Beyond Basic Connectivity in Africa: Africa Online's continental approach

Karanja Gakio
Co-founder and VP for Technical Operations,
Africa Online, Inc.


Africa Online is an Internet Service Provider whose objective is to provide access and value added services throughout Africa. Today, we can confidently claim to be the largest ISP in Africa outside South Africa.

Over the past two years, we have deployed successively more complex and geographically dispersed services all over the continent. Starting with vanilla electronic mail services to Nairobi in 1994, we now provide a matrix of services ranging from PPP dialup access, to store and forward services using the uucp protocols, leased line services using whatever capability the local PTTs can deliver, and comprehensive Worldwide Web development services. Today, we operate pops in Nairobi and Mombasa in Kenya, Abidjan in Cote d'Ivoire, and Accra in Ghana. In developing these facilities, we have encountered and overcome a number of challenges, some of which are unique to each country of operation, and others which are general challenges faced by any ISP operating in the African context. Our agenda remains aggressive and continental in scope, and we intend to continue our efforts to deliver to those potential customers in Africa the best in Internet technology and services in order to enable them to interact and compete effectively in the global marketplace. This paper describes the status of our activities largely from a technical viewpoint, and offers some insights into how we handle uniquely African challenges.


Africa Online is an offshoot of previous pioneering activities in electronic mail delivery and community building, and African news delivery via the Internet by our founders. In 1994, the company was incorporated in the USA and in Kenya, with the goal of providing access to Internet mail to users without computer expertise. These needs included the provision of reliable electronic mail, and tools, and other associated services at reasonable prices.

During this pre-Web period, the interactive Internet experience was not a priority. In any case, many sub-Saharan PTTs did not provide "high speed" private leased lines. On the e-mail front, one transport protocol (Fido) was becoming dominant for delivery services in many areas of the continent.

The popularity of Fido was due to its meager requirements for hardware and operating systems. One only needed a DOS PC of about 286 microprocessor capability (and an international telephone budget) in order to plug into the global Fidonet network, and ultimately via fido-to-Internet gateways (via uucp) onto the global Internet. This arrangement worked well but was fraught with frustration for those who were not computer experts. Amongst the system's problems were address schemes which required a lot of work in order to ensure compatibility with Internet addresses, the inability to run reliably on platforms other than DOS, and user interfaces which could never aspire to impress non-technical users.

We recognized at that time that in order for e-mail and associated technologies to become more widespread in Africa, and to become a practical necessity for people other than personal computer technicians it would be essential to develop a system which eliminated all of these issues and more. We consequently picked an e-mail front end and back end solution which addressed all of these issues very elegantly. It supported all computer platforms natively on the client side, standard Internet e-mail addressing, chats, and electronic conferencing. It was based upon the uucp store and forward scheme, a solution which enabled us to completely bypass Fido (and the complexities of gatewaying between these two protocols) and at the same time enabled us to offer very reliable e-mail services based upon UNIX servers and the associated flexibility and reliability. Upon introducing this service in Nairobi in early 1995, we realized how correct our conviction had been--our user base began to grow almost immediately.

This approach solidified our approach to Internet networking in Africa-to provide worldwide quality services using standard technologies while making allowances for specific local needs.

We have used this approach ever since in all our African operations with continuing success. Along the way, we have learned useful lessons and - because of our continental reach - we have also encountered some issues that local ISPs do not necessarily have to deal with. This paper describes some of these challenges, with the hope that we can contribute to a process which will make it easier to continue to deploy the Internet continent wide in order to take African Internet connectivity beyond just the basic.

Present status

Africa Online is staffed primarily by high caliber local staff and a few expatriate staff both at our headquarters in Boston, and in our local offices in Africa. Today we have about 180 employees worldwide. We are a subsidiary of Prodigy, Inc., which also owns the US online service Prodigy Services Corporation.

We emphasize the use of Internet open standard technologies throughout our organization wherever possible. For servers, we apply commercial platforms such as Sun Microsystems Solaris OS on both Intel and Sparc architectures, along with Microsoft Windows NT on the Intel platform. RADIUS is standard for authentication and authorization. Netscape and Microsoft web, mail and associated servers are also standard in our stable. On the client side, we emphasize support for as many platforms as possible, particularly Microsoft Windows 3.11 and 95, and Apple Macintosh. On the access side, US Robotics and Cisco systems underpin our systems. Our priority has been to plan for dialup access at 28.8Kbps.

