Database Recovery Mechanisms

One of the most critical aspects of a Database Management System (DBMS) is the ability to recover from failures or errors that may occur within the database. Since a database stores essential data for various applications, it is crucial to have mechanisms in place to ensure data integrity and availability. Database recovery mechanisms provide the necessary tools to handle potential failures and restore the database to a stable and consistent state. In this article, we will explore the various methods used for database recovery.

Types of Failures

Before diving into the recovery mechanisms, let's understand the different types of failures that can impact a database:

  1. Transaction Failures: These failures occur when a transaction cannot proceed due to logical errors or external factors like power outages or hardware malfunctions. Transaction failures can leave the database in an inconsistent state.

  2. System Failures: System failures refer to the failure of the entire DBMS. It can be caused by hardware or software issues, network failures, or even natural disasters. These failures can result in data corruption or loss.

Recovery Mechanisms

Database recovery mechanisms can be broadly categorized into two types: Deferred Database Modification and Immediate Database Modification.

Deferred Database Modification

Deferred database modification is also known as deferred update or write-ahead logging. In this mechanism, the changes made to the database during a transaction are not immediately applied to the actual database. Instead, they are only recorded in a log file called the transaction log. When a transaction is committed, its changes are applied to the database, making it a permanent part of the database. If a failure occurs before the commit, the changes can be undone by reading the transaction log.

The main advantage of deferred database modification is its simplicity and efficiency. However, the recovery process may take longer since the changes need to be reapplied from the transaction log.

Immediate Database Modification

Immediate database modification is also known as immediate update or write-through logging. Unlike deferred modification, immediate modification applies the changes directly to the database as soon as a transaction executes. To ensure recovery in case of failures, immediate database modification uses a combination of two primary techniques: logging and checkpointing.

Logging involves recording both the old and new values of modified data. This logging process allows for undoing changes made by uncommitted transactions during recovery.

Checkpointing is the process of creating a point of synchronization between the database and the transaction log. It involves writing a record in the log that specifies the current state of the database. Checkpointing ensures that the recovery process does not need to consider transactions that occurred before the last checkpoint.

Immediate database modification provides faster recovery since the changes are already applied to the database. However, it requires more storage space due to the constant need to update the transaction log.

Recovery Algorithms

To recover from failures, database recovery mechanisms utilize specific algorithms. Here are a few commonly used algorithms:

  1. Undo/Redo Algorithm: This algorithm is used in deferred database modification. It performs undo operations to rollback uncommitted transactions and redo operations to repeat committed but unapplied transactions.

  2. ARIES (Algorithms for Recovery and Isolation Exploiting Semantics): ARIES is an advanced recovery algorithm used in immediate database modification. It offers features like a multi-level log, fuzzy checkpoints, and a redo-only recovery scheme to handle different types of failures efficiently.

  3. Shadow Paging: Shadow paging is a recovery algorithm that uses a shadow copy of the entire database. During recovery, the shadow copy is used to replace the damaged or inconsistent database. Once the recovery is complete, the shadow copy becomes the new database.


Database recovery mechanisms play a vital role in ensuring the reliability and availability of data within a DBMS. Whether using deferred or immediate database modification, the chosen recovery mechanism should align with the specific requirements of the system. Implementing appropriate recovery algorithms further enhances the efficiency and effectiveness of the recovery process. With robust recovery mechanisms in place, organizations can rest assured that their databases can handle failures and remain consistent, even in adverse situations.

noob to master © copyleft