- Transactions enable you to ensure read-consistency to the point in time a transaction began for all users in the Oracle database.
- Transactions enable you to preview changes before making them permanent in Oracle.
- Transactions enable you to group logically related SQL statements into one logical unit of work.
Transaction processing consists of a set of controls that enable a user issuing an insert, update, or delete statement to declare a beginning to the series of data-change statements he or she will issue. When the user has finished making the changes to the database, the user can save the data to the database by explicitly ending the transaction. Alternatively, if a mistake is made at any point during the transaction, the user can have the database discard the changes made to the database in favor of the way the data existed before the transaction.
The commands that define transactions are as follows:
- Set transaction Initiates the beginning of a transaction and sets key features. This command is optional. A transaction will be started automatically when you start SQL*Plus, commit the previous transaction, or roll back the previous transaction.
- Commit Ends the current transaction by saving database changes and starts a new transaction.
- Rollback Ends the current transaction by discarding database changes and starts a new transaction.
- Savepoint Defines breakpoints for the transaction to enable partial rollbacks.
This command can be used to define the beginning of a transaction. If any change is made to the database after the set transaction command is issued but before the transaction is ended, all changes made will be considered part of that transaction. The set transaction statement is not required, because a transaction begins under the following circumstances:
- As soon as you log onto Oracle via SQL*Plus and execute the first command
- Immediately after issuing a rollback or commit statement to end a transaction
- When the user exits SQL*Plus
- When the system crashes
- When a data control language command such as alter database is issued
By default, a transaction will provide both read and write access unless you override this default by issuing set transaction read only. You can set the transaction isolation level with set transaction as well. The set transaction isolation level serializable command specifies serializable transaction isolation mode as defined in SQL92. If a serializable transaction contains data manipulation language (DML) that attempts to update any resource that may have been updated in a transaction uncommitted at the start of the serializable transaction, the DML statement fails. The set transaction isolation level read committed command is the default Oracle transaction behavior. If the transaction contains DML that requires row locks held by another transaction, the DML statement waits until the row locks are released
The commit statement in transaction processing represents the point in time where the user has made all the changes he or she wants to have logically grouped together, and because no mistakes have been made, the user is ready to save the work. The work keyword is an extraneous word in the commit syntax that is designed for readability.
Issuing a commit statement also implicitly begins a new transaction on the database because it closes the current transaction and starts a new one. By issuing a commit, data changes are made permanent in the database. The previous state of the data is lost. All users can view the data, and all savepoints are erased. It is important also to understand that an implicit commit occurs on the database when a user exits SQL*Plus or issues a data-definition language (DDL) command such as a create table statement, used to create a database object, or an alter table statement, used to alter a database object.
The following is an example:
SQL> COMMIT WORK;
If you have at any point issued a data-change statement you don’t want, you can discard the changes made to the database with the use of the rollback statement. The previous state of the data is restored. Locks on the affected rows are released. After the rollback command is issued, a new transaction is started implicitly by the database session. In addition to rollbacks executed when the rollback statement is issued, implicit rollback statements are conducted when a statement fails for any reason or if the user cancels a statement with the CTRL-C cancel command. The following is an example:
In some cases involving long transactions or transactions that involve many data changes, you may not want to scrap all your changes simply because the last statement issued contains unwanted changes. Savepoints are special operations that enable you to divide the work of a transaction into different segments. You can execute rollbacks to the savepoint only, leaving prior changes intact. Savepoints are great for situations where part of the transaction needs to be recovered in an uncommitted transaction. At the point the rollback to savepoint so_far_so_good statement completes in the following code block, only changes made before the savepoint was defined are kept when the commit statement is issued:
SET quantity = 55
WHERE product# = 59495;
SET quantity = 504;
ROLLBACK TO SAVEPOINT so_far_so_good;
The final aspect of the Oracle database that enables the user to employ transaction processing is the lock, the mechanism by which Oracle prevents data from being changed by more than one user at a time. Several different types of locks are available, each with its own level of scope. Locks available on a database are categorized into table-level locks and row-level locks.
A table-level lock enables only the user holding the lock to change any piece of row data in the table, during which time no other users can make changes anywhere on the table. A table lock can be held in any of several modes: row share (RS), row exclusive (RX), share (S), share row exclusive (SRX), and exclusive (X). The restrictiveness of a table lock’s mode determines the modes in which other table locks on the same table can be obtained and held.
A row-level lock gives the user the exclusive ability to change data in one or more rows of the table. However, any row in the table that is not held by the row-level lock can be changed by another user
An update statement acquires a special row-level lock called a row-exclusive lock, which means that for the period of time the update statement is executing, no other user in the database can view or change the data in the row. The same goes for delete or insert operations. Another update statement—the select for update statement—acquires a more lenient lock called the share row lock. This lock means that for the period of time the update statement is changing the data in the rows of the table, no other user may change that row, but users may look at the data in the row as it changes.