LVLT and everything over IP. This article provides a good review of several interesting topics from Internettelephony.
SOFT IN THE CENTER
Level 3's revolutionary network design is reshaping an entire industry
VINCE VITTORE
About 20 miles northwest of downtown Denver, past the first few massive construction projects on Highway 36, sits a cluster of five buildings in various stages of creation where Level 3 Communications is staging its revolution. Last year at this time, the site--seemingly light years away from the Denver Technological Center south of the city where most competitive carriers reside--was not much more than a large field with a spectacular view of the Rockies, adjacent to another large field that was soon to become a resort golf club. Among some local telecom vendor sales reps, the site simply was known as the "mud pit."
A year later, the golf club is complete, Level 3 already is eyeing nearby fields to house its burgeoning staff, and company execs Kevin O'Hara and Ike Elliott have just overseen the launch of the company's first wholesale voice product. Such is life for a company that slightly more than two years ago existed as an idea but today is reshaping and exploding the core beliefs that have held the telecom field together for a century.
At the center of the new tenet is a reliance on IP technology and a distaste for most, though certainly not all, things circuit-switched. Under the new soft-switch architecture, core switching functions no longer are handled in a single unit but instead are parceled out to multiple boxes based on their specific job. The effect is a pure IP network, but one that interoperates with the existing public network.
"The service interfaces with traditional circuit switches by making our IP network one big tandem switch, even though we're not using any circuit-switching technology," says Elliott, senior director of voice and access network engineering.
Two years ago, such a statement would have bought Elliott a one-way ticket to public network oblivion. Today, it is almost a mantra for the revolution that is forcing traditional circuit-switched vendors to buy any company with IP expertise, rearranging the priorities of virtually every carrier and ultimately leading to an array of new services.
Fiber to spare
Directly related to this new technical model is the reality that everything carriers do must be based on the new economics of a competitive telecom industry where customers can walk away at any given moment. Translated into Level 3's day-to-day operations, it means creating a worldwide network for a lot less money than circuit-based carriers.
"The objective is to build the lowest unit cost," says O'Hara, Level 3's executive vice president and chief operating officer. "Everything we do is an economic decision." In the core of the network, that means putting in softswitches for 40% to 45% of the cost of a circuit switch.
For incumbents, that fact alone is a wake-up call. But O'Hara says Level 3 isn't basing its model solely on cheaper cores. Level 3's economic model works on two basic assumptions: The unit costs for telecom are dropping every year, and the opportunity for price flexibility on traditionally fixed costs, such as fiber, are high.
"Many other carriers are building into their economic models fixed fiber costs," O'Hara says. "We think that is wrong because fiber itself is improving, and the cost is dropping. Everybody is in a variable cost game. You're penalized for not putting in fiber."
Putting glass in the ground with a vengeance is exactly what the company has been doing. Instead of building out a network of two or three strands across the country the way most backbone providers do, Level 3 is constructing a network with 12 strands of fiber--six for its own use and six for leasing to other carriers--and the company uses only one of the 10 to 12 conduits it's burying, leaving plenty of room for expansion (Figure 1).
Moreover, the company isn't marrying itself to any particular type of fiber, generally using the latest fiber regardless of previous generations already in the ground. "Each generation of glass gives you another opportunity to lower your unit cost," O'Hara says. Extrapolating the improvements up to the most recent generation of fiber, Level 3 has an 80% cost advantage over carriers that put in single-mode fiber five years ago. Table 1 shows Level 3's network construction schedule for the U.S.
"The dynamic of pushing costs down is foreign to incumbent carriers," O'Hara adds.
Protocol proctor
Nowhere does Level 3 differentiate itself more from incumbents than in its network architecture, where the decentralization of switching functions and packetization of voice at the edge are reshaping the industry. On the network edge, the company takes analog voice and converts it into packets as quickly as possible, using a variety of protocols. Indeed, if Level 3 has had any immediate impact on the public network market, it's in protocols.
