Getting a handle on interoperability
Teamwork. What a concept. Supposedly, that's what Boy Scouts, Girl Scouts, Little League, high school football, the cheer-leading squad, Army (Navy, Air Force, Marine, Coast Guard) boot camp, and even marriage are all about. Yet, practically speaking, the virtues of teamwork have never really been extolled, let alone practiced, in the cable industry to any widespread, sustained degree. That is, until now.
On the rolling hills between Denver and Boulder, Colo. toils a somewhat disparate, but very talented, group of engineering professionals (representing both MSOs and vendors alike) who have suppressed their naturally competitive natures to form a dynamic technology team. This team may be laying the groundwork for a seismic change in the way the industry gets down to business and meets the competition head on.Standard bearers
Over the last few years, as cable data modem technology has developed from a mere pipe dream to a practical reality, operators have virtually swooned with anticipation over the new revenues that high-speed data services could bring to the industry. Yet their giddy fall to the floor has always been broken by one harsh reality—money. Money to develop, manufacture and deploy these high-speed modems throughout North America.
Standardization became the rallying cry. If the modems had a basic degree of technical conformity, consumers could buy them and take them whenever and wherever they moved. That's why "plug-and-play" soon became the mantra for those who were determined to move massive amounts of money off operators' already overburdened financial books.
Taking the stage at a Western Cable Show just a few years ago, a virtual chorus line of operators and vendors alike vowed to come up with a high-speed spec that would make "plug-and-play" a reality. That brought about the Multimedia Cable Network System (MCNS) group and its resulting Data-Over-Cable Service Interface Specification (DOCSIS).
According to Bob Cruickshank, director for digital network technologies at CableLabs, initial reaction to DOCSIS was not all that enthusiastic. "When we met with all the vendors for the first time," says Cruickshank, "in what we called our Interoperability Summit right after NCTA last year, the biggest beef from the vendor audience was, 'How the hell are we supposed to test this thing? There's no tester available. There's no test plan. You guys should not have released a spec without a test plan.' And we said, 'Oh, great. You wanted us to wait another couple of years while some competing technology just takes us out?' Obviously, we couldn't do that. We had to follow a new model.
"And what we proposed to them was that each one contribute part of what they had into what we called our incubator. What they had went everywhere from intellectual property to vendor resources such as engineers or manufacturing expertise or whatever."
"So, we opened their eyes to the idea that we could cooperate. That we could create the business. That we could create the economies of scale and get things down to a commodity level.
"After that, they could shoot back to their Darwinistic survival-of-the-fittest mode. But, we got them to agree to at least cooperate so that they all didn't waste their time while some other competing technology took them out, because they took too long to get it done."The logistics of cooperation
While everyone involved finally realized there were definite advantages to this teamwork concept, old habits were initially difficult to overcome. To ease the transition from conflict to cooperation, Cruickshank was tapped to represent MSOs' interests in the effort, and modem pioneer Rouzbeh Yassini, CEO of YAS Corporation and better-known as the founder of LANcity Corp., was brought on as an executive consultant to CableLabs to coordinate vendor participation.
"The general idea for us," says Yassini, "is to get the job done. And the job is to make sure we have one retail product out the door. We're not forcing a schedule or product maturity to happen, or forcing arbitrary tests to be passed. And everything is piling into an Acceptance Test Plan (ATP)."
While many feared this cooperative effort might turn into a nightmare of lost R&D and competitive advantages, the intellectual challenges of this daunting effort quickly overcame those concerns. Says Yassini: "The first impression you always get from the vendors is, 'Oh my God, this is our confidential stuff. I'm not going to let my competitors see this.' So, on the first day of work you can see them come in, almost like they've got their machine guns loaded. It seems like they're hovering over their boxes, keeping track of who's getting too close.
"Yet, by the end of the week, the engineers are totally involved in their work. There's all kinds of troubleshooting going on. Every engineer is helping every other engineer to find solutions. There's no other way to do it."
Yassini and Cruickshank have worked hard to keep things simple, open and on track. The rules of engagement, as they call them, are concise (one page), direct and unequivocal, says Cruickshank. While participants aren't required to sign the rules literally, they're bound by them, nonetheless. "We knew," says Cruickshank, "we had to create a protective environment, a technological Switzerland, if you will, so that people would feel comfortable. So, we set up guidelines and rules of engagement. And we just tell them that if they violate them, they're out of the first certification batch.
