首页    期刊浏览 2025年08月28日 星期四
登录注册

文章基本信息

  • 标题:How efficient is your backhaul link? - The Unbundled Network
  • 作者:Jim Guillet
  • 期刊名称:Telecommunications International
  • 印刷版ISSN:1534-9594
  • 出版年度:2001
  • 卷号:Nov 2001
  • 出版社:Horizon House Publications

How efficient is your backhaul link? - The Unbundled Network

Jim Guillet

Local loop traffic aggregation, through the use of ATM switches, can significantly improve the business case for delivering high-speed internet access over ADSL networks.

Deciding which is the best network architecture for aggregating local loop traffic will depend on the applications themselves. These can range from voice to multimedia applications (such as mobile, PBX, frame relay and internet traffic). And as each application has a varying quality of service requirements, traffic aggregation needs to support a multi-service environment.

Traffic aggregation typically occurs on the backhaul link (ie, between the central office or base station and the operator's point of presence on the core network). How this backhaul link is designed and managed will be crucial to the service provider's business case, since it will largely determine the costs and levels of customer service it can offer.

One of the key applications is, of course, high-speed internet access (HSI). And without an optimised aggregation and backhaul strategy, the business case for providing HSI weakens. Central to achieving the optimum network architecture for the delivery of HSI are network edge concentration and flow-through service provisioning.

HSI basics

To appreciate the business case for multi-service aggregation generally -- and HSI aggregation specifically -- a quick review of HSI is in order. The principles behind HSI traffic aggregation are the same whether the fixed local loop access is wireline or wireless. For the purposes of this article, however, the focus is on wireline and ADSL (asynchronous digital subscriber line) technology since it is this transmission medium which currently supports the most HSI traffic.

The basics of HSI via ADSL, as illustrated in Figure 1, are as follows:

* An ADSL modem on the customer premises provides the interface between ethernet and the ADSL network termination. The modem provides an ethernet bridging function, performing TM segmentation and reassembly of the customers' IP traffic in and out of a ATM virtual circuit (ATM VC).

* A digital subscriber line access multiplexer (DSLAM) provides the ADSL line termination and aggregates the ATM virtual circuits from all subscriber loops onto a DS-3, OC-3 or STM-1 AT uplink.

* DSLAM uplinks feed into an ATM backbone, across which each customer's ATM VC is backhauled to the appropriate ISP where it is terminated on a broadband access server (BAS). The BAS provides P-level session management, including authentication, access control and administration. Behind the BAS are the routers, switches and servers hat comprise the internet.

Backhaul utilisation

Initial residential deployment for HSI over ADSL typically impose a maximum downstream speed of 500Kbps and an upstream speed cap of 128Kbps. Business deployments for HSI, which command a higher price, allow higher speeds depending an local loop length.

Despite these bit rates, the peak, simultaneous usage of residential customers is in the order of 15Kbps. With the vast majority of DSLAM deployments supporting 200 to 500 subscribers (a few large sites support up to 2,000), the corresponding peak load on the DSLAM's uplink varies from 3 to 7.5 Mbps. Depending on the up-link speed -- 45 Mbps or 155 Mbps -- the peak, simultaneous uplink utilisation for residential HSI is 2 to 20 per cent.

Taking into account that not all subscribers will be using service simultaneously, over-subscription of data services is usually engineered at four to eight times (ie, an operator sells more capacity to individual subscribers than is actually available on the backhaul link). With residential HSI, however, over-subscription can be as much as 50 times although, in practice, approximately ten times over-subscription is the norm. This partly explains why an ATM backbone is so often used to aggregate HSI traffic at the network edge. Aggregating ATM virtual circuits (VC) from multiple DSLAMs provides a significantly higher utilisation of backbone facilities, compared to having the DSLAM interconnect directly with the BAS over a leased line.

Even though the DSLAM itself aggregates residential HSI traffic from hundreds of ADSL subscribers, backhaul utilisation can be improved by at least another factor of ten using ATM switches.

Capex optimisation

Another consideration in the business case for HSI aggregation is the optimal mix of IP session managers, commonly referred to as broadband access servers (BAS) and ATM switches. The economic question is: since the BAS supports direct termination of ATM interfaces (as well as IP session management), why not co-locate a BAS with each DSLAM, ignoring the ATM network entirely?

Not surprisingly, the answer is cast. It's an old maxim in the networking trade that the price per port generally increases as its location in the network moves toward the core. In addition, the price increases with the platform's intelligence. Both apply to the BAS when compared to the relatively inexpensive ATM edge switch.

By allowing the ATM network to aggregate HSI traffic from multiple DSLAMs and backhauling it to a few centralised BAS locations, net capex is reduced. For example, the market price for a BAS' ATM interface is at least triple that of an ATM switch. From the back-haul utilisation discussion above, aggregating the traffic from ten DSLAMs onto an equivalent sized pipe through an ATM switch, and passing this highly utilised pipe into a single interface on a BAS reduces capex by more than 50 per cent.

