Digital Subscriber Line Access Multiplexers are fast becoming an important issue for carriers, because the DSL-based broadband market is finally beginning to take off. The message is that, at last, customers want broadband – and an increasing number are turning to DSL to get it.
All these DSL lines have to be terminated on DSLAMs in central offices or carrier points of presence (POPs) so that all that bandwidth-hungry data traffic can be pulled off and backhauled to the carrier backbone networks. And that is a lot of capacity to be provided and supported. That, in turn, translates into a lot of headaches for carriers trying to make money out of broadband while minimizing their costs.
This means there are lots of opportunities for vendors with smart backhaul and backbone technologies to relieve the looming pressure that broadband is going to put on the carriers’ networks. But how are carriers to figure out the best approach? The following report aims to offer some up-to-date guidance.
There are two main contenders for the second-mile network upstream of the DSLAM: Asynchronous Transfer Mode (ATM) and Ethernet.
ATM backhaul is well established and fits naturally with the use of ATM in the first mile, downstream of the DSLAM towards the carrier. It’s today’s default, status quo technology. It’s well understood, works well, and there is a lot of it about. So, many carriers may be tempted to stick with it.
But there is a new contender – Ethernet. Not the good old LAN Ethernet but the newer metro-hardened variety. And its enthusiasts see it as the real solution that carriers need.
A great feature for the supporters of Ethernet for DSL backhaul is that it forms part of a bigger picture of a public Ethernet evolution, in which Ethernet plays a major role in metro and access networks for supporting a wide range of services beyond Ethernet itself.
The basic argument for a public Ethernet approach is that it offers Ethernet-based solutions, which provide flexibility, scaleability, and low equipment cost to meet the short-term need for new bandwidth upstream of the DSLAM. It also provides operators with a technology platform that can be migrated into the first mile (between the subscriber and the first network node) as bandwidth requirements increase in the future – both over copper in existing PSTN networks and over fiber in greenfield deployments.
So, how do ATM and Ethernet compare in the critical DSLAM backhaul? This report gives a rundown on the pros and cons of the two approaches – and supplies a quick introduction to DSL that will be useful if you are a little rusty on what this part of the access network is all about.
If you are leaning towards Ethernet, as far as the regions upstream of the DSLAM are concerned, read on to see what the options are.
Here’s a hyperlinked summary:
Why DSLAMs Matter to Carriers
Terminology: What Are DSLAMs Anyway?
The Status Quo: ATM Upstream
The New Contender: Ethernet Upstream
The Operators' Alternatives
— Geoff Bennett, Director, Light Reading University
This report was previewed in a Light Reading Webinar presented by the author and sponsored by LM Ericsson and the Metro Ethernet Forum. The archived presentation may be viewed here: Upstream of the DSLAM: Beating Broadband Bottlenecks
The same issue was discussed by a panel of experts from Ericsson, Alcatel SA, Hatteras Networks, Nortel Networks Corp., and Siemens AG at the CeBIT tradeshow in March. A video archive of the discussion, which was moderated by Peter Heywood, Light Reading Founding Editor, may be viewed by clicking on this link:http://212.227.183.71/kunden/nic/1303_panel/index.html
This month’s Research Poll is also about broadband bottlenecks. Click on this link to give your own views, and get the views of other readers: Broadband Bottlenecks
Finally, a Light Reading Webinar, scheduled for April 10, will focus on a related topic – next-generation broadband remote access servers. Click on this link to register for the live event: Next gen B-RAS – The Money Makers
Worldwide, there were more than 25 million DSL-enabled lines by the end of 2002, and the annual rates of growth are hitting sky-high levels in some national markets: China and Japan are growing at about 200 percent and 100 percent, respectively, while growth among the European leaders runs from about 150 percent for Switzerland, through 100 percent for the U.K., and down towards 50 percent for the Netherlands and Italy. Overall, South Korea, the U.S., Japan, and Germany had the largest national installed bases of DSL by late 2002: S. Korea and the U.S. each had over 5 million DSL lines in service, Japan about 3 million, and Germany about 2.5 million. The strong position of the U.S. is noteworthy, as there’s a common perception that cable modems are completely dominant there.
