Home > Telecoms > NGA discussion topics > EALA > Ethernet in the Access Study

Background and Context to the Study

As a part of the discussion on the key characteristics of Ethernet Active Line Access (ALA), Ofcom commissioned a report from consultants CSMG to study the use of Ethernet in access networks from around the world. The purpose of this study was to identify and research a number of international case studies where Ethernet access is available and to compare the technology in these products with our requirements of Ethernet ALA. With the outcomes of the study, we wished to:

1. understand the technology deployed in these Ethernet networks
2. understand how Ethernet is used, and
3. put our requirements of Ethernet ALA into context, against other Ethernet products

Ethernet ALA is a set of technical requirements which enables Communications Providers (CPs) to offer competitive services to customers through a wholesale bitstream product that is as close as possible to the infrastructure. It can be used where it is not practical for shared passive access to the physical network. As an Ethernet bitstream product, Ethernet ALA should provide CPs with scope for innovation and control, as they would be able to do with access to the infrastructure.

The five key characteristics of Ethernet ALA are:

• Flexible range of aggregation points, as the most economical point to aggregate traffic will vary between communications providers.
• Support for flexible Customer Premise Equipment (CPE), which may determine the consumer experience
• Multicast, for the economic distribution of audio-visual content
• Quality of Service, for prioritising traffic where appropriate, in order to provide an acceptable user experience to customers
• Security, to ensure that data is carried safely over this network

This Ethernet in the Access study reports on seven case studies from around the world. These networks use a range of infrastructure technologies to deliver high speed access for residential and business services. The table below outlines the Ethernet networks in the case studies, the type of services offered and the infrastructure that was used.

Network	Country	Business or Residential Services	Network Technology/ infrastructure
Colt	Europe	Business	Ethernet in the First Mile/Copper
Optimum Lightpath	USA	Business	Fibre Rings/Fibre
Telus	Canada	Business	Point-to-Point/Fibre
iliad	France	Residential	Point-to-Point/Fibre
BBned	Netherlands	Residential/Business	Point-to-Point/Fibre
KPN	Netherlands	Residential/Business	Point-to-Point/Fibre
NTT	Japan	Residential/Business	Passive Optical Network and Point-to-Point/Fibre

The service characteristics of each network were compared against the ALA requirements, to better understand the ways in which the key characteristics of Ethernet ALA could be met. Other capabilities of the networks were also noted, to provide more information about the different models of Ethernet use in the access network.

Summary of Study

The research found evidence of all five characteristics of Ethernet ALA in the case studies. Some of these characteristics, such as flexible aggregation and Quality of Service, were better supported than others. Security provision, especially where point-to-point networks were rolled out, was also strong. The other requirements, of multicast and the support for flexible CPE, were also present in these networks, but take up of these options were more limited.

The differences between the networks can be related to the choice of infrastructure technology and the type of customers served by the network. Generally, there were differences between residential and business products. However, as the networks studied all support the five key characteristics in some way, it also shows that Ethernet ALA could be met by different infrastructure and network technologies.

This table outlines the case study findings in the support of Ethernet ALA requirements.

• Aggregation and Interconnection

Flexible aggregation is strongly supported in these case studies. Many networks provide a range of options for aggregation, from local aggregation to national points of interconnect. Some networks, such as Telus, also promote multiple aggregation points for resilience, at two, or even three, locations. Although a 1GE handover is the standard interface a range of interconnection bandwidth is available, some networks offer handover at 100Mbps and 10GE as additional options.

At the aggregation point, data streams from different CPs are managed by using VLANs. Traffic from residential customers is classed per Class of Service and where VLANs are managed across an NNI they are allocated per end point. These are managed by using Q-in-Q or label stacking.

• Flexible support of CPE

The choice of infrastructure in the network architecture may affect the technology requirements at the CPE. The CPE could be a wires only termination, such as that offered by KPN, or a fully featured Ethernet switch, as in BBned’s network. It could even be an integral part of the network, such as in the ring topology used by Optimum Lightpath.

If a CP only supplies a Layer 2 NTE, service providers or end users are expected to provide additional CPE. Therefore a two box model, with clear demarcation between the CP and infrastructure provider, is very common. In many cases, the responsibility for customer presentation is left to the service provider and their CPE solutions.

