Although home networks are merely geek chic today, new products are emerging to help drive them into the mainstream. These new devices, many based on Linux, will drive an explosion of interest in home networks in affluent, but not necessarily tech-oriented, households. As in any revolution, however, various players are each pushing their own visions of how the technology will work, creating conflicting information for potential customers.

Many important vendors recently created the Internet Home Alliance to help promote home networks. These vendors include diverse players such as networking giants Cisco Systems, Sun Microsystems and 3Com; consumer-focused companies Best Buy, CompUSA, Panasonic and Sears; and chip makers Motorola and Texas Instruments. Although these vendors are working together, each of them is trying to solve a particular problem: sharing a broadband connection, for example, or creating a music network (see “Linley on Linux”, page 186 of October 2000). Indeed, many early efforts will be single-purpose networks driven by a particular vendor's solution. Once home networking becomes established, however, there will be many new uses. For example, the TiVo DVR (digital TiVo video recorder) does an excellent job of storing TV programs on its hard drive for later viewing. But there is no practical way to record programs between two televisions in different rooms. A home network would allow a single DVR to drive video streams to multiple televisions.

Ultimately, I expect the home network to look much like the network in a small office, with a single server connected to a number of clients. The server will have two main components: the residential gateway and the storage server.

Many vendors are touting their expertise in residential gateways, but few are available today. These devices are similar to WAN gateways, connecting the internal home LAN to the Internet, typically through a broadband connection such as DSL or cable modem. Unlike a standard broadband modem, however, the residential gateway has a LAN interface to distribute data to multiple clients.

The storage server will consist of a processor, network connection and one or more hard drives, providing data storage for all devices in the home. This server could back up the hard drives of any PC in the home; store digital photographs, music and video; and stream audio or video content to any network client. It should be programmable enough that it can host applications, such as TiVo, that autonomously obtain broadband content or perform other housekeeping functions.

Client devices include PCs, Macs and Linux systems that use the residential gateway to access the Internet and use the storage server to back up important files or simply as shared data storage (ideally, the storage server should use a RAID design so the inevitable hard-drive failure does not eliminate vital data). Photographs or music files can be created on a PC or downloaded from the Internet, then transferred to the storage server.

Once on the storage server, a photograph can be displayed on any TV screen or monitor in the house via a simple video client. A video client has only a processor, a network interface and a video output. The client downloads an image from the storage server, decompresses it and displays it. The video client also downloads and displays a video stream from the server. Similarly, an audio client connected to an amplifier and speakers can play any music file on the storage server. The audio client has a similar bill of materials as the video client. Either client should sell for less than $100 in high volume, making it cost-effective to have several throughout the house.

Other potential network clients are Internet appliances, non-PC devices that have a web browser and e-mail. Video games connect to the network to enable multiplayer gaming. Even appliances such as refrigerators and thermostats might connect to the home network to share usage information and operate more efficiently.

This architecture centralizes the storage and main compute resources in the storage server, where disk space can be shared efficiently and upgraded easily using internal or external drives—“Honey, can you pick up a few gigabytes on the way home?” Initially, single-function products such as TiVo and AudioReQuest will employ their own hard drives, but these units will ultimately give way to low-cost thin client devices.

Linux is likely to play a key role in this growing network of consumer products. The storage server in particular needs a robust but inexpensive operating system with built-in networking capabilities. Since the storage server is likely to run more complicated applications that organize and manage audio, video and web content, it needs a platform with good software-development tools and standards. Linux is an ideal fit.

The residential gateway may be a simple networking device, but it is likely to run a firewall and perhaps other software as well. Linux may play a role here. Some of the client devices may run embedded Linux as well; the key issue here is maintaining a small memory footprint for reduced cost. Certainly, Linux is a good choice for Internet appliances with web browsers.

One concern with home networks is the choice of physical connection. The most cost-effective choice today is to use a home's existing phone wiring (HPNA). I don't see this as a good long-term solution, because most homes, particularly outside of the US, don't have enough phone jacks. The long-term winner is likely to be a wireless solution such as 802.11. The 802.11b standard provides enough bandwidth for several audio streams but only one video stream. The forthcoming 802.11a version is needed for multiple simultaneous video streams.

In the short-term, these two media, as well as power-line networking (HPPA), will slug it out, creating confusion in the market. The solution is to support two or all three standards in the residential gateway, creating a mixed-mode LAN compatible with clients of different types. This method will increase the cost of the gateway but ease consumer concerns.

As these problems are resolved and costs begin to drop, home networks will flourish. Some studies project that more than one-third of US homes will have a broadband Internet connection by 2004. These homes will be prime candidates for home networks. As music, photos and video all become compressed digital files, the advantages of a home network will become clear. The growth of home networks will create opportunities for new Linux-based devices, such as residential gateways and storage servers.