COOK Report on Internet
It is a privilege to address you today. However the idea that I as a foreigner could make anything like a full range of meaningful statements about design criteria for the Russian Internet struck me as somewhat presumptuous. Consequently I asked Andrei Sebrant, Moscow based physicist, good friend, and network addict to assist me. The first four areas that follow are his input.
Andrei wrote to me: "In what follows I will try to sum up strategic problems of Internet development in Russia as seen by a user, omitting standard complaints about poor services and bad line quality. I mention only those that are *not* mentioned in the conclusion of your conference paper for the NATO meeting. Network interconnection and predictable government policies are *very* important issues, but you have already written just the right words about them - I do not want to repeat them here.
The Internet in Russia is in a unique situation. It is being built from scratch, not evolving from pre-existing infrastructure as in the US. Consequently I think we avoid (as much as possible) the painful stage of developing a network interface that can be navigated by only UNIX wizards.
Many authors of papers about the Internet link its recent growth to the appearance of tools like Gopher or WWW which make it possible for a non technical user to use the Internet productively. Therefore if we assume that the benefits to our society are directly proportional to the number of internet users, it makes sense to build Russian Internet from the very beginning making it as user-friendly as possible.
However, these services (WWW, Gopher, ftp) are real bandwidth-eaters. That means network designers should estimate traffic and the necessary bandwidth (of both domestic lines and the gateways abroad) assuming users will use not only e-mail, but will _routinely_ transfer large files, use interactive graphic interfaces, etc. A mistake at this stage will result in a senseless traffic jams much sooner than one would expect.
Another facet of user-friendliness is the language. The Russian Internet must be bi-lingual. Not surprising, all Russian WWW servers already are. Here again from the very beginning the question of supporting several popular methods to encode Cyrillic must be solved.
Russia is a country greatly hindered by the poor quality and insufficient amount of its telephone infrastructure. The situation with the lines considerably varies from one region to another. This means that "the last mile" problem has to be addressed very carefully for every region.
Not every institute, for example, has a LAN that can be comfortably connected to the local node by a leased line. Small research institutes, schools or universities in many cases will not be able to afford a leased line at all. Thus for every regional project the cost and availability of telephone lines must be carefully analyzed.
In general, I would expect that more dial-in lines than in the US would be required. For many, a modem connected to already existing regular line will be the most reasonable and often the only available solution. But that will slightly change traditional architecture of the local hosts. We must be very careful to understand the nature of our user communities and to size network infrastructure to fit their needs.
Consistent policy for private users must be developed. Only a handful of people now have private accounts on Russian networks while the number of computers at home grows rapidly. Will they use the same infrastructure as the corporate users or should a completely separate network be created for them? The last solution seems stupid.
As soon as a person has a computer at home, he or she gradually begins to make home a "home office". I do not know the statistics of this sector development in the US, but it also should be considered before making political decisions about Internet connectivity for private users in Russia. [There are at least 5 million American computer equipped home offices. G Cook]
User support is a must. There are no network traditions in Russia nor any books or courses available. To be certain that Internet is available to more than only a few programmers, we need people ready at every site to answer silly questions. System administrators seem to be always too busy and/or arrogant to do that.
We need to train the people who will help and teach users all over the country - quite a challenge, as we can see at Glasnet, where we have to increase the staff of User Support much faster than that of the Technical Department. But this level of user support, alas, is not typical for Russian networks.
However, without a good user support we will never bring Internet to the desk of every scholar and administrator. Many will simply be too afraid to continue after facing the inevitable problems at the start and reading couple of polite messages from the Administrator where RTFM is the most polite recommendation. All this means that the training of necessary staff and expenses for maintaining reasonable level of user support must be included in any Internet development program.
Let me pick two points from my policy over view at the end on my formal conference paper on which to amplify.
First remember the technology with which we are dealing - a network in both human and technical terms. People talking to other people and using computer infrastructure to help them do so in powerful and cost effective ways. The network has a built in immune system that acts to destroy secrecy and thwart the plans of anyone who would develop policies of network operation and expansion in secret.
Certainly initial policy is formulated in private. However those who are the most successful in the network environment are those who act as quickly as possible to inform the network community of their plans *and* who listen very seriously to the community's reaction while adjusting their plans to fit the community's needs.
Finally and just as important, consider the strangeness of the economic value of connectivity. Normally to be valuable a substance like gold must be scarce. The value of connectivity between networks is precisely the opposite. The *MORE* connectivity everyone has between separate networks the more valuable to everyone that connectivity is. The Russian internet must become the *sum* of the individual pieces. No one provider can capture or control it all and to the extent that any tries the worth of all the parts of the network for all Russians - providers and users alike is diminished!
In establishing such connectivity there could be no better place to start than be resolving by year's end the problems standing in the way of a unified Moscow backbone. You have made awesome progress since my visits in 1992. While I hate to see the psychology of control getting in the way of the possibility for even more rapid progress, I am convinced that the future here is looking brighter. Thank you for inviting me.