During the last year, we introduced full Internet services in Kenya, Cote d'Ivoire, and Ghana. At the same time, we have carried out investigations towards introducing service in a number of other countries. In Kenya, we added full interactive Internet capability to our previous e-mail-only presence in April 1996. We now operate a 128Kbps international circuit (shortly to be upgraded to 256Kbps) which feeds two pops. The main one in Nairobi serves a national audience and the largest customer base in the country. A secondary one in the coastal city of Mombasa is connected to Nairobi via a 64Kbps circuit, enabling users to access the Internet with a local call. We envisage additional pops depending on consumer demand. We also operate over 20 independent Internet domains under the .KE top level domain. These provide bulk Internet services to universities, international organizations, companies and even other Internet service providers nationwide.

We introduced full Internet services in Cote d'Ivoire in August 1996. We now operate a 128Kbps international circuit (again to be upgraded shortly to 256Kbps) which feeds one pop based in Abidjan, serving the largest customer base in the country. As in Nairobi, we are also building a network of independent domains based on the .CI country top level domain, providing bulk services to third parties.

Our Ghanaian office introduced full IP services in Accra in October 1996 using a 64Kbps leased circuit (shortly to be upgraded to 128Kbps). Already this office serves one of the largest customer bases in Ghana in addition to a few customers using our bulk services. All of our operations also provide comprehensive Web development and hosting services, including our own national web sites in English in Accra and Nairobi, and in French in Abidjan.

Our view is that the access side of Internet operations will continue to evolve towards either a commodity service due to either competition or local PTTs entering the market. Ultimately, PTTs are best positioned to offer the lowest prices. In my opinion though, it is doubtful that they can match the customer care, and other services that go towards improving the customers' experience, which can be provided by a concerted organization such as Africa Online. As a result, we plan to continue to build out our enhanced services and to continue to add additional services, in order to provide a reason for customers to join us and stay with us.

An example in this area is our Web services. Over the past year, we have focussed singularly upon our access business. This year we are also exerting additional effort into our content offerings in order to include areas such as financial transactions and other complex web applications.

Our experience this past year has demonstrated commercial interest in enhanced Web services, identically to the experience of other presence providers worldwide.


Domain name registration

As a service provider operating in multiple countries, we have had to deal with a number of country level TLD custodians. The experience has been less than satisfying. Although the Internet system of distributing authority to local custodians has been implemented, custody for many African name spaces has been split between external technical authorities and local administrative contact persons. Due to varying levels of experience and expectations between these two parties, some of the countries for which we seek local domains have been singularly unable to satisfy our requests for months on end. This is one of the most serious Internet related issue we have had to deal with in the last year.

Our experience operating the TLD for Uganda has taught us that the challenges we face in other domains are largely self inflicted on the part of the domain operators. Our observation is that many of these responsibilities were distributed without adequate safeguards to ensure that local custodians do not use them to stymie the development of local subdomains. Given the low penetration rate of full Internet services in some of these countries (and consequently the low demand for customized domain services), we do not expect an improvement in this situation until greater understanding of the responsibilities are well understood continent-wide. On the other hand, since we expect to continue playing an important role in developing Internet services in many countries within the continent, we expect to help push for resolution of this issue.

Shortage of dialup capacity

On the dialup access side, some PTTs continue to face shortages in telephone capacity. In some locations, this is extremely serious and sometimes compels us to resort to the use of wireless technologies (where available) in order to boost our dialup access bandwidth. This approach, while practical, also has the limitation of requiring customers to dial in at lower speeds (in the range of 9.6Kbps) despite our ability to handle higher speed connections. At the same time, this experience enables us to gain valuable experience in the operation of Internet access using wireless technology. In other cases, it is forcing us to consider adding dialup pops in additional locations within the city of operation in order to take advantage of exchanges where additional telephone capacity is available.

Even in areas where adequate dialup capacity is available, it is usually not possible to aggregate the dialup capacity using digital circuits such as E1 due to shortcomings in the central exchange equipment. Consequently, we rely largely on numerous individual analog circuits to deliver the access ports. This also limits our ability to take advantage of modern digital equipment in our pops.

Insufficient private leased line capacity

As is the case elsewhere, we have found that bandwidth usage in our network is growing at increasing rates. Consequently, we find ourselves going back to the PTT often for increases in bandwidth on the international circuits.

Many PTTs are finding that our needs are running far ahead of their planned capacity and are experiencing difficulties providing that bandwidth.

Fortunately, many of them are quite eager so far to work with us and with our North American providers to develop interim solutions. Sometimes this involves tying separate data channels at our pops. We expect that in some countries, this will ultimately restrict the absolute growth of the Internet, but we continue to seek innovative ways of working together with the PTTs to overcome the problem.

Within specific countries, PTTs are still struggling with preparing for the provision of domestic digital leased line services, which is of great importance in building an internal internet. National coverage is still spotty in most countries, and even within the same city, PTTs are still in the process of deploying services. As a result, we are often the first or second customers for such services wherever we have a presence.

