What problems will new ISPs face ?

There has been a lot of buzz about the coming of private ISPs of late. Lots of the Big Players from the international market are eyeing the Indian market, seemingly undeterred by the delays and the ongoing tussle between TRAI and DoT. In many cases, they have already gone through much of the paperwork necessary to apply to become an ISP in India as and when the market does open up. It would be interesting, we thought, to take a look at some of the issues they would have to face to set up shop here.

The view from the ground

  • On 20th October 1997, the Cabinet cleared a policy which stated, among other things, that "India should have 1.5 to 2 million internet users by 2000 AD". Thus, the intention is there, though it hasn't quite been translated into implementation yet.

  • Numbers like these can only be achieved by a proliferation of small, medium and large ISPs. They are unlikely to happen with only a few giant players. This is a situation likely to lead to shakedowns and price wars.

  • VSNL may not even be the only existing ISP that new entrants will have to compete with - there exists a possibility that ERNET may be "allowed" to enter the fray as well.

  • A new player will have to compete with VSNL infrastructure already in place.

  • Under the current rules, potential ISPs will have to buy bandwidth from VSNL. This could lead to a conflict-of-interest situation.

  • There is a limited window of opportunity (anywhere from 5-10 years), before other technologies, such as affordable high-speed satellite based net access become available.

    Given the nature of the net and its underlying technologies, it has most often been driven by small, entrepreneurial outfits. These also have the potential to make the industry a service-driven one, rather than a hardware driven one. (that is to say, rather than your target audience being restricted to people who own PCs already, one could look at a situation where people buy a PC just to access this service.)

    The small guys

    Keeping in mind the unique position of the small operator, and the great interest in this topic, this article will attempt to explore whether being an ISP might be for you, and suggest a brief and moderately priced experiment you can embark on to see if you're ISP material.

    I'm making some assumptions about the current state of your ISP project:

  • You have determined that the business looks viable. What this basically means is that there is either little competition in your area, the competition is weak, or you have specific strategies that you think will make your effort succeed. Alas, I can help you little with this; it's a determination you have to make yourself. I recommend checking out computer publications, BBSes, and press releases in the mainstream media to locate your competitors, and assess their relative merits.

  • You love computers. If you don't, this is not the business for you. Even hardened veteran ISPs are always at their computers, programming perl scripts for their WWW sites, debugging modem problems, or following the many ISP mailing lists that have sprung up. This is not a "set and forget" business.

  • You are not intimidated by change. This business is changing, every second of every day. The rules are in constant flux. In order to succeed, you must follow a large number of mailing lists, and you must adjust constantly to changing conditions and competitive threats. If you don't like change, or if you couldn't stand 100 email messages a day from certain mailing lists, this business isn't for you.


    You will have to learn Unix. The effort you will go to in order to use less well supported operating systems, such as Windows NT, will be even greater than the effort needed to learn Unix. If you do want to use Windows NT for some reason, be prepared for a tough road, and little sympathy from your fellow ISPs when you have trouble.

    Many people interested in this business don't know Unix. One of the strongest suggestions I can give you is to learn it before you take the plunge and spend sizable sums on equipment and connections. If you have a connection that's costing you a pretty packet, and you're still trying to figure out how to do a directory listing on your shiny new system, you're in big trouble.

    If you don't know Unix, but you would like to learn it in order to become a player in this business, here are some suggestions.

    1. I assume you already have access to the internet, probably through VSNL. A VSNL shell prompt is NOT a good place to learn Unix. It's probably the most restricted Unix access in the world, with even common commands like ls being crippled. Get yourself a free account at a place that offers you Unix shell access, like m-net.arbornet.org or nyx.net -- you can use the TELNET command to reach these from VSNL.

    2. Get Linux. It's may not the Unix you'll stick with permanently, but you can get it on CD ROM very cheaply (like around Rs. 3500 for the Red Hat distribution, or even free with some computer publications), and you can run it on hardware you probably own already. If you don't want to get Linux, FreeBSD would also work fine, but note that Linux runs on a far larger selection of hardware, so FreeBSD is likely to be harder to set up and run initially.

      Don't forget to look through the online documentation that comes with your copy of Linux -- it's a valuable resource.

    3. Play around. If you have a VSNL TCP/IP account, plug your Linux machine into the net using that. You now have a "play ISP". Download and compile the latest version of sendmail from Berkeley. Install it on your system and see if you can get a friend to send you mail. See if installing and configuring complex Unix sofware really is your cup of tea before you get your expensive connection.

    PC or Workstation?

    When you start out, get a Sun or Alpha system. Yes, PCs are cheap. The problem is that they're cheap because they're not designed or engineered as well as the higher-end machines. Most PC systems are designed only for the low-stress life of running Windows and Windows applications. A Unix system puts many more demands on a system than the usual PC software.

    You can get a high-quality PC if you know where to look, but the cost can be nearly as much as a decent workstation, particulary if you're willing to consider a Sun clone instead of the real thing.

    ISP, Web Presence Provider or Content Provider?

    Recently, there has been a trend of companies springing up solely to provide space on the World Wide Web (WWW), as opposed to the traditional ISP.

    One reason for this is that the advent of SLIP/PPP has made product differentiation more difficult than in the past. Web presence provision is especially attractive if you have artistic or creative ability to put into the venture.

    Be warned, though, that the competition is getting very stiff in this area, and it's not clear whether costs are up too high and prices too low for the service to remain solidly profitable. Check local conditions before you leap, and make sure you have something special you can bring to the table.

    You might also want to explore the possibilities of generating your own unoque content (for example, a syndicated column for which you have exclusive net-wide rights) and either parcel it our to other wannabe ISPs, or directly to the end-user.

    We'll be looking at other issues faced by this fascinating new market in future columns. Your feedback, as always is welcome. You can contact me at <udhay@arachnis.com> -- I'm always happy to hear from you.


    Udhay Shankar N <udhay@pobox.com> is a Random Networking Enthusiast who collects interesting people.