Not only has the company shepherded the development of key protocols such as H.323--taking it from its early stages as a protocol designed for video communications over a LAN to a key ingredient in the embryonic stage of IP telephony--but Level 3 has served as almost a de facto driver of higher-level protocols. In looking for evidence of such leadership, one only need turn to the Softswitch Consortium, of which Elliott was a key founder and now serves as chairman. The consortium, which many vendors credit with legitimizing the Level 3 architecture vision, is designed to promote compatibility and interoperability among IP voice network elements, specifically focusing on five protocols. The group, which now counts 86 members, also serves as a think tank for advanced application development.
"We saw the group as a real win-win in both forming and driving it," Elliott says. "We had such success with the [IP device control] forum [that] we wanted to continue that--especially focusing higher in the stack. We tend to adopt protocols that have already been developed, but as is the case with most specs, you need to take it one step beyond."
Still, the group remains in the organizational stage and has much work to do, he adds. "We need to get to the point where we settle on a simple set of protocols for those access networks. We haven't achieved that level of interoperability yet."
By setting an objective of interoperability through a broad industry group, carriers should have a greater selection of vendors, further driving costs down. In fact, Elliott admits there was some self-interest involved in getting the group moving. "Certainly the Softswitch Consortium mission, which is to enable Internet innovation, is something that will benefit Level 3. We like to have a greater selection of equipment," he says.
Conversely, the early involvement of a carrier with the financial backing of Level 3 spurred many large traditional telecom vendors to get involved. Lucent Technologies, which signed a four-year contract with Level 3 last June, already had the basics of a softswitch in development but pushed it onto the fast track because of the carrier.
"Clearly, I think the softswitch would have evolved, but my experience has been when a customer like Level 3 wants to engage early, you take the challenge," says Lance Boxer, president of Lucent's communications software group. Previously, Boxer worked closely with Elliott and some of the other top Level 3 technical staff while at MCI (before it united with WorldCom). "It's much harder to fund projects with no specific customer. The product would have evolved, but clearly the early product was crafted around the needs of Level 3."
In fact, Boxer gives specific credit to Level 3 Chairman and CEO James Crowe for forcing traditional telecom vendors to rethink the way they do business. But the transition was not always easy, Boxer adds.
"Jim pushed us very hard to think very different," he says. "The first thing he did was push us with different objectives. Most customers demand things in broad terms. Jim was very specific. Ike [Elliott] and Jim and a lot of others had the audacity to challenge the industry. It's a little bit problematic for us, but on the other hand, that's the way the world is."
We've got it licked
Ironically, with all the revolutionary thinking and network scheming coming out of Denver's suburbs, one of the biggest initial objectives for Level 3 is to look exactly like existing public network operators, albeit using its own IP backbone. And until last month, when the company rolled out its first IP voice product, Level 3 looked more like a large Web-hosting company. Its (3)Voice service initially is targeted at resellers that want a public network service at IP-type prices.
"We're not rolling out a voice-over-IP product just to have one. It's to have a product that is indistinguishable from the public network," O'Hara says.
Getting to that public network par point has meant acting differently, even from its IP voice carrier brethren, many of which have approached the market with dirt-cheap pricing and questionable quality of service. It means only sending voice over its own backbone and not using any compression, which in turn requires more backbone bandwidth.
"The quality of service problem is something that we've licked," Elliott says. "Only going over our backbone is a very important distinction. We don't use any compression on our network for this service because compression is a source for quality degradation."
Not everything is perfect, particularly when trying to control quality over access networks it doesn't own, he adds. "The big issue is access networks and how to maintain quality over those access networks."
In certain cases, such as markets where the company doesn't have a local access point, Level 3 will hand off calls to the public network. Still, the company has made enough of an impact in the quality issue to make it virtually disappear in customers' minds. "The neat thing about it is that we've finally made voice over IP a purely economic decision for the customer," Elliott added, noting that the company has been using the service with employees since April 1999.
Just when--or even if--Level 3 will launch a retail-oriented voice product is unknown. Like the technology behind the company, much of the strategy has changed since its founding in late 1997.
In those heady early days, armed with $3 billion in financing, the company appeared intent on competing with long-distance and local carriers. But that changed during the course of the first 18 months as Web-centric customers flooded the market looking for hosting services. And with the introduction of voice, Level 3 could be positioned to provide a variety of services across multiple markets, though it's trying not to be everything to every customer.