Paucity of support for new digital services at national PTTs

Many sub-Saharan African countries only started providing local and international digital private leased circuits last year. As expected, their level of experience with this technology continues to be low. These PTTs still have not built out the organizations required to operate this technology. Consequently, we have had to work very closely with them in order to assist in coordinating their adherence to setup protocols, and to help them attend to the importance of covering issues such as around the clock monitoring, trouble logging, response, escalation and resolution.

Fortunately, all the PTTs with which we have worked have been very eager to cooperate, despite continuing to face resource shortages which make their tasks difficult. We expect the PTTs to continue to improve their performance in this area in the months and years to come.

Unreliable electricity supply

One of the most important ingredients for an ISP that expects to provide a good customer experience is stable electrical power. Some of our areas of operation face unreliable power supply either in terms of availability over time or cleanliness of the supply. Some territories ration power to the extent that we cannot expect power for almost 100% of business hours.

Backup power generators (with additional backup) which start and stop automatically are therefore absolute requirements in such areas.

Paucity of skilled technical personnel

Considering the newness of Internet technologies, it is natural to expect that experienced personnel will be difficult to come by. On the other hand, clearly bright, motivated personnel are available in most of our target countries. We are therefore committed to training our personnel to world standard. The most important technical skills required revolve around Internet standards and practice, UNIX, Windows NT, routers, and NAPs. Of all these, the easiest to fill with good quality personnel is Windows NT (due to the wide availability of Microsoft-certified training centers, and the familiarity and ease-of-use of the Windows interface). Nevertheless, we spend considerable resources to ensure that our personnel are up to par with any in the world in any of these areas. We arrange for training either internally or using external sources either within Africa or outside.

Paucity of technical support services for hardware and software

As a result of Africa's position as a global economic backwater, potential African Internet users suffer from being singularly ignored by most hardware and software vendors. The areas of acute need include modems, PCs, and a variety of software. Even where the equipment is available, we have found quality technical support to be in short supply. Certainly, some manufacturers, such as Microcom, Microsoft, and Compaq have relatively strong representation on the continent. Nevertheless, continent-wide, the coverage is spotty and is concentrated in the countries that show appreciable commercial promise. In some countries, this situation has forced us to take an active role in making available peripheral hardware such as modems, while being careful not to create the impression that we can provide extensive support for them. One advantage we do provide is that we can call upon our global organization to help local offices resolve technical issues quickly, if necessary by contacting the original vendors directly.

Challenges related to Worldwide Web deployment

As is the case elsewhere, we find a lot of excitement over the Worldwide Web. Numerous projects are under way to put together web sites addressing varying needs. Although general Web development is deceptively simple (HTML), we expect that the Web will only come to fruition when business-critical Web applications start getting deployed. In one respect, the challenges in Africa are similar to those elsewhere such as the availability of easy to use integrated development environments for creating Web applications, and the availability of bandwidth with which to make these applications available.

Other challenges are more local in scope, and range from the small national audiences, lack of a deep understanding of the capability or limitations of Web technology, lack of third party infrastructure (such as real-time credit card payment schemes). We continue to experiment with numerous technologies and concepts, and expect to end this year with positive results from a concerted effort to bring a deeper Web experience to both our access customers, content partners, and to the Web world at large.


Our approach is to assume that fundamentally, African needs in terms of Internet access and enhanced services are no different than the needs of other parts of the world. To be sure, unique African conditions make the tasks of providing identical capabilities to those available in other parts of the world more difficult, but not impossible. As noted above, we have developed excellent relationships with the PTTs in the areas where we operate. Our experience continues to be that regardless of the constraints under which the PTTs operate, they are just as eager to see these technologies take foot in Africa as we are. They take pride when a new dialup or leased circuit is availed and then put to use.

For Internet services to grow beyond basic access in Africa as elsewhere, both access and value added services will have to demonstrate their ability to generate revenues for the providers and value for the end-users. In this respect, the Worldwide Web is still an endeavor waiting to succeed by the revenue measure. Our experience so far has been there is a lot of interest in Web related service for both commercial and personal purposes. However, revenues continue to be elusive just as is the case worldwide. We intend to pay very close attention to this aspect of our business in the coming year.

Given these factors, we at Africa Online intend to continue deploying our services continent-wide by adding a number of countries this year and more in the following years. In doing so, we expect to play a fundamental role in the development of the Internet in Africa.

Main | Conference Program | Registration | Sponsoring INET 97 | Workshop for Developing Countries | African Networking Symposium | K-12 Workshop | Technical Tutorials | Hotels | Airlines | Tours | Malaysia | Media | Site Credits