Database abstraction layer: Difference between revisions

Revision as of 14:47, 24 October 2019

Latest revision as of 05:14, 24 April 2024

(6 intermediate revisions by 6 users not shown)

Line 1:

{{Short description|Application programming interface}}

{{More citations needed|date=September 2014}}

A '''database abstraction layer''' ('''DBAL'''<ref>{{cite book|author1first1=Tim |last1=Ambler|author2first2=Nicholas |last2=Cloud|title=JavaScript Frameworks for Modern Web Dev|url=https://books.google.com/books?id=2IfDCgAAQBAJ&pg=PA346|year=2015|publisher=Apress|isbn=978-1-4842-0662-1|page=346}}</ref> or '''DAL''') is an [[application programming interface]] which unifies the communication between a computer application and [[database]]s such as [[MSSQL|SQL Server]], [[IBM DB2|DB2Db2]], [[MySQL]], [[PostgreSQL]], [[Oracle database|Oracle]] or [[SQLite]]. Traditionally, all database vendors provide their own interface that is tailored to their products. It is up to the application programmer to implement code for the database interfaces that will be supported by the application. Database abstraction layers reduce the amount of work by providing a consistent API to the developer and hide the database specifics behind this interface as much as possible. There exist many abstraction layers with different interfaces in numerous programming languages. If an application has such a layer built in, it is called '''database-agnostic'''.<ref>{{cite web |url=http://searchdatamanagement.techtarget.com/definition/database-agnostic |title = What is database-agnostic? - Definition from WhatIs.com}}</ref>

== Database levels of abstraction ==

=== Physical level (lowest level) ===

The lowest level connects to the database and performs the actual operations required by the users. At this level the conceptual instruction has been translated into multiple instructions that the database understands. Executing the instructions in the correct order allows the DAL to perform the conceptual instruction.

Line 18:

=== External or view level ===

The external level is exposed to users and developers and supplies a consistent pattern for performing database operations.

<ref>{{Cite web|url=http://www.dmst.aueb.gr/dds/etech/db/abstr.htm|title = Levels of Abstraction}}</ref> Database operations are represented only loosely as SQL or even database access at this level.

Every database should be treated equally at this level with no apparent difference despite varying physical data types and operations.

Line 30:

An example of a database abstraction layer on the language level would be [[ODBC]] that is a platform-independent implementation of a database abstraction layer. The user installs specific [[driver software]], through which ODBC can communicate with a database or set of databases. The user then has the ability to have programs communicate with ODBC, which then relays the results back and forth between the user programs and the database. The downside of this abstraction level is the increased overhead to transform statements into constructs understood by the target database.

Alternatively, there are thin wrappers, often described as ''lightweight'' abstraction layers, such as OpenDBX<ref>{{cite web |url=https://www.linuxnetworks.de/doc/index.php?title=OpenDBX |title=OpenDBX |author==<!--Not stated--> |date=24 June 2012 |website=linuxnetworks.de |access-date=26 July 2018}}</ref> and libzdb.<ref>{{cite web |url=http://www.tildeslash.com/libzdb/ |title=Libzdb |author==<!--Not stated--> |access-date=26 July 2018 |year=2018 |website=tildeslash.com}}</ref> Finally, large projects may develop their own libraries, such as, for example, libgda<ref>{{cite web |url=http://www.gnome-db.org/ |title=GNOME-DB |author==<!--Not stated--> |access-date=26 July 2018 |date=12 June 2015 |quote=Libgda library [...] is mainly a database and data abstraction layer, and includes a GTK+ based UI extension, and some graphical tools.}}</ref> for [[GNOME]].

== Arguments for ==

=== Development period ===

Software developers only have to know the database abstraction layer's API instead of all APIs of the databases their application should support. The more databases should be supported the bigger is the time saving.

=== Wider potential install-base ===

Using a database abstraction layer means that there is no requirement for new installations to utilise a specific database, i.e. new users who are unwilling or unable to switch databases can deploy on their existing infrastructure.

=== Future-proofing ===

As new database technologies emerge, software developers won't have to adapt to new interfaces.

=== Developer testing ===

A production database may be replaced with a desktop-level implementation of the data for developer-level unit tests.

=== Added Database Features ===

Depending on the database and the DAL, it may be possible for the DAL to add features to the database. A DAL may use database programming facilities or other methods to create standard but unsupported functionality or completely new functionality. For instance, the DBvolution DAL implements the standard deviation function for several databases that do not support it.

== Arguments against ==

=== Speed ===

Any abstraction layer will reduce the overall speed more or less depending on the amount of additional code that has to be executed. The more a database layer abstracts from the native database interface and tries to emulate features not present on all database backends, the slower the overall performance. This is especially true for database abstraction layers that try to unify the query language as well like ODBC.

=== DependencyArguments ===

=== SpeedIn favor ===

A database abstraction layer provides yet another functional dependency for a software system, i.e. a given database abstraction layer, like anything else, may eventually become obsolete, outmoded or unsupported.

Software* Development period: software developers only have to know the database abstraction layer's API instead of all APIs of the databases their application should support. The more databases should be supported the bigger is the time saving.

Using* Wider potential install-base: using a database abstraction layer means that there is no requirement for new installations to utilise a specific database, i.e. new users who are unwilling or unable to switch databases can deploy on their existing infrastructure.

As* Future-proofing: as new database technologies emerge, software developers won't have to adapt to new interfaces.

A* Developer testing: a production database may be replaced with a desktop-level implementation of the data for developer-level unit tests.

Depending* Added Database Features: depending on the database and the DAL, it may be possible for the DAL to add features to the database. A DAL may use database programming facilities or other methods to create standard but unsupported functionality or completely new functionality. For instance, the DBvolution DAL implements the standard deviation function for several databases that do not support it.

=== MaskedAgainst operationsit ===

Any* Speed: any abstraction layer will reduce the overall speed more or less depending on the amount of additional code that has to be executed. The more a database layer abstracts from the native database interface and tries to emulate features not present on all database backends, the slower the overall performance. This is especially true for database abstraction layers that try to unify the query language as well like ODBC.

Database abstraction layers may limit the number of available database operations to a subset of those supported by the supported database backends. In particular, database abstraction layers may not fully support database backend-specific optimizations or debugging features. These problems magnify significantly with database size, scale, and complexity.

A* Dependency: a database abstraction layer provides yet another functional dependency for a software system, i.e. a given database abstraction layer, like anything else, may eventually become obsolete, outmoded or unsupported.

Database* Masked operations: database abstraction layers may limit the number of available database operations to a subset of those supported by the supported database backends. In particular, database abstraction layers may not fully support database backend-specific optimizations or debugging features. These problems magnify significantly with database size, scale, and complexity.

== See also ==

* [[Object-relationalObject–relational mapping]]

== References ==

{{reflist}}

<references />

{{Database}}