Distribution of the Decision Evaluator for the Cattle Industry across the WWW
Christopher J. Kaiser
Accomplishments for this project were only incremental. The final objective of having a web based interface live on the internet and tested by classroom teachers and students was not met. As proposed the simulation code, which had branched into Unix and Windows versions was merged and running on the Linux cluster with only one or two features not completely working. A web based interface which collected input for the model was completed but not fully tested. Work on formatting output and returning it to the client was ongoing at the time of project termination. All of this code resided on gborbs.agri.missouri.edu, a cluster of eight work nodes and four terminal client stations located in the Animal Sciences Unit of the University of Missouri.
Work on the code stopped in March, 2001 after I left the University in January, 2001 and funds for the research technician working on developing the code ran out. After I left the university, the cluster remained on line despite there not being a competent system adminstrator available to maintain the machine. After September, 2001, I took the machine down due to security concerns related to lack of administration. To my knowledge, the code is still on this machine, though the cluster may have been converted to other uses by this time. In May, 2001, I received a backup tape for this computer which should contain all code developed. However, I do not have a drive capable of reading the tape at this time. A second copy of the tape was taken by Dr. William Herring, another former faculty member of the University of Missouri who is now with the University of Florida.
Other activities related to this project involved establishing a larger collaborative network of researchers. These activities were centered in two areas. Organizing a USDA regional coordinating committee and working to obtain additional grant money to expand basic research on and development of the simulation program. In the first area, I chaired an ad hoc committee of several researchers who had worked on the development of DECI. The intent of the committee was to coordinate and consolidate current efforts, define the direction of new development, and elicit further interest in model development. Approximately ten university and USDA researchers attended meetings in Fort Collins, Colorado in the fall of 1999, and Denver, Colorado in the fall of 2000. Too my knowledge this committee has not met since I left and is no longer active. The effort to expand research and obtain additional funding involved development of a large, multi-institution USDA grant proposal. These ! ! efforts were unsuccessful.
The main lesson learned from this effort is that while there is an expressed interest in this kind of effort from several entities, including universities, the USDA, and education and producer organizations, none consider it of a high enough priority to provide adequate funding and intellectual resources.
The primary reason for failing to meet software development goals was my failure to hire and adequately supervise a quality research technician. This position was only funded at a half-time level and it was initially difficult to attract a qualified person at this level. For the first year of the grant, a senior computer science student was hired on a part time basis. While he did a very good job of configuring the computer cluster for use as a server for the DECI model, he failed to recognize the importance of timely development of software code. In the second year of the grant, he was replaced with a post-doctoral mathematics student who did an excellent job in the limited time he had. In less than eight months, he developed all of the interface for obtaining inputs and most of what was needed for displaying results. The only major area of development left was completing a database for storing model output. Work in this area was ongoing when funding was exhausted.
It should be said that a secondary reason for failure of this project was a
lack of institutional support. I was unsuccessful at developing collaborative
efforts with other people at the University of Missouri due to a combination
of my inability to articulate the need for this kind of research and some institutional
apathy. Therefore, I was unable to keep this project on a high enough priority
level among my responsibilities as well as secure resources which were supposedly
available from the department.
The computer hosting the DECI program is no longer on line, due to my leaving the University of Missouri and the Animal Sciences Unit deciding to de-emphasize research in animal breeding and production systems.
No cost over runs were experienced. Development efforts were terminated when funding from ADEC was exhausted. Funding for research technician salary was inadequate due to a lack of code development by the first technician hired.
The University of Missouri provided computer hardware and facilities to host the DECI model. This cluster of eight work nodes and four terminal clients was mostly purchased with start up funds associated with my position. One work node was purchased with another faculty member's operating funds and the four client stations were purchased with my operating funds. A tape drive for backup purposes was purchased with ADEC funding. In addition, the University of Missouri provided administrative overhead for the project. Approximately, 30% of my time was spent on activities related to the project. This included time spent working with the research technicians, development of code for the underlying model, work with the USDA ad hoc committee for coordinating research and development efforts on the model. All funding for my position came from the University of Missouri. No other source of funding was involved in this project.