热度 24
2015-1-22 19:16
2908 次阅读|
3 个评论
Every time I reflect on the ubiquitously connected environment in which all of us and our ‘things’ now exist -- and the names we use to describe it -- I am reminded of a classic short story by science fiction writer Arthur C. Clarke: " The nine billion names of God ." In it, two computer technicians are hired by the monks at a Tibetan lamasery who believe they have been assigned the task of listing all the names of God -- nine billion by their estimation -- after which the world will come to an end. For three centuries, the monks have been at this task, writing down by hand all the names they have found. They figure they have another 15,000 years to go unless they use modern technology. So the monks hire the technicians to set up a computer to automate the process. Having completed their assignment, the technicians are about to board an airplane to leave Tibet and -- you guessed it -- the lights in the sky blink off and the world ends. If task instead were for the monks to write down the many names used in the past to describe what we now commonly call the "Internet of Things" and then write down all the meanings and interpretations of that term, I doubt the world would have come to an end, even in 15,000 years. Our craziness about naming our connected computing environment may have begun in 1998, when the Internet Engineering Task Force (IETF) formalized IPv6 as the successor protocol to IPv4, which used 32-bit addresses and provided about 4.3 billion unique addresses to the TCP/IP protocols. It was determined that given population trends and the lowering costs of both personal computers and servers to within the price range of most households and small businesses, it was clear that IPv4 was going to run out of addresses soon. IPv6, which uses a 128-bit address, allows about 3.4×10 38 addresses, or more than 7.9×10 28 times as many as IPv4, making it theoretically possible to assign not only every person in the world their own URL, but all their pets and personal things as well -- just about any ol’ “thing” that could be connected. The Internet of Things became popular with IPv6, perhaps due to the shock of that realization that we have driven ourselves crazy coming up with names for the new environment and the entities in it. Internet 'appliances' Some of the names have included smart devices , netcentric computing devices , network computers , and ubiquitous or pervasive computing devices . Then there were a number of names that built on the definition of “appliance”, which is simply a machine designed to perform a specific function (i.e., kitchen appliances such as ovens and refrigerators). For example, information appliances came into common use to describe such and connected devices, such as smartphones, which had moved far beyond their original function as a dedicated voice communications device to include a range of functions traditionally done on the home PC: writing, email, and viewing photos and videos. But once marketers got involved “information appliance” began to be used for almost any embedded device. Another popular name, Internet appliances , met the same fate. That term originally described a dedicated device that made it possible for a non-techie to access specific internet services. But amongst the consumers and even the manufacturers, this term began to be confused with information appliances and was then conflated to describe smartphones. Looking for some way out of this craziness, I looked for inspiration to physics, where phenomenon, devices, and systems are named after people who had a role in discovering a physical effect: Volts, Amperes, Ohms, Hall effect, Josephson junctions, and so on. Another possibility I thought about suggesting was for the members of some IETF or other official working group to sit down and read Finnegan’s Wake by James Joyce and find a made-up name total devoid of previous intellectual history. That is where Nobel Prize winner Murray Gell-Mann found the word Quark , which he used as the name for a new fundamental particle that one of his papers on subatomic physics predicted. One name I liked then and still do is Tier-0 devices . This term has a long history in computing, dating all the way back to mainframes in the '60s and '70s. The term derives from early mainframe/client/server designs , where Tier-3 machines were centralized mainframes and minis, Tier-2 were servers, and Tier-1 were desktop systems, originally smart and dumb terminals and later desktop computers, laptops and smartphones. The name Tier-0 was originally used to describe any computing device smaller than a desktop or smartphone with the main task of embedded processing of real-time events related to controlling devices. Then marketing got involved, and makers of PDAs, set-top boxes, video game consoles and cell phones used it to describe their offerings. Beyond foolishness But now, in the new world of the Internet of Things, we are beyond all that foolishness, aren't we? Fat chance. Now, rather than a confusion of names what we have a multiple often contradictory definitions of what the Internet of Things is. For the general public, and unfortunately for many journalists as well, even those covering electronics and software development, the broadest possible and most meaningless interpretation is used: if it is a thing and it is connected in any way to a network, Internet-enabled