Light the city, lighten the load: Europe's optical networking markets are not as buoyant as they once were but supply still rarely outstrips demand at the metro level - The Bandwidth Economy

The optical networking market has seemed a bleak place for some time now -- service providers are in trouble and dramatically reducing spending on optical gear. In turn, this has left vendors' balance sheets looking, well, unbalanced.

Certainly, long-haul equipment shipments have tailed off in Europe but really this was to be expected with something like 15 pan-European networks either built or being built or planned a couple of years ago. Most of them duplicated the London-Paris-Amsterdam-Frankfurt routes to such an extent that self-inflicted price cuts to gain customers hamstrung the business case and put paid to quite a few of these networks (or at least the operating companies -- the network assets are still there, getting snapped up on the cheap, furthering the drop in new long-haul optical spending). It is a huge generalisation to say that there is a bandwidth glut in Europe; it's perhaps more true to say there is too much of the same sort of bandwidth. But, as we will see, there is still not enough of other sorts.

Metro supply and demand issues

Most particularly, there is not enough metro bandwidth -- it is here that demand almost always outstrips supply and, if you will excuse the pun, where there is some light at the end of tunnel for optical networking. The metro network, the zone between the home or corporate network and the service provider backbone, has come into focus as an entity unto itself. More specifically, the metro network consists of:

* last mile: connecting businesses and homes to a central office (CO);

* metro edge: aggregating and grooming traffic and applying services; and

* metro core: transporting traffic between COs

The metro network -- or city network -- is undergoing a metamorphosis, driven by dramatic growth in the number and types of companies needing metro communications solutions. These include large corporations, banks, media companies and content distributors, universities, local governments, public services as well as a plethora of service providers -- incumbent and competitive telcos, ISPs 2G/3G wireless operators -- as well as specialist storage and security service providers. The demand for more metropolitan bandwidth, fibre and services is not in question and this is true not only of major capitals but in tier 2 and tier 3 cities as well. Globally, efforts are now being focused on re-engineering urban infrastructures to meet this demand.

In forecasting how such re-engineering will take shape, it is worth considering the changing market conditions and emerging technologies that impact on metro network evolution.

For the users listed above, and for major corporations particularly, their networks and the type and sensitivity of their data traffic are changing. Data is increasingly concentrated into various locations within the metro area -- in enterprise buildings, hosted data centres, and COs and POPs -- and is accessed from the same metro area. More data is staying local to the metro area because:

* service providers have built and continue to build their data centres in major metro areas to handle local traffic demands (eg, WorldCom, Verio and Cable & Wireless);

* major companies often have several sites, as well as their data centres, within the metro region and these sites 'talk' to one another;

* companies are increasing their connection speeds via DSL, OC-3 and broadband wireless, which pulls more data onto their networks;

* business applications are becoming richer, using more graphics and video;

* many companies deal heavily with other companies in the same metro area and so business-to-business traffic continues to grow;

* there is a rise in adoption rates of hosting applications such as e-mail, financial applications and groupware to a third party; and

* there are increasing numbers of residential customers subscribing to broadband services.

On the service provider side, the growth of the internet causes urban problems because investment in local infrastructures has generally lagged behind the 'land-grab' build out of long-distance networks. There is an urgent need to establish infrastructure within cities based on technologies such as DWDM that will create metropolitan networks comparable in capacity to long-haul networks. Urgent not only from a demand viewpoint, but also in terms of revenues -- the metro space, with its proliferation of valuable corporate customers, is where the money is for service providers.

While prices for wholesale international bandwidth have fallen by as much as 50 per cent on many major routes according to some carriers, bandwidth prices within cities have fallen much less dramatically. The result is that the cost of the final few kilometres within the metropolitan area is frequently as high as the cost of the many international kilometres of an end-to-end link.

Technology issues

The mismatch between demand for, and the relative value of, metro bandwidth and the capability of metro networks to supply it is because legacy metro networks have not been built to handle such volumes of data traffic and have been caught in no man's land' between the speeds available within the LAN on one side and in backbones on the other.

LANs have matured. The steady increase in the powers of ethernet, from 10Mbps to 100Mbps and 1 Gbps, and now nearing 10Gbps, aligns LAN speeds with even the fastest core speeds (10Gbps ethernet, for example, with OC-192). However, the last mile connections of businesses and residents are mostly TDM copper-based, which provides relatively low speeds. Desktops may have 100Mbps ethernet for internal connections but for the majority of businesses the connection from the LAN to the WAN or internet is dial, DSL or T1. The local loop has become a mere sipping straw that cannot satisfy the growing metro area bandwidth thirst.

This puts increased pressure on the metro edge and core networks to handle the incoming larger data streams. It is no coincidence that Korea, arguably the most advanced broadband nation in the world, has been a highly productive market for vendor Riverstone Networks, a metro ethernet specialist, now supplying the top three Korean broadband service providers.

As it stands at the moment, legacy metro networks are structured around too many layers and devices. Service providers will look to collapse network layers, simplifying the networking model by placing IP over optical and removing now-superfluous equipment and protocols in between. ATM has not disappeared in the network backbone, but new networks are being designed strictly around IP and the service-generating potential of MPLS.

