HOW
WE PUT TOGETHER OUR OWN AFRICAN ISP ON A SHOESTRING
Mark Moore and Casey Smith
The Source ISP, Plot 20, Main Street, Box 2205, Jinja, Uganda
Published by the African
Technology Forum
Introduction
Building a Local PoP
One Year Into the Project
Conclusion
References
A recent session of the African Development Forum in Addis Ababa noted that the World Wide Web’s spread across Africa is happening at an alarmingly slow rate. Much of the world, it seems, took little notice of that declaration, perhaps because Africa has more pressing problems at hand than poor download speeds. It is easy to dismiss these discussions and technologies as lacking importance or as inappropriate in light of more pressing needs such as education and primary health care. This article does not seek to patronize the masses in Africa by raising the issue of connectivity to the status of other more basic, pressing needs that can be addressed by the use of appropriate technologies. Obviously, technologies that directly address those areas deserve our focus and energies as priorities.
We are convinced, however, that there is no emerging use of technology that holds greater potential to indirectly and directly affect change in the above-mentioned areas and others than that of Internet connectivity. In large portions of rural Africa the issue of getting “connected” is not an urgently important one, but in the urban centers all over Africa it has long since taken root as a high priority need for millions of Africans. Internet cafes and public access centers are thriving in nearly every country. European and US based companies are investing in former state owned telecoms all over Africa. The Africa One project, an undersea cable that will surround the continent, holds out great promise for the future. The percentage of users is growing at rates faster than almost any other area of the world, and companies like Africa Online are being featured in international magazines such as Red Herring for their tech savvy.
In spite of the improvements and a few bright spots on the horizon, discussions concerning connectivity only serve to perpetuate the age-old image of Africa as the "Dark Continent". Most of the continent is a long way from getting practical, affordable, connectivity to the World Wide Web. What’s worse is the impression that such connectivity is hopelessly high-tech, along with the prohibitive costs associated with the necessary equipment. Both seem to inhibit sharp young African entrepreneurs from launching projects to promote connectivity in their areas.
The thrust of this article is to encourage others in parts of Africa that connecting their community might not be as difficult as they think. If a country has open information and communications technologies (ICT) laws, then communities may be able to provide such access for themselves. We put in a local point of presence (PoP) in Uganda for about US $9,000; some $40,000 less than what the World Bank and others estimated it might cost. We had no prior experience and very little expertise to rely on, yet we managed to throw together our own non-profit Internet service provider (ISP) that is currently serving hundreds of dial-in users and offering public Internet access at a reasonable cost. Here is how we did it:
BUILDING A LOCAL PoP
One of the biggest factors inhibiting many areas from getting connected is the fact that there is no local PoP. In other words, there is no local number to dial to get connected to the Internet. In Uganda, for example, where ICT policies are open and progressive, most areas still remain without affordable access because all the PoPs are in the capital city of Kampala. That means anyone outside of Kampala, roughly 80% of Uganda's population, must make an expensive long distance phone call to get their mail. Such phone calls are not affordable, for example, for schools and health centers that wish to get connected. What is worse, there is little or no economic incentive for providers to invest in the equipment and technology needed to provide local PoPs for scattered up-country users. It only makes sense that if 95% of users are in one area, the focus of efforts will be in that area. In Uganda there are large areas where phones are working and capable but long distance calls to PoPs prevent affordable access for the majority of would-be users.
It was the above scenario that got us into the ISP business. We approached a Kampala-based ISP and pleaded with them to put a local PoP in our town of Jinja. They responded by telling us how such an investment was not in their plans and not affordable for them. “After all,” said the manager, “we already have all the business in Jinja anyway. Why should we put a local PoP in there when it seems everyone is already dialing into our server here?” We left his office discouraged and convinced that our region would have to continue indefinitely dialing long distance to collect simple e-mail messages.
Over lunch that day in Kampala we sat with a friend who had a rough idea about the equipment needed to put in a dial-in service. On a napkin we made a list of about ten items, and after a little research on www.eBay.com (an established on-line auction service) discovered that we could buy everything we needed for about $4,000. We found an old Portmaster (a PoP access port) by Livingston for $600. We found a rack of sixteen 28 kilobaud (kb) modems for $200. We found a used Cisco 1600 series router, a hub and a few other odds and ends. We even discovered that we could load the Linux operating system (OS) on an old Compaq Pentium 120 megahertz (MHz) PC that we had lying around and use it as our server. About a month after that, we made a brief scheduled trip to the United States where we recanted the above story to some friends over dinner. The friends felt compelled to help, and before we left the restaurant we received $7,000 in donations to buy our equipment. Within days we had an eBay shopping spree and returned to Uganda. The Ugandan government charged us some import duty and an additional $2,000 for an ISP license. We invested most of the rest of our money on paying for technicians and dial-in phone lines (we started with five).
Initially we could only afford five phone lines at a rental and installation cost of $50 per line. These lines cost us about $10 each a month to maintain since no outgoing calls are ever made on them. We were the first in our region to utilize Uganda Telecom’s new ability to provide numbers in hunting groups. This allowed us to give just one number to our dial-in customers, while allowing them five opportunities to connect. The time frame was about four weeks to get the phone lines terminated, or connected. A few weeks of initial problems followed as we and the local telecom learned more about the technology we were attempting to use. (This was not due to a lack of knowledge on their part. They have many capable technicians but most of those familiar with data are in Kampala where UTL does a significant amount of data business.)
Perhaps the biggest cost saver for us was choosing Linux as our operating system. Linux was cheap and readily available. We found that technicians were more familiar with Unix (the more common, proprietary ancestor of Linux) OS-based programs that they were with the Microsoft Windows NT operating system, which we initially purchased and returned unopened. We are currently using SuSe 6.4, which we have found to be hassle free and relatively simple for technicians to learn. Our server has run for a solid year without stopping due to problems related to the OS.
