THE H-REPORT:

ATTACK OF THE $500 KILLER NETWORK COMPUTERS

Time-Warner Cable's Full Service Network

by Christy Hudgins-Bonafield

Interview with Jim Ludington, vice president of technology for Time-Warner Cable's Full Service Network (FSN). The FSN is the first trial to use server-based technology with a cable network to deliver video movies and games on-demand, as well as home shopping and other interactive services.

Time-Warner's technology partners include AT&T (provided an ATM switch); Scientific-Atlanta (manufactured customer terminal with Toshiba and supplied headend electronics); Silicon Graphics (client operating system, applications and headend servers). Andersen consulting serves as systems integrator, and network management is provided by Objective Systems Integrators' NetExpert. Investment partners in the Full Service Network include Time-Warner Inc., U S West, Toshiba Corp., and ITOCHU (the world's largest trading group).

The Orlando-area project, aimed at 4,000 households, got off to a slow start, but is now picking up steam. Network Computing's Christine Hudgins-Bonafield explores whether a $500 cable client is feasible given Time-Warner's experiences.

Q: I understand that the client-side boxes you use in the Full Service Network cost $3,000. Is a $500 box feasible?

A: The $3,000 figure is inaccurate.

Q: Then what is the figure? Is that too much or too little?

A: I can't say whether it's too much or too little. We are expecting the cost to come down drama tically. I've been told by vendors that a box in the $300 to $500 range isn't unreasonable in three years or less. You have to realize our box is a prototype.

Q: Do you believe a $500 box is feasible?

A: Absolutely. What you are looking at with the Full Service Network, the box we are using today, is the testing of services you'll use two years from now. This box and the services are becoming possible. They will structure the network and what it does. Whether these boxes cost $250, $500 or more depends on what interactive entertainment becomes as well as other services. The rest of those services may or may not become part of that box.

Q: What will be required in these boxes?

A: The marketplace will decide.

Q: Who provides the prototype boxes you are now using?

A: A partnership of Silicon Graphics Inc. (SGI) and Scientific Atlanta. Scientific Atlanta is the manufacturer and a lot of the platform is a brute force integration of Scientific Atlanta's cable converter and SGI's platform at the customer site. SGI also provides server hardware and software and modulation equipment.

Q: Do you need a full client-side operating system for video streaming?

A: If you want pure MPEG decoding, I suspect you don't need much on the client side. But if you want video, audio and graphics and interactivity, it raises the level of the client-side application box.

Q: Is there an operating system today that is well suited to this task?

A: Not today. We basically built the OS we use today. We built it on both sides, taking pieces of what worked for the computer and cable TV platforms.

Q: Will you continue with your homegrown technology?

A: We want to get what the OS will do. We don't want to be the developer of it.

Q: Who is optimizing operating systems for this purpose?

A: Microsoft, Oracle, our Time-Warner operation and a few other trials. Sun. Those are the folks hacking away at it.

Q: Who has the le ad?

A: Time-Warner. Our true advantage is understanding the challenges we have to overcome.

Q: What is the biggest technical challenge?

A: Integration. [This technology] was infantile two years ago and is barely established today. We had a heck of a time integrating the computer side with switching and cable.

Q: Do any of the vendors addressing these issues have an advantage?

A: We haven't focused on the others yet.

Q: Do you need a full multithreaded, multitasking OS to support interactive applications?

A: If you are doing true interactivity, multithreaded and multitasking OSes are important. The big picture perspective is that you can break services apart and say: "Here's one for the TV and here's one for the PC." What will happen is that we'll have services rising up on both platforms and starting to cross over. For example, I don't want to turn my PC on when I get out of bed in the morning, but I want a stock quote service. I turn on my TV set and I get it from the same service provider. I think services will blend together. That's how this will come together. It's not that the PC overpowers the Internet. The clear winner is the content and service provider.

Q: Is the Java language important to you?

A: Not specifically to me. I don't know technically what it is solving for me. Interoperability and portability is terrific once we get to that point, but this sort of thing is extremely difficult to do.

Q: How do we get from here to a multithreaded, multitasking box for $500?