In short, growing numbers of customers want broadband – and DSL has become an important way for them to get it. As you might expect, the tech-savvy crowd who read Light Reading are well up there among the early adopters. An online straw poll taken during the Webinar on which this Report is based showed that 83 percent of the participants had broadband at home, and nearly 45 percent were DSL users. Another 10 percent wanted broadband, but couldn’t get it, as they had fallen afoul of their carrier’s broadband strategy – which often reflects a preference for shareholders over customers.
All these DSL lines have to be terminated on DSLAMs in central offices or carrier POPs so that all that bandwidth-hungry data traffic can be pulled off and backhauled to the carrier backbone networks. And, with DSL penetrations already at 10 percent to 15 percent of main lines, depending on where you are in the world, that is a lot of capacity to be provided and supported efficiently. That, in turn, translates into a range of technical issues on the network upstream of the DSLAM that depend on the carrier’s environment – as illustrated in Figure 1.
“When you put all the numbers together you can categorize the DSL market into three different broad categories: emerging networks, near-term-growth networks, and mature networks,” says Nan Chen, president of the Metro Ethernet Forum and director of product marketing at Atrica Inc. “And these demand three different strategies for service providers to implement Ethernet backbone technologies and for vendors to penetrate these technologies.”
So it looks as if there are lots of opportunities for vendors with smart backhaul and backbone technologies to relieve the looming pressure that broadband is going to put on the carriers’ networks.
To clarify the terminology, it’s best to start with a brief overview of a DSL network architecture and to describe the role that the various active devices play – in particular, the DSLAM.
Again, ”DSLAM” stands for DSL access multiplexer. DSL – or Digital Subscriber Line – is a set of technologies that are intended to allow access to significantly more capacity in a local loop of copper cable than would be available from, let’s say, a modem.
Figure 2 shows a typical local-loop configuration without DSL.
On the left is the residence, although it could equally well be a business, receiving either standard residential-type DSL or one of the forms – such as symmetric DSL (SDSL) – tailored for business purposes.
Inside the house is a telephone jack, with a phone plugged into it. There is also a PC in the house, and to access the Internet it uses a modem, also connected to the telephone jack. The local loop itself runs from the house to the telephone exchange; but the details here vary a lot among different countries, which means a range of active or passive devices or components may appear in the local loop. In other words, anything but the phone wire may be an issue for DSL somewhere in the world.
At the exchange, the loop terminates on the main distribution frame, from where it is passed to the local switch to give the user dial-up access to the POTS (plain old telephone service) network – e.g., the Internet exchange carrier – including PC access via modem to the user’s ISP.
Adding DSL to this conventional POTS setup leads to Figure 3.
First, the carrier installs a DSLAM in the exchange; subsequently, each residential subscriber must have a DSL modem installed. These come in many forms: internal PC cards; external passive boxes that draw power from the USB port; and combined systems like a DSL modem and wireless router. An important component in this setup is a simple passive filter (F), which allows the loop signal to be split into two frequency bands – a lower baseband that carries the POTS voice channel unchanged and an upper broadband region that carriers a high-frequency signal encoding (in an elaborate fashion) the data channel. At the exchange DSLAM, the voice and data channels are separated again.
From the viewpoint of the carrier, all the varieties of DSL (such as ADSL, SDSL, and VDSL) are downstream technologies. That is, they operate within the section from the exchange to the home or business. This report focuses on a different part of the network – from the exchange into the core of the network – and the technical options that carriers have in this portion, which is upstream for the carrier – as shown in Figure 4.
Today, the standard approach is to use ATM technologies on the upstream side.
Figure 5 shows how ATM works out as an upstream solution.
Out the back of the DSLAM is an ATM port, typically running at 155 Mbit/s into an ATM switch. This ATM switch will simply aggregate DSLAM ATM connections from many exchanges and merge them into a higher-speed ATM link – typically 620 Mbit/s. These bit rates are, of course, common because of the underlying Sonet/SDH transmission hierarchy that will usually be supporting the network.
