Explain a database lockout and the deadlock phenomenon. Contrast that to concurrency control and the timestamping technique. Describe the importance of these items in relation to database integrity.

Question

What will be an ideal response?

Answer 1

In a centralized database, a lockout is used to ensure data currency. A lockout prevents simultaneous access to individual data elements by different information processing units (IPU). When one IPU requests access to a data element, a lock is put on the file, record, or element. No other IPU can access the file, record, or element until the lock is released.

In a partitioned database, lockouts are also used to ensure data currency. It is possible, however, for multiple sites to place locks on records that results in a deadlock condition which prevents transactions from processing. All transactions are in a wait state until the locks are removed. A deadlock cannot be resolved without outside intervention from the user's application, the DBMS, or the operating system.

In a replicated database, a large volume of data flows between sites, and temporary inconsistencies in the database may occur. Concurrency control ensures that transactions executed at each IPU are accurately reflected in the databases of all other sites. A popular method for concurrency control is to timestamp transactions. Transactions that may be in conflict are assigned a system-wide timestamp. Then, the identified transactions are processed in timestamp order.

Both database lockouts and concurrency controls are designed to ensure that the transactions are completely processed and that all transactions are accurately reflected in the firm's databases. Failure to implement these controls can result in transactions being lost, being partially processed, or with inconsistent databases.

PTS: 1