Background and Rationale
The Internet and the World Wide Web have, in a few short years, become the worlds premier intellectual resource, hosting over a billion pages of information and providing unparalleled collaboration opportunities to on-line academics.
With its enormous resources and collaborative capacities, the Internet poses a historically unique chance for African universities to gain a more equal footing with their sister institutions in the more developed countries. Given that more and more academic resources are moving to the Internet and in some cases being made available only on the Internet it is imperative that African universities become connected soon if they are not to be rendered irrelevant in the modern academic world. Yet many African universities are years, perhaps decades, away from reliable and robust full Internet connectivity.
However, even a full Internet connection does not guarantee that the resources of the Internet will be available to scholars at African universities. Where users pay for their connect time or the amount of data they download from the Internet, the cost of connectivity will limit which resources they access. Some universities are likely to be able to afford only slower Internet connections, limiting their use of multimedia materials and large file downloads.
As well, for those who have not experienced the past ten years of Internet development, the dizzying array of options and materials are certain to hamper their progress. African librarians are starting from scratch as new Internet connections are created and, operating under the same time and cost constraints as their clients, may require years of training and experience to catch up with their colleagues in the West.
There are many infrastructural impediments to Internet connectivity that are unique to the African context: power failures, equipment failures, regulatory restriction of communication technologies, expensive or unreliable telecommunication technologies, and lack of foreign exchange with which to pay for connectivity. For many African universities, their first years on the Internet may involve only occasional, unreliable connections with long periods of being off-line while waiting for broken components or broken budgets to be repaired.
Finally, there's the perennial issue of content: little of the information which is currently available on the Internet is produced in Africa. One South African researcher claims that less than 0.23% of Internet content originates in Africa. Yet the results of African research is of great interest to other African scholars, who oftentimes must travel to the West to get access to African materials.
A Proposed Solution
A more reliable way to make substantial Internet-based resources available at a low cost to universities would be to provide them with a local information store that would not be affected by an undependable or inadequate Internet connection. CD-ROMs have been used in many areas, but their capacity is limited to 650MB each. Larger disk capacities are obtainable using multiple CD-ROMs or large hard drives. Yet all these options are difficult (not to mention expensive) to upgrade and to distribute in a timely manner.
The option we propose would combine three common off-the-shelf Internet technologies: proxy servers, satellite broadcasts, and push technology.
A proxy server acts as an intermediary between computers on a Local Area Network (LAN) and the larger Internet. Requests for Web pages from computers on the LAN go first to the proxy server, which makes the request to the Internet and passes the results back to the original requestor. However, as it does, it caches a copy of the newly received file onto its hard disk. When the proxy server receives a second request for the same file, it is able to deliver it instantaneously from its local copy without making another request to the Internet. (Some organizations find that 70% of Internet requests are handled locally by the proxy server saving them significant Internet traffic )
Utilizing satellites to broadcast academic documents and information into African universities involves using "edge" technologies which are only now coming to the fore in the U.S. and Europe. It represents a inexpensive way to make millions of documents available on universities' local area networks so that they can reserve their expensive Internet bandwidth for immediate communications and custom Web searches.
The Local Information Store
In our proposed system, the universities who wish to participate will provide a computer to act as an Internet proxy server on their institutions Local Area Network. The proxy server will run a modified piece of software (which we will create in conjunction with participating universities) which will serve up a wealth of information stored on the servers very large (80GB+) hard disk and, when an Internet connection is available, also handle requests to the larger Internet. (This machine could also be host to a multiple CD-ROM player, making access to established CD-ROM resources available to those on the LAN.)
The information stored on the proxy servers hard disk will include mirror copies of actual Web sites on the Internet. These sites would be chosen by a group of librarians from a consortium of African and American universities who will receive consulting from project librarians. Their decisions will be informed by data from the servers themselves, which will track the frequently visited sites and most requested documents, as well as requests from their users (a Web form will be created to expedite such requests on-line.) The proxy server will also keep track of requests for Web sites not included in the local information store.
The local information store will also contain Web pages created by member universities, search tools, news and updates from around the world, resources for teachers, students, and administrators, discussion groups, and copies of freely-distributable (public domain) software.
The proxy server will also include limited email functions and a list serve replicator. This service allows for a single incoming message to be distributed to hundreds of users on the local area network, giving all local users the opportunity to subscribe to supported list services without generating additional bandwidth.