The next device in Figure 5 is the broadband remote access server (B-RAS). This is a special device that is needed to concentrate the huge numbers of end-user PPP (point-to-point protocol) sessions coming from the DSLAM onto a much smaller number of PPP sessions going into the core network.
The operation with ATM is roughly as follows. The DSLAM is the link from the home into the network. From a home connection – perhaps browsing the Web or connecting into a company VPN (virtual private network) – the user will be creating various PPP sessions. Each PPP session is maintained between the Internet Protocol devices at the ends of the link. So, if there are 5,000 homes plugged into this DSLAM, and each home is generating three PPP sessions, there are 15,000 active PPP sessions coming out of the back of the DSLAM over an ATM virtual circuit.
Many current ADSL (asymmetric DSL) installations will use a format called PPP over ATM (PPPoA). There is also the option of PPP over Ethernet over ATM (PPPoEoA). Either way, there are lots of PPP sessions that need to be terminated.
Now, when the ATM switch collects lots of 155-Mbit/s connections from lots of DSLAMs, there will be thousands of unterminated PPP sessions running on this link. This may not be a scaleability issue for ATM, which is designed to maintain large numbers of active VCs, but it’s certainly an issue for a typical IP router. The router in this part of the network is the first IP device after the CPE (customer premises equipment) PC, and that means it has the job of terminating those PPP sessions and merging them into a much smaller number of PPP sessions that feed into the core network.
Conventional routers can’t do this kind of termination. It requires hardware support and specially designed boxes. So a class of device has evolved to deal with this job – the B-RAS.
This description suggests that there’s a rather forced marriage of ATM and IP technologies. Why use ATM in this design?
In a typical ADSL network the upstream link out of the DSLAM is quite heavily oversubscribed, because it’s an expensive link for a typical carrier to install and maintain. So they pile on as many customers as they can get away with. The thinking goes that, if this link is deliberately oversubscribed, they’d better use ATM for this connection as it’s the only mature, carrier-grade, QOS-capable network technology.
But, in addition, this is a public network service. Carriers will need to be sure they can offer the service at very high levels of reliability. Aside from the obvious issues of good design practice in the hardware, software, and network architecture, a key service element is the ability to locate faults as quickly and precisely as possible. For instance, if a carrier knows that a link from the ATM switch to a given exchange is down, an engineer can be dispatched to fix it. But which end of the link is causing the problem – the exchange end or the central office end? It makes a lot of difference in travel time if they get it wrong. ATM is always implemented over Sonet/SDH in ADSL networks, and the OAM (operations, administration, and maintenance) features for all of these technologies are second to none.
The final issue is often the clincher. If you look on the back of the DSLAM you will almost always find an ATM port. Anything else is nonstandard, and probably not widely deployed.
Put these points together, and you get a fairly strong case for using ATM upstream of the DSLAM, at least initially for big incumbents with large legacy networks and lots of embedded Sonet/SDH. But the telecom environment isn’t standing still, and the seemingly unstoppable rise of Ethernet as an alternative carrier transport solution suggests that, longer term, Ethernet is going to reach down from the core network to the DSLAM. The question is whether this is going to happen in a large way any time soon, or whether there are lingering issues that prevent real deployment today.
A great feature for the supporters of Ethernet for DSL backhaul is that it forms part of a bigger picture of a public Ethernet evolution. This builds on the technology advances in metro Ethernet from the 10 Gigabit Ethernet Alliance (10GEA) initially, and then through the current work of the Metro Ethernet Forum and the developments for Ethernet in the first mile (EFM). Figure 6 shows some of the various applications that a public Ethernet network could support.
“DSL backhaul is one of several applications that could be combined with, for example, Ethernet feeding hotspots, and Ethernet feeding FTTB [fiber to the business] applications for both residential and businesses, as well as enterprise applications,” says Peter Linder, technical expert for Ethernet broadband access at LM Ericsson (Nasdaq: ERICY). Ericsson, for example, is developing solutions for deploying EFM to complement its Ethernet DSL access solution.