or not, it is an Internet of Things device. The problem is the word thing. It is just a word we all use to describe any inanimate object, the ultimate in what screenwriters would call a high-concept description of an idea in a few succinct words, often to the point of obscuring any really meaningful information. Originally, the term Internet of Things had specific meaning. I first came across it in the late '90s in reference to radio frequency ID tags, where the originators of that technology drew inspiration from IPv4 and its use of URL identifiers that were assigned to each server on the Internet. Their idea was to make it easier for companies to keep track of "things" -- packages, components, subsystems -- by attaching RFID tags to them, each with a unique number. Then in 2004, the term was broadened to include the 6LoWPAN wireless extension to IPv6 to describe the linkage of specific kinds of "things" to the Internet with unique URL identifiers. Usually these were embedded things: microcontrollers, sensors and systems in which communications were machine-to-machine in nature -- no humans need apply. Now, in technical circles, the Internet of Things has come to imply a connected world in which all electronic things will be connected via IPv6’s TCP/IP protocol stack, with each identified by its own unique URL. But not everything will be so connected, for a variety of reasons. Even now, the IoT is beginning to fragment into many multiple subdomains: an industrial IoT, a building automation IoT, a defense and aeronautics IoT, and a consumer IoT -- each of them have about as much in common with each other as various sects of Christianity have to one another, which is often very little. One reason for this segmentation -- and separation -- has to do with the TCP/IP protocol itself. Developed originally in a time when the concern was connecting computing systems operated by humans, its underlying assumptions were based on human reaction times and expectations. Because it was important to the humans using it, all TCP/IP does is guarantee delivery. It is not overly concerned with when delivery occurs. The segments of electronics that still fit comfortably within that definition are personal computers, smartphones, and such new wearable consumer IoT designs such as smart watches and health and fitness monitors, which assume the close interaction of things with humans. But TCP/IP is basically asynchronous and is neither real-time nor deterministic, both of which are key requirements for many embedded things, particularly in the industrial segment. For others, such as in military/aerospace applications, the prime concern is security. There the idea of operating in an environment in which everything is connected without some exclusions is unacceptable. In automotive the problem is safety and how to operate in a "mixed criticality" environment in which systems with a high degree of safety requirements can work with systems -- most of them from consumer electronics -- which have no such requirements. I suspect that such segmentation will continue, but all within under the same broad IoT umbrella, as meaningless and devoid of useful information as it is. It would be nice, though, if the electronics industry paid as much attention to standards for naming things as we do to the specifics of various standards we use. I think Skip Ashton, vice president of software at Silicon Labs, had it right when in a recent interview with Junko Yoshida he said it was important to be less focused on the Internet and more on the “things” it contains. Ashton said that knowing intimately what "things" are supposed to do and how they think and behave will be the key to solving one of the IoT's most pressing issues: application layers. “Over the past 18 months, the industry has launched numerous consortia,” he said. “Every entity says it's targeting the 'interoperability' of things at home, but each is obviously concentrating primarily on its own interests.” But how do we eliminate that scattershot effort? One way is to come up with a common set of names and definitions. An IETF or IEEE standards group or procedure for names? But unless it was government mandated we would still be in a mess. How about doing what some industry groups and companies do, create a logo for the use of a term, such as “IoT Inside,” that a company could use on their products only if they met certain requirements? But all of this assumes that there is some sort of common desire to come up with a nomenclature that will be accepted and that actually means something. I don't think that exists. There is just too much invested right now on the marketing hype side in keeping things about the "Internet of Things" as vague and undefined as is possible for as long as possible, maybe even for the next 15,000 years. We still have time to sort things out, though. Maybe not 15,000 years, but at least two or three hundred years given the current deployment rate of IPv6. IPv4 is still carrying more than 96% of Internet traffic worldwide as of May, 2014. And more than 15 years after it was formalized, only 4 percent of the users are using IPv6 to access Google services. By my calculation, if IPv6 continues to grow at that rate it will be sometime in the twenty-fifth century before all of the miracles of the Internet of Things are fully realized, just in time for Buck Rogers to use.
标签: ipv6