A: I don't know that you will get one box that's interoperable for video, gaming and information services and does that in a sort of multitasking environment. I don't know if you will get that in three years. You do get this entertainment piece to get information and entertainment and another piece for the PC. We may have one box for each, or boards for each. We won't have a nice integrated solution in three years. There won't be a ki ller box, OS or board. [You'll be able to access PC online services from one room and one device and access interactive TV from another and telephony services will also be thrown in.]

Q: Samsung and Lucky Goldstar have been mentioned as potential box providers. What do you know about their involvement?

A: The closer this gets to being a commodity is when you will see the entry of companies like Samsung and Lucky Goldstar. I don't even think we'll have an RFP until well into next year.

Q: Oracle has promised to have a prototype by early next year. What interest do you have in that?

A: It's exciting for us...to see the provider of a box that actually works with an OS and an integrated solution.

Q: How do you now use Oracle's technology with your systems?

A: It's our backend database server. One of the main reasons we're in this business today is to understand all the uses of our Full Service Network services. To do this, we need an extremely effective way to capture transactional data. We use Oracle for subscriber management, business support. It's the back-end solution we are billing from.

Q: Will you use Oracle's multimedia server?

A: We are not and have no plans in the near future to use that. We have a solution working today. We'll learn what we did right and wrong in the network architecture, the user interface, and so on, over the next 18 months.

Q: Are the most important lessons learned to date?

A: Systems integration. We had to bring together three different technologies, cultures and geographies-and make a simple transparent service. To the consumer it's TV. We learned to keep focused on the fact today it's TV, but to provide a flexible system. This was an extremely rigorous process. I think we've done a terrific job in pulling this together.

Q: What are some of the pros and cons of Larry Ellison's vision for caching video at servers throughout the network?

A: As you look at an architecture that gr ows, how you distribute is to serve lots and lots of people is important. If there are high-bandwidth files that are taking a lot of hits, you put it close to the user. To do this and move assets and determine which assets to move is a science. It's a true advantage we'll have over postulating about it. You'll have to put data in lots of places, depending on the network, usage and class of service. Today we aren't caching much. Our server is in one place. But we know how to move this information from one place to another and make it available. That's the model we are using to look at the next-generation architecture.

Q: I believe your trial calls for 4,000 households. Once you've achieved that, what will you do.

A: Next year we'll test packaging and pricing, more than the technology. We'll be in more maintenance mode. We don't know all the services we'll offer right now. We do have the News Exchange [from Time Inc.] and we will transmit data from New York to Orlando for that service.

Q: When will you move to other areas of the country?

A: Our target is Orlando for the next couple of years, but we are preparing for the day we roll out beyond Orlando. In the interim, we are upgrading all our plant with fiber. By 1998 we'll move the Full Service Network beyond Orlando. Being nationwide by the end of 1998, we'll have 85 percent of our base completed. It's a process we started a couple of years ago.

Q: Oracle tells me that the caching approach works much better in cable networks than telephony nets. Why is that?

A: We have more bandwidth in cable, and cable doesn't need to deal with all the regulatory issues. In Florida we are laying in residential telephony service and high-speed access to online services. We are ramping up now. We already provide alternate access. We won't provide the telephone service until the first of the year. We've been providing alternate access for seven or eight years in some cities.

Q: How much emphasis are you putting on provid ing cable-based services for businesses?

A: We are concentrating on the residential market today. We'll get to 4,000 homes this year. Businesses could be Phase II...The best examples of where we'd go first with businesses would be the school systems. We'll connect schools with homes and ultimately provide access to businesses to access experts in business to enhance education in the school. Also, parents at home will be able to understand what homework assignments are. It's too soon to say when we might be ready to launch schools and businesses, but hopefully next year we'll be well on our way to formulating plans or getting this installed.

Q: How much bandwidth is there from the home in your trials?

A: It's 1.5 Mbps shared by 25 users from the home and 750 MHz downstream, but the shared 25 users is flexible.

Q: When will you be offering Internet access?

A: We already are in Elmira, N.Y. This is separate from the Orlando trial. It's just Internet access in this location and telephony in Rochester, N.Y. The ultimate goal is to layer them in over the same cable network.

December 15, 1995