Asymmetric connectivity rant


With the relatively recent proliferation of various means of low-cost, always-on, high-speed Internet access intended for home users (IP-over-cable, DSL, satellite, wireless), many people have written much about the benefits that low-cost, widespread access to these technologies will bring to individual users. Much of what has been written is true, and I do believe that attempts to make network computing ubiquitous are generally Good Things. However, most articles and discussions about widespread, low-cost, high-speed access have failed to mention one of the most important philosophical issues with the implementation (but not necessarily the design) of these systems: asymmetric bandwidth allocation.

Until fairly recently, individuals or (more commonly) organizations which desired permanent Internet connectivity would obtain this connectivity by means of leased lines, usually through the local telephone company; companies and institutions with more than a handful of employees still do this today. This type of connection would typically be sold as being capable of transmitting data at a particular rate, such as "56Kbps" (DDS), "128Kbps" (2-channel ISDN), "1.544Mbps" (T1), etc. This data rate specfied the maximum rate of transfer for information in either direction (upstream or downstream), and the rate indicated the theoretical maximum, which would be slightly reduced due to protcol overhead, etc. This sort of symmetric connectivity allowed users to send data to the outside world and receive data from the outside world at equal ("symmetric") rates.

Dialup users, too, have historically used technologies which provided for symmetric bandwidth allocation. Initially, this came in the form of dialup shell accounts, wherein a user's computer essentially acted as a dumb terminal over a connection speed of 300-28800 baud. Files could be uploaded and downloaded at the same rate using zmodem or similar protocols. With the advent of the web, many users switched to dialup SLIP or PPP access, which also worked over conventional POTS lines and which also operated in a symmetric fashion. In the late 1990s, however, this changed somewhat with the development of so-called "56K" dialup modems, which operated at an "upstream" speed of approximately 28Kbps and a "downstream" speed of approximately 53Kbps, under ideal conditions which were almost never met. In most cases, users of so-called "56K" modems connected at approximately 28.8Kbps, for essentially symmetrical speed in both directions. Users of ISDN lines could connect at 64K or 128K in both directions.

More recently, however, we have seen the increasingly widespread adoption of various alternative forms of Internet access for home users, which are designed to provide constant connectivity (as opposed to the intermittent connectivity provided by dialup facilities), usually at higher speeds than would be possible with standard dialup technology, and at lower prices than have historically been charged for dedicated circuits such as 56K DDS and T1. The most common forms of these technologies use either existing cable-TV facilities and wiring or DSL technology which works over standard copper telephone wiring. Of the various forms of DSL which are currently in use, the most common technology for residential users has been ADSL.

While the widespread adoption of (relatively) cheap, high-speed, always-on Internet connectivity for home users is clearly a Good Thing, I am quite concerned about one near-universal aspect of the implementation of these so-called "broadband" access technologies. Specifically, ADSL services and nearly all IP-over-cable services limit the potential upstream data rate to a small fraction of the potential downstream data rate. For example, the standard Verizon (nee Bell Atlantic, nee Nynex, nee New England Telephone) ADSL configuration allows for a maximum of 640Kbps downstream, while only offers a measely 90Kbps upstream. Some IP-over-cable providers are slightly more generous, allowing (in the case of Mediaone - northeast region) up to 1.544Mbps downstream, and 300Kbps upstream.

"So what?" you say, "it's still faster than dialup"! Yes, this is true, but it is exemplery of the result of a flawed assumption which is held by most of the providers of these residential connectivity services, which is likely to have a great (and detrimental) impact upon the use of this medium for communications in the short-term and long-term future. Essentially, the argument for asymmetrical bandwidth allocation is that users tend to download more stuff than they upload, and therefore demand more downstream bandwidth. While this is, in many cases, a basically correct assumption, the logic is flawed, and ignores the fact that the Internet is unique as a communication medium and was designed for two-way communication. By stifling users' upstream traffic (and thus subtly discouraging the use of their service for upstream traffic), providers are essentially shifting the nature of this medium from a two-way medium of communication unlike anything else in history to a one-way medium, much like television, radio, and printing.

In short, I fear that the widespread acceptance of asymmetrical bandwidth allocation will reduce the power of the Internet as a one-to-one (or, more accurrately, many-to-many) communication medium, and instead turn it into a broadcast medium, for which the protcols on which the Internet is based are not well suited, and which simply duplicates existing broadcast media.

