Abdulrahman A. Mohamed Mobile: +254 713 500 814 Email: abdulrehman@tum.ac.

ke

TECHNICAL UNIVERSITY OF MOMBASA

CCI 4301: ADVANCED DATABASE MANAGEMENT SYSTEMS
WEEK 1: INTRODUCTION TO ADVANCED DATABASE CONCEPTS

OUTLINE

• 1. Introduction to Transactions

• 2. Properties of Transactions

• 3. Models of Transactions

1. Introduction to Transactions

Definition: A transaction in a database is a logical unit of work that consists of one or more

database operations. These operations can include data retrieval (read), data modification

(write), or a combination of both. Transactions are essential for maintaining the integrity

and consistency of a database.

Explanation: Transactions are used to ensure that a series of database operations are

executed as a single, indivisible unit. The ACID properties, which stand for Atomicity,

Consistency, Isolation, and Durability, govern the behavior of transactions

Examples: Let's consider some examples to illustrate the concept of a transaction:

Example 1: Money Transfer Suppose you are developing a banking application. To transfer

money from one account to another, you need a transaction to ensure the following:
Abdulrahman A. Mohamed Mobile: +254 713 500 814 Email: abdulrehman@tum.ac.ke

• Deduct the specified amount from Account A.

• Add the same amount to Account B.

• Ensure that both operations are executed together, so if one fails, the other is also

rolled back.

BEGIN TRANSACTION;

UPDATE AccountA SET Balance = Balance - 100 WHERE AccountNumber = '123';

UPDATE AccountB SET Balance = Balance + 100 WHERE AccountNumber = '456';

COMMIT;

If any of the updates fails (e.g., due to insufficient funds), the entire transaction is rolled

back, maintaining data consistency.

Example 2: Online Shopping In an online shopping application, when a customer places

an order, a transaction ensures the following:

• Deduct the purchased items' quantities from the inventory.

• Record the order details in the order history.

• Ensure that both actions are completed together.

BEGIN TRANSACTION;

UPDATE Inventory SET Quantity = Quantity - 2 WHERE ProductID = '789';

INSERT INTO OrderHistory (CustomerID, ProductID, Quantity) VALUES ('12345', '789',

2);

COMMIT;

If any part of this transaction fails, the entire transaction is rolled back to maintain the

integrity of the inventory and order history.

Transactions are a fundamental concept in database management, ensuring that database

Abdulrahman A. Mohamed Mobile: +254 713 500 814 Email: abdulrehman@tum.ac.ke

operations are reliable and consistent, even in the presence of errors or system failures.

They play a crucial role in various applications, from banking systems to e-commerce

platforms, where data integrity is paramount.

2. Properties of Transactions

Definition: Transactions in a database management system (DBMS) exhibit specific

properties, often referred to as ACID properties, which ensure the reliability and integrity

of database operations. ACID stands for Atomicity, Consistency, Isolation, and Durability.

1. Atomicity:

• Definition: Atomicity ensures that a transaction is treated as a single, indivisible unit

of work. It either executes in its entirety or not at all.

• Explanation: If any part of a transaction fails (e.g., due to an error or system crash),

the entire transaction is rolled back to its initial state, ensuring data consistency.

• Example: In a banking system, transferring money from one account to another is

an atomic transaction. If the transfer fails at any point, the entire operation is rolled

back, and the money remains in the source account.

2. Consistency:

• Definition: Consistency ensures that a transaction brings the database from one

consistent state to another, preserving data integrity and defined constraints.

• Explanation: The database must satisfy integrity constraints, such as referential

integrity or domain constraints, before and after a transaction.

• Example: In an e-commerce system, if a product is out of stock, the system must

Abdulrahman A. Mohamed Mobile: +254 713 500 814 Email: abdulrehman@tum.ac.ke

ensure that no orders can be placed for that product.

3. Isolation:

• Definition: Isolation ensures that concurrent transactions do not interfere with each

other. Each transaction appears to execute in isolation, even though they may run

concurrently.

• Explanation: Isolation levels, such as Read Uncommitted, Read Committed, and

Serializable, define the degree of isolation between transactions.

• Example: In a reservation system, two users booking seats on the same flight should

not see each other's reservations until both transactions are complete.

4. Durability:

• Definition: Durability guarantees that once a transaction is committed, its effects are

permanent and survive system failures.

• Explanation: Committed changes are stored in non-volatile storage (e.g., hard disk)

and are not lost, even in the event of a power outage or system crash.

• Example: In an email system, once an email is sent and acknowledged as sent, it

should be stored safely and not lost, even if the server crashes.

5. Serializability:

• Definition: Serializability is a higher-level property that ensures that a set of

concurrent transactions is equivalent to some serial execution of those transactions.

• Explanation: It prevents anomalies like lost updates and uncommitted data from

occurring in concurrent execution.

• Example: In a banking system, serializability ensures that transferring money from

one account and depositing it into another is correctly ordered, even when multiple
Abdulrahman A. Mohamed Mobile: +254 713 500 814 Email: abdulrehman@tum.ac.ke

users perform these transactions simultaneously.

In summary, the properties of transactions, including Atomicity, Consistency, Isolation,

Durability, and Serializability (ACID), are essential for maintaining data integrity and

ensuring reliable operations in database systems. These properties help prevent data

corruption and maintain database reliability, even in the presence of system failures and

concurrent access by multiple users.

3. Models of Transactions:

1. Flat Transaction Model:

• In a flat transaction model, a transaction is a simple sequence of one or more

SQL statements.

• It lacks any control structures or branching within the transaction.

• Transactions are straightforward but may not handle complex scenarios

efficiently.

Example: A transaction that consists of a sequence of SQL statements, such as inserting a

new customer record, updating their order, and committing the changes.

2. Structured Transaction Model:

• The structured transaction model includes control structures like loops and

conditional statements within a transaction.

• It allows for more complex transaction logic and error handling.

Example: A transaction that processes a batch of orders, checking each one for validity and

updating the inventory accordingly. If an order fails validation, the transaction can roll

back only that specific order's changes.

Abdulrahman A. Mohamed Mobile: +254 713 500 814 Email: abdulrehman@tum.ac.ke

3. Nested Transaction Model:

• In the nested transaction model, a transaction can be divided into

subtransactions, each with its own ACID properties.

• Subtransactions can be committed or rolled back independently.

Example: A banking system that allows customers to perform multiple transactions within

a single session. Each transaction within the session can be committed or rolled back

separately.

4. Long-Running Transaction Model:

• Long-running transactions span extended periods and are typically used in

applications like online reservations or stock trading.

• They require mechanisms to handle partial failures and checkpointing to

ensure progress.

Example: An airline reservation system where a customer's seat selection and payment may

take place over a period of several minutes. The system must handle interruptions

gracefully.

Understanding these transaction models and their properties is crucial for designing

database systems that meet the requirements of various applications while ensuring data

integrity and consistency.