Utilizing wholesalers' bandwidth supply for on-demand growth.
BY KEVIN R. COTTRELL, Cogent CommunicationsVirtually all of the guidance issued by telecommunications companies over the last several weeks warns that tight cash will compress revenue and earnings over the next couple-or more-quarters. Equipment vendors specializing in optical gear have been particularly cautious, perhaps because these companies just can't win. About a year ago, several fiber giants reported revenue shortfalls that were attributed directly to their inability to keep pace with demand. Then, carriers had the cash, but vendors didn't have the product. That strategic oversight was soon corrected by investing capital to ramp up fiber and router production capacity.
What a difference a few months make. Today, optical equipment vendors are saying 'here's the product, show me the money.' So what's changed? According to financial experts, many factors are conspiring to create a capital crunch in the communications sector. At the same time, bandwidth demand hasn't slowed. Users' appetites for capacity remain voracious. Certainly, some analysts are quietly suggesting that a bandwidth glut is on the horizon, but they are in the minority. The bulk of the evidence continues to suggest aggressive growth in the need for fat pipes carrying more and more data.
So, while many service providers are strapped for cash, customers want more bandwidth. That capacity has to meet the reliability agreed upon in service-level agreements, requiring carriers to deploy even more pipes strictly for backup functions. What's a competitive local-exchange carrier (CLEC) to do?
One solution is to start from scratch and build a network that's designed specifically for packed-switched data. On the face of it, that sounds wildly expensive, but the next generation of all-optical networks is surprisingly cost-effective. For starters, a data-only network is far less expensive than a converged voice and data network. Eliminating a whole layer of voice-enabling equipment can cut bit transport costs by as much as half. Moreover, next-generation all-optical networks will deliver bigger pipes for less, especially if the network is not wedded to SONET technology. DWDM solutions both at the metropolitan and backbone levels complemented by the introduction of new terabit-class routers will get rid of a lot of the baggage typically associated with SONET and affordably address capacity requirements for the foreseeable future.
Of course, as every carrier knows, there's a lot more to building a network than initial capital costs. Ongoing operational costs can break the bank. So assuming that carriers are unable to go for a long-term solution by building a data-optimized packet network, then one alternative to mitigate the situation is to add capacity incrementally. Not surprisingly, though, the result is often a case of too little too late, while the cost of the interim solution can be far greater than the benefit. That's because bandwidth is hard to provision and expensive to support, particularly in existing voice-oriented legacy networks. And sometimes a network extension takes as long to implement as it takes to build a new network, especially when geography demands that the carrier work with multiple suppliers, which leaves the CLEC at the mercy of disparate schedules and timelines.
That brings the carrier to alternative #3. If building from scratch is not in the budget and extending the network incrementally is just an expensive band-aid, then leasing already-available capacity is a logical option. Wholesale providers with their own DWDM-based OC-192 (10-Gbit/sec) backbones can lease incremental capacity instead of entire fibers to CLECs looking to expand at minimal risk. Thus CLECs can upgrade to OC-192 rates for a price that is not only comparable to OC-3 (155.52 Mbits/sec), but also distance-insensitive. By dealing with a single facilities-based vendor, CLECs are assured rapid, streamlined provisioning, resulting in service in just a few weeks rather than several months. Data wholesalers that own end-to-end networks can offer CLECs scalability, speedy deployment, quality of service, and lucrative pricing.
Growing too fast and getting stuck with excess capacity result in an inability to rationalize sunk costs. Conversely, growing too slow and leaving customers hanging defer revenue, or even worse, enable a competitor to capture that cash from frustrated customers. The cost of not doing business or doing it poorly because a CLEC doesn't have the time or money to get its infrastructure up to speed is a lost opportunity.
