Institutional Computing and GIS Concepts, Needs and Networks for Research and Teaching By Jeff Thurston Academic and other large multi-functional institutions often have spatial information system needs that are unique, requiring creative planning for optimum use of resources. GIS can be found in many different university /college departments including agriculture, forestry, history, art, health, computing science and physics to name a few. At the same time, institutional infrastructure can be managed using GIS and other spatial tools. Architecture documents can be digitized, buildings located in 3D and emergency routes for campus interfaced to emergency services among many other possibilities. In an institutional setting there is a need to develop computer labs with similar software. This ensures all students can be taught using the same methods and that instructors can develop lessons in one format. It may not be uncommon to find two to three different types of GIS software, two to three graphics packages, various visualization software and statistical packages along with modeling software in one lab. While the number of students with computers in their homes has increased (estimated 70-80%), this does not necessarily mean that the average student has GIS and spatial software running on their home computer. Therefore campuses must provide a sufficient number of workstations. Numbers and Time Let's assume for a moment we have a campus of 15,000 students. Further assume 10%, or 1500 students, will be exposed to GIS related software and analysis in any given year. We can probably assume a semester runs 16 weeks and there will be two semesters in most cases. This will require 47 workstations with GIS for the full 32-week year. This is where things get interesting, since in this example 10% of the student population is tying up 100% of the computer lab for 100% of the time. Obviously this is not likely to occur since the other 90% of the student population will demand computer lab time - rightfully so. This is why software is only part of the equation. To resolve this other labs need to be constructed. Hardware and Security Needs Hardware needs can often extend beyond the undergraduate student lab. Specialized software running modeling processes might require considerable computing power. Data could be continually fed to quicker computers, analyzed and interpreted then routed back to the student lab. In research applications, super-computing capability for spatial data may be required. Visualization applications, which are growing significantly in the spatial sciences area, are known to gobble up huge chunks of CPU time. Since visualization often results in large data files, the ability to "move" data becomes an important and critical factor for intranet GIS. Imagine a class of 24 downloading a 2 Mb data file, each student performing a series of overlay analysis then transferring 10 files of a similar size back to the server for archiving - all within 10-15 minutes (don't forget about the other 90% of users also making demands on servers)! Since students don't actually like their grades to be handed out to everyone on the Internet, some sort of security is necessary. Often GIS data cannot be freely distributed and is subject to (in Canada at least) penalty if served up freely to the world. Therefore, there is a need for authentification on GIS servers to ensure data security. File Sizes GIS information tends to increase in file size for research projects. These projects often generate intermediate analysis that, are stored for later retrieval and investigation. It is not uncommon to move from a simple discrete point file to an interpolated continuous surface. The file sizes increase rapidly moving from point to fully visualized surface. Currently we are talking about storing pentabytes and terabytes of spatial data. And this is just one student with a relatively medium sized or partial data set. Disk storage space will become a major issue for institutions in the not too distant future, particularly as visualization increases. Other possibilities include the next generation of GIS products which will incorporate visualization more fully and be in 3D. Many of these until now have not been available for standard desktop use. Consider the following: 1 b/w 22cm X 22cm aerial photograph requires 2 Mb of space 1 hardcopy map at 600 dpi requires about 2 Gb of space GIS monitors require 1200 or more dpi Visualizations of medium size may take minutes (30-60) to render Each of these factors needs to be considered when developing a workable GIS implementation within an academic setting. Compression technologies may speed up data transfer.