We kept our set up cheap by being creative with the technology we had present. For example, initially we used a 486 laptop as a router. This router had Linux loaded with two network interface cards (NICs). We used the old laptop (literally a throw away in the USA) to separate our Internet Cafe and the dial-in users. In addition, in the Linux server we used a Rocket Port Serial Card, which affordably offered us up to sixteen connections without upgrading to a higher end router.
ONE YEAR INTO THE PROJECT
One year into our project things are going great. We are the only PoP outside of the Ugandan capital city and, as far as we know, possibly the only ones operating an ISP as a non-profit in all of Africa. We serve about 150 dial-in customers and are now offering leased-line options to small businesses. This has all been made possible because of Uganda's recent restructuring of the telecom industry. Uganda Telecom Limited (UTL), the former state-run Telecom, has done a lot to make our project possible by be willing to sell us affordable connectivity. Initially, they provided us a leased line at a cost of around $1,000 a month, but would not sell us bandwidth on their International Data Gateway. We were forced to terminate our UTL circuit with another company’s very small aperture terminal (VSAT) in Kampala; a solution that was very expensive, but effective. Eventually UTL began providing us with our entire solution, driving down our costs and wisely insuring that even more (be it a small amount) of Uganda’s international data traffic was traveling though their data gateway.
UTL, which is now majority-owned by a consortium called UCOM, (in which Telecel International is the major player) has been motivated to provide affordable access to us for two reasons. First, because they are smart businesspeople and are happy to facilitate any growth that generates more calls for them. Secondly, because they realize that their competitors (who are new to Uganda and 100% wireless) would also be happy to work with us and take more and more customers off of the old copper wire network that is exclusively owned by UTL. We realize that it is only a matter of time before UTL, and others who own the communication backbones in Uganda, offer the service that we are offering. We do not expect to be filling the niche we are filling long term. However, since we are not in it for the money, we are more than happy to see them progress toward a day when they can effectively and affordably serve our local community.
As we have grown, our technology has evolved with us. We recently scrapped our simple, inexpensive 8-port hub for a Cisco Catalyst Switch (also purchased on eBay). We got rid of the old Compaq PC and bought a more capable server with a tape back up system. Our Cisco 1600 series router gave way to a 2500 series. Our old APS power supply died and we installed a heavier duty power backup with deep cell batteries placed in series. Soon an AS 5200 series router complete with twenty-four digital modems will replace our ancient rack of modems and Portmaster. (This one was a donation from a company in the USA.) There is a significant amount of used equipment available on-line due to all the dot-com buiness failures of late in the USA. (Beyond just eBay, you can go to www.bid4assets.com , www.dovebid.com , www.cowanalexander.com and www.gordanbrothers.com for examples of what’s out there.) The exciting thing is that the majority of the above has been accomplished with funds provided by our dial-in subscribers.
We feel several applications have proven important to our success. Obviously basic e-mail and Internet connectivity are large draws for individuals. Our little Internet cafe, (made possible by a great organization called U-connect) thrown together with a series of donated Pentium PCs, has single handedly brought the price of surfing the Internet down by 400%. Users can now surf in our café for about 6 US cents a minute, something that we are very proud of. In addition, our Web-based e-mail (the ability to check e-mail from anywhere) has been extremely important and useful since the vast majority of our users do not have home PCs. Our Internet Message Access Protocol, revision 4 (IMAP4) web based mail program sets us apart from other larger big money ISPs who do not offer this service to their customers. Our customers can access their web based mail right on our LAN instead of waiting for slower US based web based providers like Yahoo.com and Hotmail.com to bounce their messages all over the world.
There have been some discouraging moments as well. We had a lot of initial excitement about the ability to do Voice over Internet Protocol (VoIP), but the Ugandan Communications Commission quickly put a stop to its use by threatening to close down any cafes that offer Vo IP. In spite of that technology being banned, many people still come in and utilize instant messenger services to chat with their friends and family around the world for just a fraction of what it may have cost them to do so in the past.
Our long-term vision sees amazing potential for small non-profit PoPs like us. From our PoP we can envision students sitting in Jinja participating in web-based classes at schools in the United States or Europe. It is exciting to think that we might play a small role in breaking down educational barriers in eastern Uganda. We look forward to using current Content and Cache technology to accomplish this as well as applications such as long range medical treatment. Doctors from more developed countries could see and visually diagnose patients. In short, we can begin to play a role in addressing those more pressing needs mentioned in the first paragraph of this article.
Our short-term vision sees possibilities that offer more practical and therefore more exciting uses of a non-profit local PoP. Imagine the opportunities and ideas available to schools in your area if they have affordable Internet access. Imagine individuals in your area being able to walk into a public Internet site and access e-mail for a fraction of current costs. Those changes are practical, simple and immediate.
Certainly there is a lot about our story that will not prove possible or practical in other contexts around Africa. Our hope, however, is that others will use our project and a few of the details provided above as a springboard for ideas to begin similar projects that can serve communities in a non-profit manner.
The
Source ISP: www.source.co.ug
Red
Herring: www.redherring.com
Africa Online: www.africaonline.com
eBay
www.ebay.com
www.bid4assets.com
www.dovebid.com
www.cowanalexander.com
www.gordanbrothers.com
Linux Online www.linux.org
Configuring multiple NICs with Linux: http://isolde.mcs.newpaltz.edu/~easwaran/Linux/net/net2.html
Internet Message Access Protocol, revision 4: http://www.vijaymukhi.com/vmis/imap.htm
Go back to the updated ATF site or to the main ATF home page at MIT.