The basic argument for a public Ethernet approach is that it offers Ethernet-based solutions, which give flexibility, scaleability, and low equipment cost to meet the short-term need for new bandwidth in the second-mile network. It also provides operators with a technology platform that can be migrated into the first mile (between the subscriber and the first network node) as bandwidth requirements increase in the future – both over copper in existing PSTN networks and over fiber in greenfield deployments.
Cost is a strong argument in favor of Ethernet upstream of the DSLAM. Capital expenditure (capex) is reduced because the expensive high-capacity ATM switches are eliminated – a crucial fact when the number of ATM PVCs required by broadband services is increasing rapidly. And, simultaneously, the high operational expenditure (opex) of provisioning thousands of ATM PVCs is eliminated. ATM also leads to complexity, as the IP/ATM/Sonet/SDH layers must be managed separately, each requiring different management tools and staff.
Ethernet is generally cheaper and easier to handle, and it can support a broad range of services – such as point-to-point and circuit emulation – and can also support a significant degree of user oversubscription, which offers further potential savings.
“Plus, integrated Ethernet optical management will save a lot of money, too,” says Atrica’s Chen. “So, all in all, there is a very strong case for Ethernet in terms of doing the backhaul, as well as being a transforming choice in the metro space.”
Figure 7 shows how the DSL access portion of the network can be evolved in three steps towards the public Ethernet model.
The first step is a traditional broadband architecture with any of the DSL combinations (ADSL, SDSL, and VDSL) being delivered by ATM over DSL in the first mile; in the second mile – basically, from the DSLAM and upwards – it’s transported by a variety of ATM variants up to an ATM switch and then into a B-RAS.
The next step in the evolution occurs from the DSLAM upwards and takes advantage of Gigabit Ethernet, a technology that did not exist when the traditional broadband architectures were set in the mid-1990s. Using Gigabit Ethernet in this way reduces interface costs in the DSLAM itself and also in the aggregation of traffic in the Ethernet switch and into the B-RAS. Standard ATM over ADSL still runs on the actual copper link in the first mile, because those are the most widely deployed technologies today. So, by taking the interim step of introducing Gigabit Ethernet from the DSLAM upwards first, all existing standards for ADSL can be conserved towards the CPE (the DSL modem in the home).
The final step is an evolution towards an EFM standard. This occurs when fiber is brought deeper into the access network, either to a node supporting EFM copper (Ethernet over VDSL) or with a fiber solution (EFM over singlemode fiber).
EFM copper increases the reach of Ethernet over copper from the current 100 to 200 meters over standard Category-5 cable; it also allows deployment over networks using existing copper pairs, so there is no need for new cabling. EFM fiber covers both 100-Mbit/s and Gigabit Ethernet over singlemode fiber.
“We have actually got standards from 10 0Mbit/s upwards for singlemode fibre, which will allow the access nodes to be placed further upwards in the network,” says Ericsson’s Linder. “What we are talking about – DSL with Ethernet backhaul – is a step in the evolution towards an all-Ethernet-based access network, where we will eventually end up with a lot of fiber in that access network.”
Ethernet everywhere is a pretty picture, but is it really necessary? There are two big reasons for suggesting that the answer is yes: growing capacity requirements and return on investment.
Capacity
One is just the sheer growth in access bandwidth that is becoming necessary for Internet services – and one of the main drivers in Internet access today is the move towards more bandwidth-hungry services. However, three parameters define the limits of the services that can be supported:
The traffic that can be supported by the DSL link itself
The speed defined by the service provider (peak best effort)
The speed that the up-link has been dimensioned for
Figure 8 illustrates the scale of the bandwidth problem.
In the early days of dial-up Internet access, the driving services were of the Yahoo! and Hotmail type, with simple text and graphics. The wire speed was 56 kbit/s, and that was in most cases the limiting factor, which meant the service offerings were very tightly bound to the wire speed. Traffic patterns from these services typically led to an upstream concentration of 20:1 — 2 to 5 kbit/s from the remote access server and upwards into the network.