Interestingly, this capex analysis holds far network access providers who extend their IP networks out to the edge as well as for those who use ATM at the edge. Using an ATM switch in front of a BAS to aggregate ATM VCs reduces capex more than 50 per cent, whether these products are co-located or not.

Flow-through service provisioning

In addition to backhaul and capex efficiency gains through aggregation, significant operational savings can be achieved through network and service management tools that automate the provisioning of HSI. With the popularity of HSI generating massive numbers of provisioning requests, faster and more accurate service provisioning will not only increase customer satisfaction, it will also reduce the work load involved in resolving front or back office customer issues.

From an aggregation perspective, the requirement is for the provisioning of each customers' ATM VC from the DSLAM, through the ATM backbone, to the BAS. Service moves and changes among multiple ISPs must also be accommodated.

Using standard element management techniques as a base line, a system that enables path provisioning -- ie, an ability to provision each network element automatically by specifying circuit end points and allowing the management tool to calculate and configure the rest -- is a significant step forward. Such a tool would increase the rate of subscribers provisioned per day and reduce the number of provisioning errors, if nothing else, by eliminating the problems that inevitably result from the practice of configuring each network element by hand.

A conservative estimate of the operational productivity improvement is 20 per cent. Assuming a US$50 ([member of]55) cost to provision an HIS circuit, such a network and service management tool would liberate US$10 ([member of]11) per subscriber. Given that the capex per subscriber for an ATM aggregation switch is also about US$10 ([member of]l1), the operational savings from the initial service turn-up alone pays for the aggregation outlay. Further operational savings from adds, moves and changes over the service lifecycle are incremental and apply directly to the bottom line.

The role of outside plant

Most HSI today is deployed from DSLAMs based in central offices. This has made competitive access relatively straightforward with multiple providers locating equipment within the same building.

But not every potential HSI subscriber is served by a central office, and a fact not lost on equipment vendors and access providers. New generations of digital loop carrier (DLC) systems have emerged that enable ADSL deployment from remote terminals located in neighbourhood pedestals.

To address the unbundling of ADSL lines deployed from within the regulated incumbent carrier's outside plant, a new network element known as the optical concentration device (OCD) has emerged. Located in the CO, the OCD terminates the ATM uplink from the new generation DLC and provides virtual access to the subscriber's loop via an ATM VC ('virtual' because it is almost impossible to allow physical co-location in a pedestal). For example, in the US, regulators have mandated that the OCD provide a DS-3 or OC-3 hand-off point to competitive carriers.

The OCD is essentially a small ATM switch. It provides aggregation and switching of subscriber ATM VCs among different access providers. As this deployment model builds out, it will provide another opportunity to extract cost from the network through statistical gain. Access providers should aggressively evaluate traffic loads at OCD hand-off points to ensure maximum utilisation. Not only will this reduce their backhaul costs in high density serving areas, it will reduce their aggregation costs as well.

Leveraging multi-service capabilities

Pursuing an optimised network strategy begins with a single, unifying infrastructure that supports overlays instead of disparate networks. With diverse traffic flows demanding different qualities of service, advanced traffic management is essential to differentiate and apply, for example, 'gold', 'silver' and 'bronze service levels. Network and service management plays a key role in simplifying the management of multi-service networks by automating the task of service provisioning and ensuring the delivery of SLAs (service level agreements).

While HSI is an excellent example to demonstrate the cost savings and service improvements possible with a good aggregation implementation, the role of the multi-service edge in the delivery of voice, frame relay and multimedia services is lust as important (if not more so) from a revenue and profitability perspective. Mobile operators also use multi-service networks to aggregate 2G, 2.50 and 3G access traffic and its transport over a unified core, while new voice networks can position multi-service platforms as media gateways, converting traditional speech flows between circuit and packet domains.

The business case for aggregating high-speed internet services is a good illustration of how costs can be extracted from the network while improving customer service. Aggregation of local loop HSI traffic optimises the utilisation of backhaul facilities and BAS equipment, reducing operational and capital expenses, while flow-through service provisioning accelerates service revenue and reduces errors.

As port of an overall multi-service network, aggregation plays a key role for all services and applications. Getting aggregation right results in an ability to connect more customers, deliver quality of service and reduce time to revenue. Combining it with a scalable, multi-protocol core network should result in an end-to-end, unified network infrastructure that will ensure the continued delivery of profitable services in the years ahead.

COPYRIGHT 2001 Horizon House Publications, Inc.
COPYRIGHT 2002 Gale Group

联系我们|关于我们|网站声明
国家哲学社会科学文献中心版权所有