"The rules say you can't access another vendor's equipment without their permission or without them being here. It says that if you're a vendor and you hook up with another vendor, and you discover issues between your pieces of equipment, you both own those issues and you'll share them with CableLabs. But nobody else will know about them. In other words, any dirty laundry stays between the people involved."
DOCSIS participants are also utilizing the Internet (www.cablemodem.com). Cruickshank says they created virtual discussion groups for participants through exclusive e-mail lists. Those interested in physical layer (PHY) issues have their own list, while those interested in MAC and higher level issues have their own list. The e-mail lists go on to cover a wide variety of issues. All correspondence in each list is archived so those coming into any particular discussion later can get up-to-speed with the whole group quickly.
In addition, there are weekly and monthly meetings. And, if a particular issue pops up, a special task force is set up to deal with it. There's also a visiting engineer program that a number of companies (e.g., Toshiba, Thomson Consumer Electronics, Scientific-Atlanta, Motorola, Shaw Communications and MediaOne) have taken advantage of.
This cooperative talent pool has proven to be invaluable in keeping the DOCSIS effort on track. "We're in a situation," says Cruickshank, "where even if we had the money, we couldn't hire the people we need. Because what we really need are the people who are developing these modem products.
"So, these people are on site here. In some cases, it's 50 percent of the time. In most cases, it's 100 percent of the time. They're treated just as if they were regular staff members on the project. They've all got responsibilities. And that's been extremely useful."Testing, testing...1,2,3
To make a "plug-and-play" certification process a reality, the CableLabs consortium has established an eight-step testing path. It has essentially completed the first four rounds and is currently working on the fifth test area. If things stay on track, the entire suite of eight tests will be completed by autumn of this year.
Steps one through three, says Yassini, covered the "bare minimum requirements" of an open-standards spec. This included tests on the PHY itself and tests for the MAC/PHY interface. Step number four dealt with chip-level operability.
"At this point," says Yassini, "we've proved (been able to test) the subsystem elements of the DOCSIS spec. What this means is that you have the spec; you can develop a chip; you can develop a prototype; you can speak PHY, MAC and a minimum OSS language to build a technology that can act as CMs (cable modems) and CMTSs (cable modem termination systems) by meeting goals one through four."
Test number five, which is currently underway, begins to bring everything together as a system, but still focuses on modem functionality. The goal of step six will revolve around modem configurability or "most of the modem functionality that can be met by hardwiring."
"The goals for these two tests," says Yassini, "are to build a 120-home node network that includes three CMTSs and 120 CMs. How we build this is that we've come up admission and exit criteria." He explains that admission criteria includes such things as symbol rates, operating frequencies, throughput rates, etc. Having these thresholds will save vendors weeks, if not months, in figuring out signal configurations.
Round number seven (currently scheduled to occur in the May time frame) will deal with modem manageability "because SNMP is the last piece of the code that people normally add to their product" says Yassini. Step number eight (slotted for July/August) will put the high-speed data pedal to the metal with formal field tests.
In anticipation of these real-world field tests, Yassini and company expect to put CableLabs' Multiple Services Compatibility Testing (or MUSCAT) Room to good use. Yassini says this 120-home network is "effectively the cable plant you would build for a small town. And, we've got the ability to put digital video, telephony and data on it. It's a controlled environment where you can play with variables, aberrations, lengths and delays, and all kinds of things. It's probably one of the largest indoor instrumented HFC plants in the world."
The first round of certification will likely cover a variety of vendors. After that, individual vendors will go through the process (see Figure on page 28) on their own.
"This first time out of the chute," says Cruickshank, "it's being done in a batch mode. We're bringing all the vendors together and using their peers in trying to identify any problems.
"Through this process, when we take that large network and put 30 of this vendor's modems and 30 of that vendor's modems together, effectively, we'll be using the manufacturers' insights and brain power to find problems with other vendors' stuff.
"Then, what comes out of that batch, is a bunch of product that the engineers collectively feel works well enough to put in the field, as well as documented procedures through the Acceptance Test Plan that details all the things we looked at, how we looked at them, and what we think the right answer is when you run a particular test."
