Wikibooks: Oracle and DB2, Comparison and Compatibility/Architecture/Oracle

• Format: Book
• Language: English
• Subjects: DML
• Online Access: https://en.wikibooks.org/wiki/Oracle_and_DB2,_Comparison_and_Compatibility/Architecture/Oracle

= Oracle = = Oracle Storage Model = The primary structures that make up the physical database on disk are • Data Files • Control Files • Undo Logs • Temporary Files Data Files hold Data Indexes Undo Segments and Temporary data. Data and Indexes can be stored in separate files or together in the same data file. Control Files contain the structure of the database. The control file is a binary file and there should be at least 3 for every database you have (for redundancy purposes). The control files look exactly alike and they should be on separate drives and controllers. The contents of the control files are such things as the database name create date and the canonical path of all the data files and checkpoint information for each database. They are used to mount the database open it and access it. They store synchronization data in the event that the database needs to be recovered. The Undo / Redo logs have 3 major components. • All the DML (Insert Select Update and Delete). All the changes made to Indexes and Tables get stored here. • All the DDL (i.e. Create Drop Alter) • All the commits The online undo logs are a record of all the changes made to the database. Because these files are so important in maintaining the integrity of the database in the event of a recovery each undo logfile is duplicated and information is written to both members (multiplexing). The log files are then organized in groups so that as one set of logfiles fills up Oracle does a checkpoint and log switch after which the next set of files in the group are written to. There will be at least two sets of logfile groups for a database but there will probably be more (up to a maximum of 255). The logfiles are written to cyclically when one group of logfiles is filled the next group in the series is written to. These files should be on separate drives and controllers (again for redundancy). When the last group in a series is written to the process begins again with the first group in the series. An additional level of security ...