Last edit: 05-03-17 Graham Wideman |
Personal |
SDSU Information Infrastructure Strategy -- What Happened? Develop Key Data Dimensions Article created: 98-07-17 |
Develop Key Data Dimensions
What kind of progress did we make on core dimensional ids?
Organization Id: I have already noted the 3-steps-forward, 3-steps-back progress on Organization Id. The university continues to use somewhere between 10 and 30 different organization id schemes. This is a crippling handicap, because it is with "organization" that responsibility, resources and accountability come together to be managed, and around which data aggregation is most importantly provided for the service of managers.
Person Id: In 95/96, in conjunction with a proposed new "One-Card" identification card, the handling of student id was considered. The choices are basically to continue to use SSN (most students already have one)... or to issue our own id numbers (but that requires a reliable authority). After a large number of meetings, the debate came down to what extent the university was bound to treat student-id as confidential information (governed by acts known as "FERPA"). It seemed to many participants that SSN needed to be treated confidentially, and that using a non-SSN-based id would free the university to share the SDSU-created id info more liberally across business units for use in a wider array of business functions. However, another group was convinced that any student info had to be treated equally confidentially, and hence a non-SSN-id was no more flexible than SSN. To resolve this issue, it was decided that management would consult with CSU legal counsel on the matter. The issue was then dropped and never resolved. Hence no advance on this topic.
Facility Id: Initially, the department responsible for naming, numbering and tracking building and rooms would periodically take it upon themselves to change the building menmonics and room numbering "as needed". All other systems that, for want of a better id scheme, relied on these mnemonics and numbers, would at some later time find out that their data was wrong and that transactions they exchanged with other systems didn't work. Luckily, we found a particularly sympathetic and savvy staff member in that department who we worked with to adopt and establish a more disciplined approach. Hopefully that will stick.
Account Numbers: The State account numbers are established by the FAS accounting system, and hence have a single authority. However, their point of connection to other entities is through the organization-id code, and the org-id code used by FAS is unique to FAS. This is further aggravated by the fact that as much as half of the university's financial picture flows from grants and contracts through the separate SDSU Foundation accounting system, which has a different set of account numbers and organization ids. (In the analysis of FAS, however, progress was made in clarifying other ids, such as those for categorizing expenses.)
Courses and Sections: These are our "product ids", and hence quite critical dimensional variables. These are captured in the SIMS system, and superficially look to be reliably authoritative. However, even these are subject to significant ad-hoc idiosyncracies, such as the same course id being used for different types of content (say "Intro to Music" with different variants for different instruments all having the same id), and the same actual section receiving multiple ids, to be used by students in different majors. Considerable improvement in the conceptual clarity of these ids would be beneficial, but awaits concerted application of I.S. discipline.
In summary, the lack of progress on organization id and person id has been a significant impediment to improving the ability to aggregate or share data across systems. While the Faclity Id example was inspirational, it's not clear it will stick if not a part of an overall culture in which benefits are being obtained (and recognized) from disciplined management of ids.