As for the actual seal itself, Yassini says there is no design yet, but "it's going to be something simple and meaningful." He says there are ongoing discussions on branding opportunities that might come out of the certification process as well. This could include raising consumer awareness by using the seal or trademark, not only on the modems themselves, but on packaging and in ads and promotions.
He notes that any revisions of previously certified modems will have to be recertified. And while CableLabs may be greatly involved in the first few rounds of certification, the process may eventually "end up like a UL (Underwriters' Laboratories) affidavit-type situation," possibly supervised by a third party.Lots of progress, a few concerns
Many of the vendors participating in the MCNS/DOCSIS process say they are pleased with the strides being made toward interoperability. "The good news is, that most of the misinterpretations of the spec are more at the higher layers now, more on the software side of things, so that's easy to change, especially as part of the development process," says Levent Gun, vice president of 3Com's Cable Access Division, a manufacturer participating in the CableLabs' DOCSIS effort.
"I think the major hurdle right now is in resolving, or verifying, the interoperability of the various silicon (chip) platforms that the vendors are working with," says Mario Vecchi, senior vice president and chief technology officer at Time Warner's Road Runner Group."...Once that gets established, I think we will have cleared some major hurdles."
"The DOCSIS specification is complex, but I think so far, things have been going pretty much on track," says Chris Grobicki, product management, Broadband Technology Division, Bay Networks. "Yes, there could always be an issue that comes up, but I think that at this point in time, any issue that would slip the time-frame out from '98 to '99 for MCNS modems has not shown itself."
One member of the DOCSIS certification board, Steve Craddock, vice president at Comcast Corporation, is very pleased with how well things have gone and is already focusing on issues that need to be addressed in the next version of DOCSIS.
"I'm absolutely thrilled about the progress that DOCSIS has made," says Craddock. "I didn't, in my wildest dreams, think we would be where we are right now. And I think a lot of the companies that have been participating in the interoperability labs deserve a round of applause. They've really done a lot of great stuff."
But, he notes, because the DOCSIS effort had to "freeze the requirements" to get this initial certification effort underway, compromises had to be made and decisions had to be deferred. "In doing that," says Craddock, "we landed on variable length packet IP, which is not geared for guaranteed quality of service (QoS). What we're doing now is good for the Web surfers, the @Home and the Road Runner subscribers. But, for that next level of services, for example, streaming media, IP voice and video, QoS is going to be a challenge."
The solution, he says, will involve going back into Version 1.0 DOCSIS and "making some changes that will allow us to continue to let the product go to market," but at the same time, allow the MSOs to get some measure of QoS so they can start to offer some of the new services.
And, from what he's seen at CableLabs, Craddock is confident that a solution will be found. "There are some very clever ways to get QoS over the MCNS variable length packet," says Craddock. "There's some extremely bright plans for this. And I'm telling you I'm absolutely impressed with some of the guys up at CableLabs and in the industry who are working this particular issue at this point in time. I mean, it's really good stuff."The retail challenge
Once the certification process is formalized, codified and cranking out seals, the biggest challenge may still lie ahead: Going retail.
Time Warner's Vecchi says that the company's Road Runner Group is looking at a number of different options to put the modems directly into the hands of consumers. To get its feet wet in retail distribution, the company will be conducting "retail-like" tests in one of its markets, offering consumers its "Road Runner in a box," which wraps up an existing cable modem; a PC labeled as "Road Runner-ready," with a NIC card and service software already installed; and an installation manual and instructions. Through this test, "we are exercising the whole business-marketing-installation cycle," says Vecchi.
Once the consumer gets the box home, the onus is on him to plug in and activate the modem, then call Road Runner to activate his account. The expectation, says Vecchi, is that the consumer will do everything.
"If the consumer plugs in the modem and has a problem because the registration process is not working (what our techs refer to as the 'blinking light' problem), then the tech would check to see if the plant is dirty, or if there is some major problem in the home wiring. Our expectation is that the large majority of homes in locations where we have deployed (service) would be able to bring this up on their own."
