||This article includes a list of references, but its sources remain unclear because it has insufficient inline citations. (September 2010)|
In computing and systems design a loosely coupled system is one in which each of its components has, or makes use of, little or no knowledge of the definitions of other separate components. The notion was introduced into organizational studies by Karl Weick. Sub-areas include the coupling of classes, interfaces, data, and services.
Coupling refers to the degree of direct knowledge that one class has of another. This is not meant to be interpreted as encapsulation vs. non-encapsulation. It is not a reference to one class's knowledge of another class's attributes or implementation, but rather knowledge of that other class itself.
Strong coupling occurs when a dependent class contains a pointer directly to a concrete class which provides the required behavior. The dependency cannot be substituted, or its "signature" changed, without requiring a change to the dependent class. Loose coupling occurs when the dependent class contains a pointer only to an interface, which can then be implemented by one or many concrete classes. The dependent class's dependency is to a "contract" specified by the interface; a defined list of methods and/or properties that implementing classes must provide. Any class that implements the interface can thus satisfy the dependency of a dependent class without having to change the class. This allows for extensibility in software design; a new class implementing an interface can be written to replace a current dependency in some or all situations, without requiring a change to the dependent class; the new and old classes can be interchanged freely. Strong coupling does not allow this.
For comparison, this diagram illustrates the alternative design with strong coupling between the dependent class and a provider:
Other forms of Loose Coupling
Computer programming languages having notions of either functions as the core module (see Functional programming) or functions as objects provide excellent examples of loosely coupled programming. Functional languages have patterns of Continuations, Closure (computer programming), or generators. See Clojure and Lisp as examples of function programming languages. Object oriented languages like Smalltalk and Ruby have code blocks, while Eiffel has agents. The basic idea is to objectify (encapsulate as an object) a function independent of any other enclosing concept (e.g. decoupling an object function from any direct knowledge of the enclosing object). See First-class function for further insight into functions as objects, which qualifies as one form of first-class function.
So, for example, in an object oriented language, when a function of an object is referenced as an object (freeing it from having any knowledge of its enclosing host object) the new function object can be passed, stored, and called at a later time. Recipient objects (to whom these functional objects are given) can safely execute (call) the contained function at their own convenience without any direct knowledge of the enclosing host object. In this way, a program can execute chains or groups of functional objects, while safely decoupled from having any direct reference to the enclosing host object.
Phone numbers are an excellent analog and can easily illustrate the degree of this decoupling.
For example: Some entity provides another with a phone number to call to get a particular job done. When the number is called, the calling entity is effectively saying, "Please do this job for me." The decoupling or loose coupling is immediately apparent. The entity receiving the number to call may have no knowledge of where the number came from (e.g. a reference to the supplier of the number). On the other side, the caller is decoupled from specific knowledge of the who they are calling, where they are, and knowing how the receiver of the call operates internally.
Carrying the example a step further, the caller might say to the receiver of the call, "Please do this job for me. Call me back at this number when you are finished." The `number' being offered to the receiver is referred to as a "Call-back". Again, the loose coupling or decoupled nature of this functional object is apparent. The receiver of the call-back is unaware of what or who is being called. It only knows that it can make the call and decides for itself when to call. In reality, the call-back may not even be to the one who provided the call-back in the first place. This level of indirection is what makes function objects an excellent technology for achieving loosely coupled programs.
Measuring data element coupling
The degree of the loose coupling can be measured by noting the number of changes in data elements that could occur in the sending or receiving systems and determining if the computers would still continue communicating correctly. These changes include items such as:
- adding new data elements to messages
- changing the order of data elements
- changing the names of data elements
- changing the structures of data elements
- omitting data elements
Methods for decreasing coupling
Loose coupling of interfaces can be dramatically enhanced when publishers of data transmit messages using a flexible file format such as XML or JSON to enable subscribers to publish clear definitions of how they subsequently use this data. For example, a subscriber could publish the collection of statements used to extract information from a publisher's messages by sharing the relevant XPath expressions used for data transformation, or the JSON Schema. This would allow a responsible data publisher to test whether their subscriber's extraction methods would fail when a published format changes.
Loose coupling of services can be enhanced by reducing the information passed into a service to the key data. For example, a service that sends a letter is most reusable when just the customer identifier is passed and the customer address is obtained within the service. This decouples services because services do not need to be called in a specific order (e.g. GetCustomerAddress, SendLetter)
Note that loose coupling is not universally positive. If systems are de-coupled in time using Message-oriented middleware, it is difficult to also provide transactional integrity. Data replication across different systems provides loose coupling (in availability), but creates issues in maintaining synchronisation.
- Coupling (computer science)
- Cohesion (computer science)
- Web Services
- Design pattern (computer science)
- ISO/IEC 11179 - metadata registry specification
- Data element
- Enterprise service bus
- Enterprise Messaging System
- Space-based architecture (SBA)
- Tightly Coupled Systems
- Loosely Coupled: The Missing Pieces of Web Services by Doug Kaye
- Service Oriented Architecture: A field Guide to Integrating XML and Web Services by Thomas Erl
- Karl Weick, "Educational organizations as loosely coupled systems", Administrative Science Quarterly, 21 (1976), 1-9 (part).
- "The Management of Organizational Change among Loosely Coupled Elements" (1982) by Karl Weick reprinted in his book Making Sense of the Organization (2001)
- James Douglas Orton and Karl E. Weick, Loosely Coupled Systems: A Reconceptualization, Academy of Management Review 15 (2):203-223 1990