of customer requirements – people asking for easier tools – and the standards bodies have stepped up and delivered some of these, adding them to Ethernet,” says Extreme’s Lunk.

Next Page: Ethernet FTTH Architectures

To pull some of the previous considerations together, Figure 6 shows a typical active Ethernet FTTH architecture that would be used to deploy triple-play services. On the left are subscribers with home gateways connected via, typically, 100-Mbit/s Ethernet over fiber to the carrier Ethernet access router, which is usually located in a neighborhood street cabinet. To satisfy the need for power, equipment can be located in street cabinets, building basements, or even on pole mounts.

The access router located in the neighborhood street cabinet is typically connected into the Ethernet aggregation network by Gigabit Ethernet. That Ethernet aggregation network would be connected over an MPLS core to a central POP, on the right of Figure 6.

Gigabit Ethernet aggregation means that distances can be fairly substantial, covering several cities. The Utopia network, for example, covers 14 cities in Utah, with a total distance from north to south of 100km.

At the central office, there are B-RAS and Radius DHCP servers, and also several routers to provide access to the Internet and services from different ISPs (shown in Figure 6 by the brown and orange lines connecting to ISP A and ISP B). So subscribers have access to different services from different service providers.

There is also a local TV headend, which provides connectivity to broadcast TV and collocated video service to provide access to VOD services, illustrated by the red and green lines. Finally there are collocated VOIP servers to provide access to the PSTN, illustrated by the blue line.

Within this architecture, subscriber requests are typically mapped to what are called per-service VLANs, with each per-service VLAN assigned a class of service, QOS, and bandwidth appropriate to the type of traffic it is handling. All this is taken care of by the underlying MPLS network.

Each per-service VLAN is then typically mapped into a Virtual Private LAN Service (VPLS) instance by the carrier Ethernet router at the central POP, and this VPLS instance is mapped across the MPLS core to the carrier Ethernet access router in the FTTH network, where it is delivered to the appropriate subscriber. VPLS is based on MPLS, and so can ensure the desired end-to-end characteristics for the FTTH service.

“In this type of network, multicast has a major role to play, and there are different architectures that can be used to deliver a multicast service. In this particular example, it is the carrier Ethernet routers that are handling the multicast to deliver the service to the subscribers,” says Lucent’s Holland. “In other networks it is quite possible to have a Layer 2 delivery network and then overlay that with a Layer 3 multicast capability. In either scenario, it is the underlying infrastructure, which is based on end-to-end MPLS, that is necessary to deliver services with the necessary QOS and so on.”

However, despite the huge drive towards FTTH, PacketFront’s Gustafsson points out that practicalities will keep the access environment mixed for a considerable time.

“Sooner – rather than later – you will run into time-to-market situations where you want to reach subscribers faster than you can pull fiber. So it is of utmost importance that the solution – hardware and software provisioning and the rest of the network – handle different types of access media. Those might be ADSL, ADSL2+, Ethernet over coax, wireless, power-line communications, and so on. And that network owner has to provide the same look and feel for all subscribers, regardless of where they enter the network.”

This sort of hybrid situation is shown in the following figures. Figure 7 shows the case of multiple access technologies being aggregated, while Figure 8 shows a deployment of FTTH mixed with fiber to the building (FTTB). The FTTB access equipment is placed in basements or in street cabinets and is attached in a ring design direct to the aggregation part of the network.

Gustafsson points out that FTTB deployed like this provides over-allocation and is a cost-effective way of giving redundancy. How many devices are placed in the ring depends essentially on the level of over-allocation on the fiber to the aggregation point.