Chapter15ReliabilityandSecurityinDatabaseServers.ppt

Agenda Reasons why backup and recovery are important to database servers Strategies for recovery from failures Ways in which DB systems define and enforce security control Potential Reasons of Database Failure 1. The database server computer crashes. 2. The database server program crashes. 3. A database client computer crashes. 4. A client program crashes. 5. The network connection between client and server fails. 6. A transaction executes a rollback operation. 7. A transaction executes an illegal operation. 8. Two or more transactions deadlock. 9. One or more transactions introduce errors into the database. 10. Data on a disk drive is corrupted. Recovery Strategies Recovery via Reprocessing: Recover the database state from a backup and reprocess all transactions that have occurred since the backup was created Drawback: does not guarantee durability by reprocessing, except in the most limited cases (i.e. applications that do not respond to direct user input) Recovery Strategies (cont’d) Recovery via Roll Forward: Recover database state from a backup and then reapply all changes of committed transactions in same order that they were originally committed using redo logs Advantage: Full durability using physically remote backups and redo logs Recovery Strategies (cont’d) Recovery via Rollback: Recover to a previous correct database state by removing the effect of corrupted transactions by using undo logs Advantage: More efficient than Rollforward when database has not crashed and only few bad transactions Recovery Strategies (cont’d) Recovery from Disk Corruption: Mostly recovery done by backup and roll forward If some disk pages corrupted might be possible to recover from disk cache Else, recreate corrupted pages using transaction logs Security in Relational Database Systems Database security starts with physical security DBMS uses two main types of security: Account security for User Authorization Access security for Protection of Database Objec