Methodical construction of sustainable SDI nodes

In this position paper, we present our group’s perspective and philosophy on the development of information systems that serve to enable spatial data sharing and spatial data processing in large, heterogeneous environments. We call such systems SDI nodes, as they are the building blocks of information infrastructure in which geo-information plays a role. We clarify our view on SDI, and identify the main reasons why the actual construction of SDI nodes is lagging behind. In the discussion, we specifically focus on the case of lesser developed countries, where the conditions to arrive at actual implementations are particularly challenging. Remedies to the causes identified are suggested, and we elaborate on how our group intends to contribute to those solutions. The problems addressed in this paper The software industry has over 40 years of experience in building information systems leading to a vast body of knowledge on how to plan, analyze, design, implement and maintain such systems. By any standard, this is only a young field of science, as is aptly demonstrated by the rapid sequence of paradigms-of-the-day (for design as well as system operation) over that short history. Yet, it is fair to acknowledge that information systems of all trades are now being built routinely due to the maturity of methods, developers and technology. ∗Our email addresses are, resp., [deby,jmorales,lemmens]@itc.nl 1 Geo-ICT is ICT with a flavour Information systems that provide geo-information capabilities are just a special case of general information system, and most of the consolidated knowledge on design methods for these systems applies to them. For instance, a commonly accepted design principle is that of ‘separation of concerns’ (SoC), under which a complex design process is broken up into phases that show as little as possible mutual overlap. This helps to manage the complexity of the design, and serves to produce intermediate designs that are themselves important products for the final system’s lifetime. In building user interfaces, the Model-View-Controller paradigm1 is an example of SoC; in database design, the distinction between user view, conceptual, logical, and physical schemas2 are examples of SoC; in service-oriented architectures, the separation between interface, protocol, service and function serves the same purpose. But practitioners in the geo-information field are not always following suit, either because they were not educated in the approach, or the technology that they apply does not provide strong support for it. As a consequence, the management of their design & implementation projects is unnecessarily complicated, and often suffers. It is also true that the (im)possibilities of the technology are not always well-understood, either by the professionals or by their managerial decision-makers. One such case arises commonly when organizations are requested to deliver information services on the basis of the geodata resources that they are holding, and believe that provision of the requested service is equivalent to handing out their data resource. In many cases, the organization can deliver the service while keeping the data to itself. The application of solid design principles and associated methods can help to clarify such situations. The status quo of SDI development in Africa The term SDI has been associated in the literature with matters of technology, policy and institutional agreement, mostly under the adagio that its success must rest on all three pillars. We do not disagree with that principle, yet feel that many of the problems associated with making SDI a reality, especially in LDCs, can often be attributed to issues within just one pillar. This paper discusses why SDI technology is difficult to put into action, and how we can improve on it. On that domain, it constitutes our research group’s creed. Guided by efforts of UNECA and EIS-Africa, amongst others, we are witnessing a growing awareness of the need for SDI initiatives, and improvements on institutional and regional coordination. The picture, so to say, is becoming clearer, but also more complex. At the same time, actual implementations of contributor SDI nodes remain relatively rare. Various reasons for this have been identified: here, we 1Trygve Reenskaug is generally recognized as the inventor of this paradigm, when he was working as a visitor at the Xerox Palo Alto Research Laboratory in the late 1970s on the Smalltalk programming language. 2This distinction is normally traced back to the ANSI/SPARC Study Group on Database Management Systems [10]. 2 want to focus on the issues of methodical and technological nature and discuss possible remedies. What we specifically aim at in our work is to build up a body of design and implementation knowledge that can be applied to systematically build SDI nodes. That body of knowledge should include: • a shared system architecture perspective, with high levels of system modularity, • standardized methods of design, • standardized methods of implementation, and • in-depth knowledge of the technology. Such knowledge must be built up, and subsequently shared amongst those organizations and individuals working with it. Our organization aims to play the role of accommodator and collaborator in this scheme. We want to be actively involved in projects of actual SDI node implementation, and develop generic solutions that others can work from. In essence, in SDI, we should not only attempt to make our data interoperable, but also our design methods as well as the technology that we apply for it. The objectives of our work A lack of resources is probably the common denominator for obstacles in SDI development. The most crucial resources are: human capacity, technical and methodical know-how, data and metadata, finances. In a typical LDC governmental organization, these resources are difficult to sustain over time, as most of them age quickly or easily leak away. The approach should thus be to minimize leaks and ageing, and minimize the amount of resources needed. We address some of the issues in some more detail. Human capacity There exists a genuine lack of IT capacity to develop and maintain larger (geo)information systems, as most geo-ICT capacity has developed in one-off (desktop) GIS projects, which technologically are very unlike durable (multi-user) information systems. The IT capacity required for the latter witnesses a strong pull in the private sector labour market in developing economies, and so organizations with an ambition to sustain their own SDI node(s) have a hard time sustaining their human capacity. Standardization of the technology and of the uses of the technology will help to improve the situation because routine solutions will have the overall effect of lowering the required know-how levels. Isolation The development of SDI is often too strictly interpreted as a geospatial domain instead of a specialized domain within the wider ICT domain. As a consequence, the methods and techniques that have been developed in that wider domain are not always applied for SDI development.