Athabasca University | AU Student/Staff Login | Invited Guest Login
COMP 602 – Week 4 – Reflection
My last version of Apartment Rental ERD can be found at: https://landing.athabascau.ca/pg/file/read/123589/dbdiagrampng
The discussion thread is at: https://landing.athabascau.ca/pg/forum/topic/119519/dickson-lams-data-model-proposal/
I am noticing everyone getting through this week with a much clearer ERD than in the beginning of the week myself included. Sometimes it is the extra time spent on analysis and research do and did help clarify the goal and thus changes the model, or just simply observing the others gave new ideas and gain new insights.
With the latest version of the diagram, I had moved around a few entities including breaking out some attributes to form new tables, removing circular references and adjusting the cardinalities of a few tables. The result is a much straight forward less complex version of the previous diagram, a much refined and easier to see the relationships between the entities.
One thing I would like to mention is the experience with the use of surrogate keys. Once I started to remove surrogate key I began to force myself to find neutral key that I never thought of before. For example, Building ID, an imagined key purely added to enforce records uniqueness that has nothing to describe the underlying data. Instead I had used Block/Lot/Plan for the latest diagram; a legal description used by land surveyor and government agencies to uniquely identify a piece of land or property. Other examples include exposure, resident and management company.
Other adjustments involved breaking off the sharing of “Contact Info” between Management Company and Resident. Instead the contact info attributes are now part and parcel of the specific entity rather than a separate entity. I think this leaves more room for specific requirements used by specific entities.
Dickson
Tag cloud
Comments
The discovery of new attributes is a really interesting side-effect of following the process. Great discovery! It helps to confirm my belief that this is worthwhile even when the end result will not be a relational database. The man who taught me systems analysis was fond of observing that companies often called him in to help build a computer system but, quite often, the process led them to realise that it was not a new computer system that they needed but better ways of working in the first place. Until we really understand the human and organizational systems we are trying to computerize we are likely not just to fail but to make things worse than they were before!