Technologies that Didn’t
Part 8 - Dial Banks
In the early 1990’s, several large network equipment vendors sold equipment targeted at large scale dial banks, including Cisco, ADTRAN, and Cabletron. These servers were often tied to banks of modems on one side, and to servers of some sort on the other, creating a dial bank. Dial banks grew in scale and sophistication on the back of Bulletin Board Systems (BBSs), run by dedicated system operators (SYSOPs) for the support of individual communities. At one time, there were tens of thousands, potentially even hundreds of thousands, of these systems in the United States alone, serving large communities of users.
While BBSs are often called the “internet before the internet,” they were really a different kind of world than the modern day ‘net. If the ‘net is a large continent connected by a well-maintained (well, perhaps moderately well-maintained in some cases) transportation system, BBSs are a group of islands scattered through an ocean. No two BBS connects to any other, they didn’t generally support email between one another, and while users did overlap, they each served a separate community.
The dial bank and BBS business model grew into early content provider giants, such as America Online and Prodigy. For a long time, you could find free disks for these services just about everywhere. In fact, when working with a customer on a case in a vendor’s technical assistance center, I sent a manager off to set up an account with one of these services to download the largest software package I could find. This kept the manager tied up long enough to work with the engineer, eventually restoring the company’s access to the Internet.
Dial banks were eventually supported by network operating systems, marking an early form of the virtual private network. Specialized servers were developed, deployed, and managed just for employee dial-in so sales folks could work while on the road, or download product information while at a customer site. Extensions to protocols were developed (such as RFC1793, which describes how to run OSPF over a demand circuit, which was aimed primarily at dial-up links designed to connect multiple networks) such as a remote office connecting to a main office to perform inventory, or process credit card transactions.
In today’s IT world, however, MODEMs have (largely) gone the way of other extinct forms of hardware. While MODEMs once ran data packets over voice channels, today voice runs over data channels, even on mobile (or cell) phones. This radical change in fortunes has one obvious culprit: the widespread adoption of the Internet, combined with modestly prices always-on specialized data connectivity. It should have been easy to see this change taking place; voice can easily be deployed as a specialized form of data service, but data is much more difficult to make into a specialized form of data service.
Lessons learned? First, what seems inevitable, today is almost never inevitable. Some technology that seems like it is necessary to the very foundation of “doing business” can disappear fairly quickly, almost without warning, if you are not paying close attention to the fundamental technologies in play. Second, if you do not want to be caught off guard, it is always best to understand the problems being solved and the financial efficiencies being gained by solving them.
Third, that which runs over the top today could well be that which runs underneath tomorrow. Technology layering makes it possible to develop and understand a technology long before large-scale deployment. It could well be that IP itself becomes an “over-the-top” technology running on some new form of “telephone network,” reversing the fortunes of the two technologies, in the near future. Many companies are already considering the question “what comes after IP.”