Growing a gigabit at a time resolves this circular argument. Rolling out Internet Protocol (IP) services over Gigabit Ethernet is a very attractive option for CLECs and carriers. An example of an Internet-optimized all-optical metro architecture is shown in the Figure. A 16-wavelength, OC-48 (2.488-Gbit/sec) metropolitan fiber ring links up to eight multitenant buildings using bidirectional transmission on a single strand of fiber. Two 2.488-Gbit/sec connections travel in opposite directions on the metro ring to a drop point at a lit building, providing redundancy.
With 5 Gbits/sec of total capacity, the service provider can provide dedicated Fast Ethernet (100-Mbit/sec) and Gigabit Ethernet (1-Gbit/sec) connections to customers in each building. The wholesale service pro vider off of the same regional network infrastructure can quickly deliver additional capacity beyond the initial 5 Gbits/sec, as de mand warrants.
In the all-data, all-the-time IP world, SONET's voice-oriented bandwidth increments don't match up with Ethernet speeds. Installing or upgrading SONET circuits and related termination equipment can take several months and easily exceed the available capital. In contrast, DWDM-based Gigabit Ether net solutions can be rolled out fast-usually in just a few weeks-not only because of the technology, but also the inherent efficiencies of dealing with a single wholesaler. A facilities-based wholesaler that owns both a nationwide backbone and metro rings has the ability to meet CLECs' demands quickly. A single national vendor, a single project plan, and a single point of contact can speed up deployment and improve accountability. That in turn enables the CLEC to make-and keep-service commitments to its users.
Table 1 illustrates how long it typically takes a traditional interexchange carrier (IXC) to meet a CLEC's various capacity requirements versus the timeline that some new data-service wholesalers are adopting.
A highly efficient and scalable Layer 3-protected IP-over-DWDM network virtually eliminates quality-of-service (QoS) issues (see Table 2). In short, this architecture enables network operators to throw bandwidth at capacity and quality issues to make these problems go away.
Wholesalers can offer unprecedented amounts of affordable bandwidth, because these new infrastructures are very efficient. Oversubscription and shared access are eliminated. Every customer can have a dedicated, protected Ethernet connection to the terabit router. Even routing customers' data becomes more efficient. Since the high transport costs associated with traditional networks are eliminated, the provider can ditch hot-potato routing; the customer's traffic can ride on a dedicated network for as long as possible before jumping off into the IP cloud. Cold-potato routing means fewer network handoffs and therefore less latency and better performance.
Today, CLECs can upgrade to OC-192 rates for a price comparable to OC-3. Carriers can therefore meet customer demand for high bandwidth without busting the bank and the lifetime obligations implied by ownership. Most data wholesalers are requiring no more than a 12-month service commitment. Table 3 shows a comparison between rates charged by a hypothetical IXC and a hypothetical data-service wholesaler.
Leasing capacity on a Layer 3-protected IP-over-DWDM network means CLECs not only get the latest technology, but also get much-needed capacity quickly and affordably. Scalability and speedy deployment allow carriers to plan for demand. Providing customers with demanded bandwidth in just a few weeks verges on real-time deployment. That also means CLECs are no longer infrastructure-driven or infrastructure-constrained. Rather, these carriers can be truly market-driven because granularity enables them to grow as required by the end customer, not as dictated by technology.
Dealing with a single national data-service provider means the incremental bandwidth is deployed practically on demand. Yet, fast deployment does not compromise QoS. In fact, these next-generation IP-optimized architectures perform better than traditional networks in terms of network availability and packet delivery. Further, there is no shared access, no oversubscription, and no hot-potato routing.
Finally, the price/performance ratio of leased capacity is unparalleled. The new Layer 3-protected IP-over-DWDM infra structures are enabling radical cost reductions. That means a new breed of data-service wholesalers that own their long-haul and metropolitan-ring networks can pass along the savings and offer CLECs unheard of pricing. A pricing model that creates a much lower cost per bit mile will not only completely alter the way CLECs grow their networks, it can also radically improve the way end customers use their networks and the Internet.
Kevin R. Cottrell is director of wholesale and carrier sales at Cogent Communications (Washington, DC). He can be reached at ([email protected]).