Dick Day, corporate vice president and general manager for the Motorola Multimedia Group, is another firm believer in the OEM retail model. "One of our end-game visions," says Day, "is that the modem is not going to be an after-market device; it's going to become an OEM device inside the computer. But guess what? It's a personal computer retail model. That's a big difference.
"Technically sophisticated customers will choose to buy it at retail and install it themselves, possibly. And then others, and I'll use as an example my wife, who I always use as my first marketing test, if it's not bundled in and really simple, she's not going to do it. So, it will have to be some type of OEM appliance. Of course, there are some technical issues with that, and that's not on anyone's short-term horizon."
Another DOCSIS proponent, NetGame Cable, is planning to OEM both cable modems and cable modem termination systems. "We don't plan to go to market with our own NetGame CMTS," says Michelle Schulman, NetGame's director of business development. "We have some OEM deals that are in the works with a few different vendors. We're also speaking to computer resellers who are interested in OEMing the (cable modem) product and probably bundling it as part of a package with their computer. That way, they'll be able to offer a network-ready, high-speed PC."
Samsung's David Lin says that, when the time comes, his company will be distributing DOCSIS-certified modems through MSOs, computer retailers and systems integrators as well. "Computer retailers will be the early adopters of these MCNS cable modems," says Lin, who is director, product management and business development, Cable Modem Lab, Multimedia Technology Center, with Samsung. "And as for system integrators, they may be even more qualified than computer retailers, because this is their business."
3Com is pursuing a "micro retailing" approach to modem distribution, says William Markey, director of marketing, Cable Access Division, for 3Com. "We will work with the local cable operator, identify the footprints of available service areas, and then activate our local retail partners to enable distribution," says Markey. In mid-February, 3Com was scheduled to activate its first retail partners by placing US Robotics-branded cable modems on retail shelves in the U.S. and Canada (3Com acquired US Robotics in 1997).
"Generally speaking, there is much hard work ahead to work through all the challenges necessary to make cable modems viable in the retail channel," says Bay's Grobicki. For its part, Bay has already begun working with cable operators on marketing strategies and collateral materials to enable them to start generating demand for the modems, he adds.
By the same token, Grobicki is hopeful that the modem installation complexity will eventually be a thing of the past. "Our experience has been that the complexity is not in the installation of the cable modem. Really, the complexity is in the installation of the Ethernet adapter card, and the software in the PC. That has to get simpler.
"How do you make it simpler? In the way the modem is connected to the PC. Currently, it's Ethernet, but there is already talk of moving to the Intel USB (Universal Serial Bus). That will be a bus that's already installed on the motherboard of the PC. So you would connect directly to that bus, without having to install an adapter card."
Technical standards. Certification. Retail sales and distribution. Anything else standing in the way of cable modem ubiquity? You bet, says Jane Zeletes, vice president of marketing for Hybrid Networks Inc. A 10-year veteran on the consumer electronics retail side of things, she believes there is a whole list of concerns to be dealt with when it comes activating high-speed data service.
"A big issue," says Zeletes, "is the activation process in terms of setting up billing for a service provider once the consumer takes the product home. That whole process has to be in place. And not just activating the product. How does the customer get registered on the network? How does the network acknowledge that and all the billing and all that gets set up? The activation itself is the easy part. It's the support processes behind it that are a concern. This is not a one- or two-month lead item. These processes are extensive."DOCSIS and beyond
The investment in time, money and talent to bring DOCSIS to certification has been astronomical. Yet, the potential revenues are mind-boggling, no matter whose calculator you use. But the overall impact of the effort, says Yassini, goes far beyond little Johnny surfing the Web at lightning speeds for a homework assignment. It goes to the very foundation of an industry that's in the process of remaking itself and the society it serves.
"You know," says Yassini, "the cable industry has been a proprietary industry for 40 or 50 years. But DOCSIS is not just a cable modem. It's really, for the first time, bringing standardization and the process of building an interoperable digital mentality to this industry.
"We're not just creating a solution for the cable modem problem. We're also establishing a process, a procedure, a mentality and a work ethic necessary to take standards and bring them to the commercial market. It will allow the cable industry to build for the next 40 years based on a digital platform. The DOCSIS effort is going to be the foundation for that technology where multi-billion dollar businesses are going to run."