The introduction of basic ADSL changed that simple picture. Driving services became download/sharing-oriented like Napster for MP3 files; even though Napster was illegal, it drove a lot of the ADSL penetration. ADSL boosted the wirespeed significantly, up to 8 Mbit/s. But, to provide some kind of limitation on what was needed in the network, the most common service offerings today are based on 512 kbit/s downwards towards the user and 128 kbit/s upwards. So the uplink dimensions of these networks follow dimension rules very similar to dial-up; 20 to 25 kbit/s per connected subscriber is quite common.
Now we have moving into a post-Napster era, with new applications like to peer-to-peer Kazaa, which allow users to exchange not only 3-minute music clips, but also larger chunks of MPEG-2- and MPEG-4-encoded content. We are also seeing higher wire speeds with the further development of the ADSL standards: still 8 Mbit/s downstream, but now increased in the upstream direction to between 2 and 3 Mbit/s with the introduction of ADSL2. As the traffic patterns change here, with new bandwidth-consuming applications, there is growing interest from end users to get a better service offering than the existing 512 kbit/s; and it is also important for operators to increase the uplink from the DSLAM and upwards through the network to support these new services.
And it isn’t going to stop here when VDSL (very high speed DSL) is factored in. Although it is impossible to predict what the future really high-bandwidth services will be, it is likely that some combination of big MPEG content willbe a driver.
“If something emerges that is a legal combination that provides MPEG content to subscribers, that would definitely drive bandwidth. We see with the interaction of ADSL2+ and VDSL standards that the wire speed is going to go up significantly, so it will not represent the bottleneck, assuming that you manage to get fiber to that particular point,” says Ericsson’s Linder. “One of the key challenges that operators face for future broadband capacity is to increase the capacity in the upstream towards the network.”
Cost
But capacity is not the only problem; there is also a big revenue challenge for carriers, as the economics of broadband are pretty difficult – lots of upfront investment and uncertainty about how much users will pay for what – including the ever-present threat of cannibalization of some existing service revenues by users taking an IP alternative, such as voice over IP (VOIP). In short, there is a first-mile issue of a costly rollout leading to a low penetration – which makes for a poor business case. Then there is a second-mile problem, which is the exact inverse — as traffic increases because customer usage increases, the cost of the second mile goes up, too. Heads you win, tails you lose.
Today most DSL offerings worldwide are based on more or less invariant service offering in the range of $40 to $50 per month for broadband access connection, and that goes for power users as well as more price-sensitive users. This is not the way to make a lot of money, as any market trader will tell you.
Basically, carriers will have to establish a price curve along which they can separate different types of customer so as to extract maximum revenues. This means that carriers need a low cost for their basic service and the ability to cheaply add extra capabilities and bandwidth to order to satisfy a range of higher-value and premium customers.
“If you can achieve a breakeven point that is lower for the basic network but, at the same time, has paved the road for additional applications coming on top of it – we believe that that is the winning concept to secure a continuously high broadband growth,” says Linder.
This translates into a requirement for a technology base that can lower network costs, thereby improving the bottom line immediately. There needs to be flexible bandwidth management to make the offering of differentiated levels of service easier and to allow additional premium high-bandwidth services.
Thus, the business model moves from an access-only offering to one with rich media content and applications, as depicted in Figure 9.
For its proponents, it is at the confluence of high bandwidth and low cost where Ethernet wins. Growing broadband penetration and increasing demands for capacity-greedy services put great pressure on the access network – but even greater pressure on the upstream aggregation and transport network. And Ethernet is a way of handling the upstream problem, but with the longer-term prospect of being deployable downstream as well, to provide a single, coherent network architecture.
In broadbrush terms of bandwidth, flexibility, and cost, public Ethernet can seemingly make an attractive case as an alternative to the traditional broadband access based on end-to-end ATM solutions. But what of the hard practicalities for carriers of switching to Ethernet to support a mass market for broadband? This would imply:
Introducing Ethernet rather than continuing on a pure ATM path
Increasing throughput in the access node to eliminate any traffic bottlenecks in the DSLAM
Increasing the capacity in the uplink and reducing costs at the same time
Optimizing the solution for IP-based services rather than a mix of IP- and ATM-based services
Migrating to an EFM architecture rather than a full-service access network (FSAN) architecture
The table below summarizes some of the practical issues that can affect carriers’ thinking about keeping with ATM or of moving to Ethernet. As usual, specific circumstances play a big role in determining the importance carriers attach to particular issues.