To date SONET/SDH is the dominant technology in the metro core but that was not intended for the burstable, rapidly fluctuating internet. Consequently, mast service providers lack efficient ways to create differentiated services at this point in the network. To realise the revenue potential of metropolitan demand, they must not only deliver bandwidth but services and this requires intelligence in the network that has not previously existed. Service providers are being pulled in different directions because of these needs. Not only must they direct greater investments into technologies that can quickly provision more dynamic services, but they must do so against a backdrop of financial conservatism.

Currently, complexity is systemic in metro networks, which, in their present incarnation, cannot be optimised, relying as they do on SONET/SDH fine-tuned for voice traffic that grows slowly and predictably. Access lines from buildings to the network are thin compared to LAN connections and backbone speeds. TDM leased lines and other copper technologies are not easily provisioned, and do not scale to the needs of the new metro model (Figure 1 shows the complexity of today's metro network).

Today, with the bulk of installed optical equipment, technicians must physically visit the hardware to configure it and to provision services. Worse yet, in order to increase bandwidth from one point to another in a multi-ring path once capacity is reached in a single SONET/SDH ring, the capacity of every add/drop multiplexer (ADM) in every ring must be upgraded. These characteristics of unintelligent optical hardware cause major problems for service providers, increasing their operational costs and prolonging the time to fulfil customer orders for new service.

The next generation

To meet these challenges, a new generation of data-responsive intelligent optical technologies is emerging. Common characteristics of new intelligent optical hardware are:

* it is designed for both voice and data;

* it deploys in mesh and star configurations, as well as in rings;

* it allows carriers to remotely configure hardware and provision services; and

* it scales efficiently in service provider networks.

In addition to Intelligent SDH/SONET, Infonetics believes that the key technologies in the evolution of the metro network are:

* WDM metro core and access: new equipment using WDM for transport around the metro area; designed especially for the enterprise last mile, it allows service providers to provide multiple services over the same fibre topology.

* l0G ethernet: further widening the bandwidth advantage of ethernet solutions, and also providing an ethernet technology for the metro core.

* MPLS: adds a layer of traffic engineering capabilities to IP/ethernet networking and will be a key instrument for offering multiple business services such as VPNs, virtual TDM lines and transparent LAN services in the metro area.

* Resilient packet ring (RPR): an emerging standard that takes advantage of the physical fibre ring topologies common in metro areas, putting ethernet packets in rings to offer the resiliency of SONET and the data efficiency of ethernet.

* Passive optical networks (PONs): bandwidth travelling over fibre is shared among multiple users via the use of splitters. This offers an economical alternative to running dedicated fibre to home or business customers.

Ethernet can run over the network in various ways and in making their technology choices, service providers will need to decide which of several emerging forms of ethernet will offer them the best combination of efficiency, flexibility and resilience. Ethernet-over-DWDM is one option. Although not yet a standard it is a method used by service providers to extend the distance between ethernet switches from LAN to MAN -- so far only Atrica has announced DWDM interfaces for their optical ethernet products.

Ethernet over MPLS over fibre is another approach and one which is being standardised by the Metro Ethernet Forum to add protection paths. Other ways of delivering ethernet include ethernet-over-SONET/SDH according to ITU standard X.86, and ethernet-over-SONET framing using GFP (general framing protocol].

Future metro scenarios

Figure 2 shows what Infonetics Research believes will be the response to the growing range of faster access options will become available to the residential and business market, and the metro core market will simplify in the number of layers while intelligence continues to be added. This will enable the metro network to become more adaptable to changing ethernet- and wavelength-based demand.

From our tracking of the optical market space, we can see that an interesting trend has emerged: the closer to the customer the faster spending is increasing; carrier spend on the edge of the network is holding steady or increasing on a quarter-to-quarter basis. Infonetics Research's quarterly worldwide market share and forecast service -- Intelligent Optical Network Hardware -- showed that worldwide revenues for intelligent optical metro network hardware hit US$1.02 bn ([euro]1.04 bn) in 10 2002, a 15 per cent increase from 4Q 2001. Optical CPE is shown to be the fastest growing category, reaching $129 m ([euro]132 m) in 1Q 2002, a 21 per cent ncrease from over 4Q 2001. Metro edge equipment expenditures rose 16 per cent, metro core/regional equipment expenditures rose 12 per cent, while optical long haul declined 28 per cent. Submarine optical expenditures also dropped by 38 per cent during the quarter.

However, whilst we forecast that the market for intelligent optical equipment in the metro network will continue to grow healthily, there is a long-term migration process that involves gradually moving services off SONET/SDH and onto wavelength-based transmission. Moreover, traditional circuit-switched voice remains a key cash stream for service providers despite the fact that margins are dropping and that it is a smaller percentage of traffic than it was. Thus, investments in technologies such as SONET/SDH will be protected and vendors of newer, packet-based technologies will have to appreciate that. SDH and SONET will need to be supported for many years yet while service providers and vendors establish and grow promising new services such as gigabit ethernet as well.

But for the near term, no single network technology will dominate metro networking --the transition from legacy to new technologies is just beginning.

Richard Webb, European market analyst, Infonetics Research. He can be cantacted at [email protected]

Infonetics Research's Optical Networks Market Share service covers the hardware, applications, and services used in legacy and intelligent optical networks (www.infonetics.com)

COPYRIGHT 2002 Horizon House Publications, Inc.
COPYRIGHT 2003 Gale Group