Demo Letter: Home Networking
This short piece focuses on IP as the enabling technology.01-Aug-1998

This essay original appeared in "Demo Letter" in the Summer of 1998. It probably needs more editing but, unfortunately, it was published more or less "as-is".

Introduction

The confusion about "home networking" reflects the confusing about what the Internet is. To many the Internet is the Web and the applications that use it.

Home Network is often confused with the Entertainment Center (a house is the device used to keep the rain off the television) or with Home Automation. The concepts are related. But the major lesson of the Internet and, for that matter, the early days of the PC, is that if we can find the right enabling technologies, we can rapidly innovate and create new products and new markets.

If the Home Networking is seen just in terms of the applications, then there is little synergy and many still-born markets

IP � The Great Enabler

For this sake of this essay, I'm making two simplifying assumptions:

  1. IPV6. This is the version of the Internet Protocol with a 128 bit address and the ability to create a local IP network without waiting for a global address.
  2. End point security. Each end point device authenticates using protection keys. Focusing on the key itself allows a choice of policies for gaining and distributing the keys. Encryption should also be common but authentication is more fundamental.

While these are realistic assumptions for new devices, they are more important as an architectural model. They simply reflect the original simple architecture updated for a larger and less innocent world.

It is this simplicity that has made the Internet Protocol such an important. The Web is but one of the simple and inevitable by-product. By just putting a destination address in front of a few bits and dropping them on the wire one can indeed move the world. Sometimes the bits don't get there. The idea of making the unreliable transport the basic service and shifting the responsibility to the end points was a crucial innovation. It doesn't second-guess the applications and forced them to accept services that they don't need or are insufficient. Packet losses are only one of the issues an application must deal with anyway. By not creating false promises, the applications are forced to be more robust. One result is the Web which is resilient rather than brittle.

The Internet itself is simply the collection of IP enabled networks. Or, in a more restricted sense, it is the transport between local networks using IP. The bulk of the traffic is within these networks. A switched 100 megabit Ethernet can carry a lot of IP traffic

Vertical vs. Horizontal

The personal computing industry has thrived because we have been able to leverage common software and hardware. Back when Dan and I first wrote VisiCalc we had to justify a $3000 computer for one program. But once you had a computer, a $10 program need to only justify the $10 purchase. Price/performance improvements of a trillion over the history of the "stored program" computer is a dramatic result.

By contrast, the communications infrastructure is a complete nonstarter. It is a vast wasteland carrying a few phone calls when the telephone companies deign to and that's about it. And we accept that just like we accepted notion that computers were behemoths supported by their own priesthood.

When we want to create new services such as wiring the entertainment center or connecting the refrigerator to the toaster, we create a new infrastructure. Each one tuned for its application. IEEE-1394 sets the rules for media streams. CEBus has magic knowledge about what devices exist and what the relationships are.

One price is that neither is useful until there is sufficient infrastructure and buy-in. The other is that innovation is stymied since these are backward looking architectures tuned to supporting the old applications. Imagine the personal computer with a Teletype as the interface.

IP changes all this. IP spans the physical media and allows a common communications infrastructure to be shared. Since the hardware is shared and de-coupled from the applications, we have a common focus for innovation.

Enabling Technologies

And this is already showing results. The ability to create networks with "No New Wires" is a good example. This gets us out of the cycle of stagnation � no services without new hardware, no hardware without services. The cost of the network is essentially zero! We can start to assume the IP infrastructure! And, as the perceived value increases, we'll be get higher speed and full time connectivity to the rest of the world via the Internet.

Just as with PC's, it doesn't matter if the market is driven by applications if it is limited by technology.

The first step is in making the physical infrastructure available and delivering IP connectivity

Now

We can now focus on taking advantage of it and experimenting with new applications and devices. We don't need to have the answer to what home automation is. We don't need to wait for committees or, worse, Congress, to tell us what DTV will be.

Given IP as the common "wiring" we can start to experiment and ship the experiments since each can be useful and doesn't require limiting future choices.

One can leverage the web. A web server is a single chip or a small amount of software. And immediately the thermostat or VCR can be viewed in any browser. Or one can use low level (UDP) packets with a trivial protocol to control a light switch. (Yes, it can be cheap and simple)

Impediments

This is all very exciting. There are still some impediments. While the simple model of IP is real, there is still many who see kludges like proxies, firewalls, NATs, VPNs as features. There are still lots of myths about the cost and complexity of IP. There are na�ve assumptions about what the applications are and how ATM, 1394, Residential Gateways, ITV, DTV are the answer thus precluding other questions.

Just like the industrial strength protocols and networking of the past turned out to be illusions these too will succumb to the marketplace.

D�j� vu All Over Again

It's 1978 and there are those still not excited about the microprocessor. And there are those of us who know we don't know what will happen but know it will be lots of fun.

And, perhaps, there is even some money in it.

Bob Frankston
Copyright � 1998 Bob Frankston. All rights reserved.