Although these comments probably appear to be somewhat reactionary, this sort of twisting of the Internet from its status as a highly democratic, decentralized communications medium in which anyone could, for very little cost, communicate freely with anyone else, into a more centralized, less free medium, likely to be controlled by the ever-decreasing number of corporate entities who control access to it. This is already happening in some areas, as cable-TV providers are trying to provide "content" such as streaming video over IP via their cable lines to customers. This is not only a terrible mis-application of packet-switched networks, but is also an indication of what the future may be like--with Internet access controlled by a few major companies (AOL/Time Warner, AT&T, local telephone ILECS, etc.), who attempt to reduce the ability of the average individual or small business to provide useful information to the worldwide community by exercising their control over what types of access services will and won't be available to the average home user at reasonable costs.

I am now a bit afraid that my comments here will be misunderstood--I'm not trying to write some anti-capitalist rant about Why Big Corporations Are Evil. That's not my intention here at all, and couldn't be farther from my own beliefs. I am very much pro-capitalist and anti-government-intervention. I also believe that there are many reasons why people would want to communicate with other people for reasons other than commercial gain. One of the major assets of this communication medium is that it allows any individual with some amount of creativity to make available his writing, photographs, movies, etc. to the world for a minimal cost and compete (in a fair manner) for the attention of other users with anyone else; I also believe that electronic commerce isn't an especially important purpose of a worldwide computer network; much more important is the sharing of ideas among ordinary individuals.

Unfortunately, it seems that this ability to provide information is being stifled by the access providers. While this may or may not be a specific agenda of theirs (in fact, I do not believe that it is--I suspect that it is just a result of technology which has been developed in response to flawed assumptions about what consumers want), it does appear to be a problem whose importance will only increase. After all, the last thing that most of us want is for this wonderful, unique communication medium to be turned into some big impersonal electronic shopping mall, where lots of plastic junk is available for sale, but where none of the users can engage in intelligent conversation with others.

Obviously, the so-called "content-provider" groups of commercial enterprises like the idea of asymmetrical connections, as it reduces the ability of individuals to compete with them by offering their own "content" (information, writing, pictures, whatever) of higher quality and without the commercial motivation that the "content provider" companies have. The idea of human beings who actually want to engage in some sort of dialogue with other human beings is foreign to many of these companies which are in the business of providing "content," as there is little commercial incentive to encourage this (no good way to profit from it and no way to encourage people to consume more of it), despite the fact that the ability to do such a thing is one of the major advantages of the Internet over other communication media.

I will not mention here the other problems associated with many ISPs with respect to terms of service which often prohibit users from running "servers," and/or dynamic-IP address allocation policies which make it near-impossible to effectively run any type of server off of such a connection. In short, I feel that such restrictions also tend to stifle communication; if the concern is one of bandwidth usage, then I completely support billing systems which take usage into account (maybe with a flat charge for a reasonable transfer limit in both directions, and additional charges per gigabyte of data transferred above that limit). I realize that bandwidth costs money and that users should pay for same; I fail to understand the justification for current policies which penalize individuals who wish to run low-volume servers, yet allow software and music pirates (or legitimate users of PD software and recordings) to transfer huge files without limitation. I realize that the cable infrastructure is such that static IP addressing would result in a massive waste of IP addresses, although this problem could be solved with dynamic routing. Still, there is no legitimate justification for DSL service which uses dynamic IP address allocation.

Although few seem to agree with me at this point, I do find the philosophical issues associated with asymmetrical bandwidth allocation to be very scary indeed, with respect to their long-term effects on human communication. I don't have a solution, unfortunately, other than to encourage users to look hard at connectivity options available to them and to take advantage of SDSL (which is a better technology than ADSL; it is more stable and allows symmetrical upstream and downstream transfer rates) service where available, and to boycott any DSL or cable ISP who uses dynamic IP addressing and/or has a blanket "no servers allowed" policy. I currently pay approximately three times the cost of low-end ADSL service in order to get 192K SDSL connectivity with static IP addresses, routed service, and other goodies, and I encourage others to do the same. Only by demanding reasonable terms of service can users expect the price of such services to come down over time, and the supply to increase in the future. Government regulation (particularly the misguided form of regulation which often comes when professional politicians attempt to understand large-scale data networks) is not the answer here. As consumers, we must make clear to those who seek to profit from us that provision of inexpensive symmetrical connectivity, without mindless usage retrictions on anyone who might possibly want to do something that might damage someone else's revenue stream, can be made to be profitable for the providers of such access.


Return to Home Page
Valid HTML 3.2! Last Modified: Tuesday, 27-Mar-2001 14:54:16 EST