DBMS Transaction Processing

Updated March 4, 2023

Introduction to DBMS Transaction Processing

DBMS Transaction Processing is a logical technique of unit processing that involves one or a collection of database access operations. In a husk, this database transaction indicates real-world events related to nay initiative. In DBMS, all kinds of database access operations that are detained between the launch and finish transaction statements are measured to be a single logical transaction.

During this transaction in the server, the database remains inconsistent. Simply once the database is dedicated/committed then the state is altered from one consistent one to another state. Hence, we can define that DBMS Transaction processing as a means of distributing information processing up to distinct, inseparable operations that are denoted as transactions that either complete or fail as a whole. This transaction should not be in a transitional or, partial state.

Syntax:

In DBMS, a transaction defines a program that comprises a group of database operations, which when implemented as a reasonable unit of data processing. The actions accomplished in a transaction contain one or multiple of database actions such as retrieve, insert, or update data.

Operations of Transaction

Succeeding are the core operations of DBMS Transaction:

• Read(X): This read operation is applied to read the X’s value from the database server and keeps it in a buffer in the main memory.
• Write(X): This write operation is applied to write the X’s value back to the database server from the buffer.

Let us take an illustration of debit transaction in DBMS from an account that involves the below operations:

R(X);
X = X – 1000;
W(X);

Assuming that the value of X before the initial transaction be 5000.

Then,

• The initial operation states the value of X from the database and supplies it in a buffer.
• The next operation will reduce X’s value by 1000. Hence, the buffer will include 4000.
• After this, the final operation will perform a write command to write buffer value to the database. So, X’s absolute value will be 4000.

But sometimes it happens that due to any hardware or software failure or power cut off, etc. the transaction can be halt before the operations in the set.

Suppose, in the above transaction, if the debit transaction is failed after the second operation run then the value of X will stay 5000 only in the database that is not acknowledged by the bank. Therefore, this problem can be solved by two operations that are significant.

• Commit: For saving the work done enduringly.
• Rollback: Helps to undo the work completed.

How to Process the Transaction in DBMS?

A transaction in DBMS is stated as a set of tasks where an individual task is the least processing unit that cannot be further allocated. So, a transaction is a minute process that is either implemented into completion fully or is not implemented at all. Here, if a transaction is retrieved consisting of data only but not having any data update then it is said to be a read-only transaction. Every high-level operation may be separated into a number of low-level tasks or processes.

For instance, a data operation for an update can be categorized into three different tasks mentioned as follows:

• Read_Item(): This command reads data elements from storage to main memory.
• Modify_Item(): This command alters the item value present in the main memory.
• Write_Item(): This command writes the changed item value from main memory to storage.

Note that the database access is limited to the operations like read_item() and write_item(). Similarly, for all other transactions, the operations read and write practices the simple database operations.

Whereas following are the low-level operations implemented in a DBMS transaction:

• Begin_Transaction: This specifies an indicator that defines the start of the execution of a transaction.
• Read_Item or Write_Item: Here, the database operations can be enclosed with the operations of the main memory as a portion of the transaction.
• End_Transaction: It is an indicator that defines the end of the transaction.
• Commit: It defines a signal to identify that the DBMS transaction has been absolutely successful in its totality and will not be incomplete.
• Rollback: It defines a signal to identify that the DBMS transaction has been unsuccessful and therefore all temporary deviations in the database are incomplete. A committed transaction will not be turned back.

Examples

DBMS transaction processing is followed as a set of operations accomplished on the data records stored in the server database that results in an alteration in the database state.

This DBMS transaction is completed through the ACID principles, which includes:

• Atomicity: Performs totally or not at all.
• Consistency: Supports uniformity of data.
• Isolation: Executes multiple transactions individualistically.
• Durability: Durability of data changed.

Following are the Transaction states in DBMS mentioned as:

• Active: This is the initial state in which the transaction arrives in the active state and stays here while executing operations like read, write, or others.
• Partially Committed: The above transaction arrives in this state after the execution of the transaction effectively.
• Committed: This state is entered after the transaction is completed fruitfully and then system checks have delivered commit signal.
• Failed: A transaction moves from active or partially committed states to this failed state when it notices that usual execution will no longer progress or system instructions fail.
• Aborted: This state is achieved when the transaction is moved back after the failure state and also the database has been re-established to its state which was previously when the transaction was initiated.

For example, we have an employee of any bank who transfers Rs.7000 from A’s account to B’s account.

This transaction processing is small but involves many low-level actions such as:

A’s Account:

Open_Account(A)
Old_Balance = A.balance
New_Balance = Old_Balance – 7000
A.Balance = New.Balance
Close_Account(A)

B’s Account:

Open_Account(B)
Old_Balance = B.balance
New_Balance = Old_Balance + 7000
B.Balance = New.Balance
Close_Account(B)

This can be a basic demonstration of processing a transaction in DBMS.

Conclusion

In DBMS, Transaction Processing has been intended to conserve database integrity i.e. the uniformity of related data objects, in a recognized consistent state. Transactional Processing databases denotes the databases which have been structured for optimizing the performance of transactional processing, also often stated as OLTP, i.e. Online Transaction Processing. The DBMS transaction should support the ACID property to confirm exactness, data reliability, and entirety.

This is a guide to DBMS Transaction Processing. Here we discuss the introduction, how to process the transaction in DBMS along with the examples respectively. You may also have a look at the following articles to learn more –

1. Relational Calculus in DBMS
2. Concurrency Control in DBMS
3. Data Independence in DBMS
4. Serializability in DBMS