• Multicast

Some of the networks in the study do not provide a multicast functionality, especially in wholesale networks, as they have not seen a demand for such a service. Multicast is more likely to be offered where retail services are provided. Some providers only offer multicast as an ‘on-net’ service within their own network. Others, like KPN and BBned offer a multicast service which traverses the Network-Network Interface (NNI), so that other CPs can also provide multicast services using their technology. BBned offers both Layer 2 multicast (in Rotterdam ) and Layer 3 multicast (in Amsterdam ).

• Quality of Service (QoS)

The research found a range of different approaches for the support of QoS. Residential services, such as those from BBned and KPN, are offered with different Class of Service. This allows the CP and infrastructure owner to use the backhaul and access bandwidth efficiently, while still managing high priority services. Providers of business connectivity, like Optimum Lightpath and Telus, tend to sell products with guaranteed bandwidth for each connection or VLAN. In addition to these two options, NTT also provides a best effort service for wholesale internet access and layer 3 QoS for their IP telephony service.

• Security

The extent of support for security in the access network depends on the physical arrangement of the infrastructure, as well as virtual and logical separation of the transport paths. Point-to-point networks, where there are dedicated end to end links for each customer, were rated most highly, as there is a physical separation between different customer traffic streams. In addition to the physical separation, many of these networks also provide a logical separation of data, as VLANs are used to keep different streams of traffic apart. Passive Optical Networks were seen to be less secure than point-to-point architecture as multiple end users share the same optical fibre. However, this was also viewed as a low security risk, as CPs can implement other security settings at higher layers, such as encryption.

• Other Properties of Ethernet Access networks

A range of end user speeds are available from 10Mbps up to 1Gbps and some operators are planning services up to 10Gbps. Typically, residential fibre products have a line rate of 100Mbps and high speed business lines up to 1Gbps. However the research also found that end user speeds were limited by the medium of the access network. In the Colt network, using copper, multiple copper pairs (up to eight) were required to provide high speed access with Ethernet in the First Mile, as each copper pair is only capable of carrying 5Mbps of data. Faster end user speeds, potentially unlimited speeds, are possible on optical fibre.

There are different approaches to inter-operator processes. Some, often those for CPs with residential services, like KPN and BBned, use systems based on XML (Extensible Markup Language). This allows CPs to directly interface with the operator’s systems, and their requests can be processed automatically. Other operators, usually those with a smaller volume of requests such as business operators, would use a manual system. The customer presentation could be a web based form, like that used by Optimum Lightpath, but the requests would be processed by the operator manually.

All networks use flat rate charging to price their access products, although traffic based charging is considered to be an option for the future. Some networks, like Telus, charge by the distance, to encourage operators to roll out their own networks and to interconnect at more locations. Products from others, like Optimum Lightpath, are not as sensitive to distance. Additional services, such as high QoS products and multicast are considered as ‘add-ons’, and therefore further charges are required for these services.

Conclusions

This research has shown that the key requirements of competitive Active Line Access could be met by an Ethernet access product. It also supports the decision to choose Ethernet as the technology for ALA. The Ethernet networks in the case studies deploy different types of infrastructure (Point-to-Point, PON, Fibre Ring and Ethernet in the First Mile) over copper or fibre. Therefore wholesale products based on Ethernet ALA requirements could also be offered over a range of infrastructure. Infrastructure Providers are free to choose the infrastructure technology for their NGA network and will be able to offer a standardised ALA service for interoperability with CPs.

Ethernet has been used in different ways in these networks. Apart from Iliad, who offers a dark fibre service, other wholesale services are provided through Ethernet based protocols, such as Active Ethernet, Point to Point Protocol over Ethernet (PPPoE) and Ethernet in the First Mile. The range in technical protocol, even in this limited number of case studies show that infrastructure owners have a number of Ethernet based protocols to choose from when deciding on the technology of their network.

These case studies confirm that the five key characteristics of Ethernet ALA can be met in some way with Ethernet in the access.

• For flexible aggregation, current networks offer a range of aggregation locations and some infrastructure providers encourage aggregation at multiple locations for resilience.
• In the flexible Support of CPE a two box model of CPE is common. The providers of residential networks provide an NTE and CPs or customers are responsible for the additional CPE.
• Multicast is available as an additional option for some retail services, but there are additional charges.
• There are a range of approaches to providing QoS. Residential services prioritise traffic using different Class of Service, while business services tend to provide guaranteed bandwidth.
• The security of the network depends on the level of physical separation in the network, although logical separation is also use to separate different types of traffic virtually. PON was seen to be less secure as point to point networks as there is a shared infrastructure.

• Ethernet in the Access - International Case Studies [pdf]
  CSMG, September 2008

Back to top