Table 1: Carrier Feedback on the Two Alternatives
The first point that comes across is that in many cases there is idle ATM and Sonet/SDH capacity in the network that carriers would like to use up first, rather than investing in new forms of backhaul. And that makes a lot of sense. However, there are carriers that are saying it is important for them to reduce the concentration factor or the contention in the DSLAM to provide for new services – and this is a strong point for Ethernet.
Another frequent concern is that there is a lot of existing process investment in the ATM backhaul related to service and network provisioning that carriers do not want to jettison. Opposed to this view are carriers that seek aggressive rollout and growth on the back of the low rollout costs that Ethernet gives.
Similarly, carriers using ATM backhaul have no learning curve for their staff. They know the ATM gear, they are familiar with the equipment, and they have deployed it for a number of years. However, some carriers are prepared to argue that the advantages of Ethernet are so great that its adoption is eventually inevitable and it is better to do it sooner rather than later, even if it involves a learning process.
The fourth issue is service related. Proponents of ATM-centric solutions often say that derived voice is a key element of their business plan and that they would like to provide voice over ATM. Carriers more interested in Ethernet-centric solutions often see VOIP as the key element – and they realize that means large amounts of bandwidth for which they need to find a cost-effective solution.
Finally, as Ericsson’s Linder points out, different carriers have different visions of their long-term network goals, and this influences attitudes towards the two technologies. “An element that we often come across is related to: What are we migrating towards?” he says. “A migration with an ATM backhaul approach often has as a target a fiber revolution, ending up with an ATM PON [passive optical network] and ATM VDSL. An Ethernet-backhaul approach is often the starting point for a journey towards an Ethernet-in-the-first-mile solution, with fiber deeper in the network and EFM as a target migration for that fiber revolution.”
So, after careful consideration, you have decided to buy the Ethernet story as far as the upstream regions from the DSLAM are concerned. What are your options?
From the greenfield perspective, if you are thinking of deploying an Ethernet-backhaul technology, there are multiple ways to do it, namely:
DSLAM as a bridge with VLAN mapping
DSLAM as a straight bridge
DSLAM as an IP router: This is quite straightforward, as it’s basically a DSLAM incorporating an IP router, and there is no shortage of kit from which to choose – for example, Cisco Systems Inc. (Nasdaq: CSCO), Copper Mountain Networks Inc. (Nasdaq: CMTN), and Lucent Technologies Inc. (NYSE: LU) each have products in this class. Subscribers’ PVCs are terminated and routed, and all IP traffic is aggregated onto a single Layer 2 uplink, although individual subscriber identities may still be determined at the B-RAS by the source IP address (this is B-RAS dependent). PPPoE cannot be used across the router to the B-RAS – it just doesn’t work with this architecture – but PPPoE is still OK to the DSLAM. This is actually a good way to provide basic IP functionalities, although this architecture is not optimized for enabling IP services in a B-RAS. But it works well in small networks with no B-RAS and a single ISP, or for carriers that simply don’t want to provide IP services.
DSLAM as a bridge with VLAN mapping: Here the DSLAM bridge applies Ethernet framing and appends a different 802.1q VLAN tag for each subscriber. The Ethernet switches basically tunnel the connection transparently with no knowledge of the individual 802.1q tags, so there is no need for any kind of provisioning in itself and individual subscribers are not separated by the B-RAS at Layer 2. This approach would work well typically with PPPoE and other Layer 2 protocols because of the Layer 2 transparency. It is possible to offer a simple form of QOS for business subscribers by using the 1q tag as an identifier.
DSLAM as a straight bridge: This is very straightforward, as it is a pure Ethernet model, and no logical subscriber separation is maintained at Layer 2, so Layer 2 is bridged transparently through the network. PPPoE and other Layer 2 protocols may be used as well. This architecture works very well with PPPoE, and the subscriber identities are maintained by PPPoE session IDs.
