| In this work we first compare two approaches to relational database design and discuss the advantages and drawbacks of each approach, with the purpose of finding an approach to designing a relational database whose contents are from different sources and subject to constant changes. We conclude that the database designed based on the second approach (storing structures as values) is more flexible to changes. We further extend the discussion to a more general issue: the storing structures as values approach is not one unique to database design; that is, it has applications in many other areas, such as object-oriented design and computer programming.; We design and conduct several experiments under the Oracle7 Server, running in the Solaris UNIX environment to investigate and compare the execution times of certain aggregate queries on the databases designed based on these two approaches, as well the execution time of converting the database designed based on the second approach to the database designed based on the first one. We discover that the difference of the query response times on these two databases is not significantly large compared with the execution time of converting the database. Based on the previous discussion and experimental results, we use the second approach to design and develop a weather data processing system as an example application, because the weather data are usually from different stations and updated constantly. |