The Master Information Store
The master copy of the information store will be kept on a master server at an institution with a consistent and fast connection to the Internet. Those charged with maintaining/updating portions of the master information store will be able to edit their Web pages and databases directly via the Internet.
Field copies of the information store will be updated daily via satellite broadcast, with one signal updating all the servers in the satellites footprint. We will develop a protocol that will allow the updates to be broadcast multiple times, staggered over several days so that systems that are not consistently operational have the best chance of receiving the updates. Updates will be serialized so that each system will track the updates it receives, ignoring duplicate update transmissions and reporting instances of incomplete or missing updates back to the master server for later rebroadcast. System administrators will have the option to disable portions of the information store they do not want to receive. In addition, system software and configuration updates will also be delivered via satellite broadcast.
Outbound requests or messages will either be sent directly via a standard Internet connection or stored for later transmission via whatever form of connectivity the university has available (satellite, telephone, short wave, cable, diskette, etc.).
Off-Line Data Warehouse
Finally, the master server will be able to host significantly more material than is actually broadcast to the field information stores. Large databases and information archives of infrequently used items, either on the wider Internet or stored on the master server, will be made available to participants via a delayed request system -- the off-line information store. The master server will broadcast indexes of the off-line information stores to the local information stores. When a participant selects a document from the off-line store index, a request to the master server for the document will be automatically generated and the participant notified of the delay. When the master server receives the request (milliseconds or months later) it will queue the request and send the requested document to the participants email account (via the Internet if available, otherwise via broadcast.).
The system will automatically monitor the requests to the off-line store and detect when a document is requested frequently enough to warrant its inclusion in the field information store.
Once the system is in place, theres no limit to the number of possible master servers (or channels.) Depending on demand and storage constraints, we may create master servers that are focused on a general area (medicine, law, engineering, etc.), giving participating institutions the choice of which server to replicate as their local information store. Of course, much of the topical information will be the same, but the core database will be configurable to suit the particular discipline.
Project personnel will work with participating universities to develop their capacity to digitize their collections. Utilizing a combination of on-line and on-site training, we will coach African librarians in the procurement of equipment, the use of the hardware and software, and techniques for high quality archiving of academic materials. We will hold biannual workshops to demonstrate digitizing techniques as well as promote professionalism and inter-university collaboration.
We will also place considerable focus on training participants to develop databases.
Each participating university will be allotted space on the master information store server for their Web site and will receive assistance from project staff in Web site design and maintenance.
The University of Iowa will assist with the task of identifying resources for the master information store and securing permission for their rebroadcast to field servers. We plan to integrate these activities into our library sciences program, inviting our African colleagues to join in to develop their skills, and seek out the experts in various fields to lead the procurement of materials.
There are numerous projects already in place which have collected documents and secured resource permissions for African scholars -- yet they have no method for delivering the content. Instead, they rely upon the African scholar having full Internet access. We would work with these groups to deliver their content via satellite.
Some content, especially that currently stored in databases, may need to be repackaged to work on our system. If the donor is unwilling or unable to reconfigure their materials, we will assist them in doing so.
This system will deliver to all nodes on the participating universitys Local Area Network:
Libraries adopting this system will receive:
System Maintenance and Reliability
This system will be maintainable and reliable because it uses off-the-shelf hardware for everything -- including the satellite receiver. The server will be a standard Pentium PC with a large hard disk. The most vulnerable part of the system will be the hard disk, which can be mirrored and/or backed up to tape for added security. In the event of a PC failure, the hard disk and the satellite connection simply needs to be moved to another functioning PC. In the event of a hard disk failure, a substitute hard disk can be installed and restored or a replacement disk can be expedited from another participating university or project headquarters (disk replicating software will be included in the system). In the event of a satellite receiver failure, a backup receiver would need to be secured by the receiving institution. (One could be kept on-site or retrieved from a national or regional depot.) System maintenance will be coordinated from project headquarters, with ample training provided to library staff to handle emergencies.
Given the uncertainties faced by many African universities, this proposal represents a stopgap measure that may turn out to be a good long-term solution as well. The eventual acquisition of full high-speed Internet connection may not necessarily preclude a university's further participating in this project, since any university will still save significant amounts on costly bandwidth by storing documents on their local area network.
Indeed, there may be greater value to participating librarians of having a focused Intranet that is a subset of the larger Internet that has grown out of the collective wisdom and experience of many African university librarians.
Given the vast footprints of today's satellites and the inexpensive components proposed, this solution might well serve hundreds, if not thousands, of libraries throughout Africa and the rest of the developing world.