DSLAM as an L2TP LAC: In this model, DSLAMs terminate PPPoE tunnel sessions inside L2TP tunnels. So the session for all ISPs uses the same tunnel, and carriers can obviously create backup tunnels for resilience. Centralized B-RAS farms terminate tunnels from DSLAMs. The entire network between the DSLAM and the B-RAS is IP based.
As there are already a lot of ATM DSLAMs in place, you are likely to need a migration plan from ATM to Ethernet backhaul. When you do the migration, you want to preserve the existing networks – which means, typically, that you have to install an aggregation device between the existing ATM-based DSLAMs and the new Ethernet-backhaul switches to translate the ATM into Ethernet. This particular box could employ any of the above greenfield solutions.
The aggregation device must do the following: It has to support Ethernet/ATM connections via DS3, DS1, OC3, and STM1 connections; it has to support Ethernet, Gigabit Ethernet, or Fast Ethernet uplinks; and it has to provide one or more of the Ethernet DSLAM functions mentioned above – either as an IP router, a bridge with VLAN tagging, a straight bridge, or an L2TP LAC. There are many such devices in the marketplace already.
QOS and provisioning are just as crucial for Ethernet backhaul as they are for ATM backhaul.
Says Nan Chen of Atrica: “In terms of QOS, obviously, this is not plain old Ethernet service we are talking about here. This is an Ethernet that has a strong orientation for a connection-oriented service, where people are defining services as either point-to-point or multipoint-to-multipoint. In fact, the newly defined services by the Metro Ethernet Forum are being tested and agreed by members with CIR [committed information rate] and also PIR [peak information rate] types of services. In other words, this Ethernet, either in the backhaul or providing just generic Ethernet services in metro networks, will be able to provide QOS on top of it.”
There are multiple mechanisms for supporting QOS. Multiprotocol Label Switching (MPLS) as an underlying mechanism to provide the connections is one, but you could provide the services with other connection-oriented technologies (even a VLAN), as long as you can identify a connection. “Once that connection is identified, the network can be provisioned to provide QOS just like any other network,” says Chen. “So Ethernet can be a highly reliable and QOS-enabled network.”
Figure 10 shows a simple approach, in which 802.1p provides explicit priority information for packets in the Ethernet backhaul network to support QOS. Essentially, the DSLAM performs a mapping between ATM PVCs on one side on the copper wire and the Ethernet uplink, over which 802.1p prioritization is used for the different traffic streams. So the traffic, even though it goes over Ethernet, is always being prioritized.
As Figure 11 illustrates, provisioning IP is easier when there is a B-RAS, either inside or outside the DSLAM, as you can provision different services. What’s nice about Ethernet is that you don’t lose granularity in terms of identifying the subscribers, as they can be identified by a VLAN ID, an MPLS tag, or whatever – and this can be used as the basis of applying QOS to that particular subscriber.
Finally, backhaul Ethernet is potentially a big step towards the wider provisioning of broadband in small towns and communities, because it alters the economics of small-scale installations. The current problem is well known: Providers need to get hundreds of subscribers before it becomes worthwhile to install a DSLAM, thereby restricting the availability of broadband access.
There are two prime issues here. First, with existing DSLAMs, which are to a large extent chassis-based and designed for large central offices of very high densities, there has been an issue of downward scaleability. In a lot of cases, that has meant that 200 to 300 subscribers per site is the minimum entry point for deploying solutions. However, there are now solutions in the market that go down to as low as 20 subscribers.
But the other issue that has been limiting the penetration of small sites has been the availability of high-capacity transmission towards those sites. Many carriers believe it is too expensive to deliver the transport capacity required for broadband into those sites. But there are today solutions in the market that will allow carriers to feed those sites either over a spare fiber with Ethernet straight away; or through upstream E1 connections by reusing capacity in existing PDH or Sonet/SDH networks. There are also solutions based on wireless transport.
“So I think right now we are coming to the point where new technologies enabled by Ethernet will be able to drive down the price points and economics for going into smaller sites that we haven’t seen before,” says Peter Linder of Ericsson.