1. Introduction Design of data management layer is four-step process [1] Select format of storage; Alternatives are: Files (Sequential and Random)  Relational databases  Object-relational databases  Object-oriented databases [2] Map problem domain classes to selected storage format  Problem domain objects drive object storage design [3] Optimise storage for efficient performance [4] Design appropriate data access classes and manipulation classes to: Separate problem domain classes from storage format  Handle all communication with the database 2. Files Files contain lists of data, e.g. bytes, characters, strings, integers, records(which contain fields), etc. Usually supported via programming language-dependent operations, e.g. as defined by methods of pre-defined classes FileInputStream, FileOutputStream for byte streams in Java, FileReader and FileWriter classes character streams in Java, RandomAccessFile in Java. Files may be used as masterfiles, look-up files, transaction files, audit files, history files. Sequential access files are accessed by sequential file operations, e.g. read, write, search Such files are very efficient for sequential processing via sequential operations But are very inefficient for random processing, i.e. on average 50% of contents must be accessed to find desired random record 2.1 Sequential Files May be: Unordered  Data is in sequence that it was written to the file, i.e. data is organised serially  New data is added to end of file  Ordered  Data is in sorted order, e.g. ascending order of some component in the data 

… 下载原始文件以阅读完整文档。