"We want to get very good in one horizontal layer," O'Hara says.
One area the company mostly will play to is the application service provider (ASP) market. In fact, the addition of voice likely will prevent the company from competing as an ASP itself. "Increasingly, customers want to buy bundles but don't want to provide everything," O'Hara says. "ASPs want to do a few things only but want to do them well. We think voice itself is an application. How you enhance voice is a good application for ASPs."
Still, the company may dabble in some of those applications.
"There aren't really tech barriers," Elliott says. "There are some time-to-market barriers. We're going to see an explosion of innovation for Internet-enabled voice applications."
Pinning Elliott down on specific applications is virtually impossible, though he admits to interest in unified messaging.
"I see [a] real opportunity for apps in the business market that save people real time and make them more productive," he says. "Certainly unified messaging is a piece of that. Certainly speed dialing is a piece of that. I don't know how many times I've looked up someone's name in Outlook [rather] than turn around and manually dial their number. The applications that we can envision are just the tip of the iceberg."
And as wireless carriers move into the IP realm, Elliott envisions Level 3 working with wireless carriers and offering specific value-added services. "One of the real advantages of this voice-over-IP movement is how the IP layer abstracts. The Holy Grail is to get the same type of services to work regardless of the access device."
Regardless of which strategic direction the company moves, market environment will play a major role.
"We don't think we're going to be perfect," O'Hara says. "We think we're going to be better than the competition."
--------------------------------------------------------------------------------
More letters for the alphabet soup
Vince Vittore
If there's one thing the telecom field doesn't need, it's more acronyms. But for Level 3 Communications to move from concept to reality, many key protocols had to be developed, adding even more names and acronyms to the lexicon.
While not joined in any formal relationship, the Softswitch Consortium has been heavily influenced by Level 3's development. "The way Level 3 looks at all these protocols is that we don't choose winners," says Ike Elliott, senior director of voice and access network engineering for Level 3. "We try to get as much traffic on the network so we try to hit all the major protocols."
The group currently is working on five protocols:
H.323. Though Level 3 is not using it within the network, H.323 is the cornerstone technology for the transmission of voice-over-packet networks, specifying the components, protocols and procedures. In the IP telephony environment, H.323 is used in a variety of devices, including gateways to convert and compress analog voice into packets.
Session initiation protocol (SIP). Often mistakenly identified as a direct competitor of H.323, SIP is a protocol used between end systems and proxy servers. Among the services it enables are call forwarding, calling number delivery, personal mobility (the ability to reach a called party under a single, location-independent address even when the user changes terminals), terminal-type negotiation and selection, terminal capability negotiation, authentication and blind and supervised call transfer. In Level 3's architecture, SIP is used as a protocol between soft switches.
Real-time transfer protocol (RTP). RTP, according to the Softswitch Consortium, "is primarily designed to satisfy the needs of multiparticipant multimedia conferences." In more general terms, the protocol "provides end-to-end delivery services for data with real-time characteristics, such as interactive audio and video. Those services include payload type identification, sequence numbering, time stamping and delivery monitoring." RTP also allows for data to be shipped to multiple destinations using multicasting technology. It does not guarantee quality of service, but it allows a receiver to reconstruct the sender's packet sequence.
SGCP/IPDC/MGCP. Among the most important in the Level 3 scheme, this group of protocols, which has evolved into media gateway control protocol (MGCP), provides call control architecture similar to--though not exactly like--SS7 protocol functions in the circuit-switched world. Though MGCP is well into development, Level 3 initially has used IPDC. "The MGCP is something that we're doing a lot more interoperability testing with through the Softswitch Consortium," Elliott says.
Real-time streaming protocol (RTSP). RTSP acts as an application-level protocol for controlling the delivery of data with real-time properties such as voice. RTSP acts as a "network remote control" for multimedia servers. According to the consortium, "RTSP establishes and controls either a single or several time-synchronized streams of continuous media, such as audio and video. It does not typically deliver the continuous streams itself, although interleaving of the continuous media stream with the control stream is possible." |
