blob: 868e01229dec32745c536473970c425cbaa7fc89 [file] [log] [blame]
<?xml version="1.0" encoding="UTF-8"?>
<org.eclipse.epf.uma:ContentDescription xmi:version="2.0" xmlns:xmi="http://www.omg.org/XMI" xmlns:org.eclipse.epf.uma="http://www.eclipse.org/epf/uma/1.0.3/uma.ecore" epf:version="1.0.0" xmi:id="_lbGQwMM3EdmSIPI87WLu3g" name="layering,_0gpkAMlgEdmt3adZL5Dmdw" guid="_lbGQwMM3EdmSIPI87WLu3g" changeDate="2007-02-26T12:37:43.312+0000" version="1.0.0">
<mainDescription>&lt;p&gt;
Layering logically partitions the subsystems into a number of sets, with certain rules regarding how relationships can
be formed between the layers. Layering provides a way to restrict inter-subsystem dependencies, with the result that
the system is more loosely coupled and, therefore, more easily maintained.
&lt;/p&gt;
&lt;p&gt;
Consider the number and purpose of the layers carefully. Do not over-complicate the solution by defining more layers
than are needed to meet the needs of the solution. More layers can always be added in the future to meet new
requirements. Removing layers is not always as easy and may introduce risks into the project.
&lt;/p&gt;
&lt;p&gt;
The criteria for grouping subsystems follow a few patterns:
&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;b&gt;Visibility&lt;/b&gt;&lt;strong&gt;:&lt;/strong&gt; Subsystems may depend only on subsystems in the same layer and the next-lower
layer.
&lt;/li&gt;
&lt;li&gt;
&lt;b&gt;Volatility&lt;/b&gt;&lt;strong&gt;:&lt;/strong&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;b&gt;In the highest layers&lt;/b&gt;, put elements that vary when user requirements change.
&lt;/li&gt;
&lt;li&gt;
&lt;b&gt;In the lowest layers&lt;/b&gt;, put elements that vary when the implementation platform changes (hardware,
language, operating system, database, and so forth).
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Sandwiched in the middle&lt;/strong&gt;, put elements that are generally applicable across wide ranges of
systems and implementation environments.
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Add layers&lt;/strong&gt; when additional partitions within these broad categories help to organize the
model.
&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;b&gt;Generality&lt;/b&gt;&lt;strong&gt;:&lt;/strong&gt; Abstract model elements tend to be placed lower in the model. If not
implementation-specific, they tend to gravitate toward the middle layers.
&lt;/li&gt;
&lt;li&gt;
&lt;b&gt;Number of layers.&lt;/b&gt; For a small system, three layers are sufficient. For a complex system, five to seven
layers are usually sufficient. For any degree of complexity, more than 10 layers should be viewed with suspicion
that increases with the number of layers. The table that follows gives you a few guidelines.
&lt;/li&gt;
&lt;/ul&gt;
&lt;p align=&quot;center&quot;&gt;
&lt;strong&gt;Guideline for number of layers according to number of classes&lt;/strong&gt;
&lt;/p&gt;
&lt;div align=&quot;center&quot;&gt;
&lt;table
style=&quot;BORDER-RIGHT: rgb(128,128,128) 1px solid; BORDER-TOP: rgb(128,128,128) 1px solid; BORDER-LEFT: rgb(128,128,128) 1px solid; BORDER-BOTTOM: rgb(128,128,128) 1px solid&quot;
cellspacing=&quot;0&quot; bordercolordark=&quot;#808080&quot; cellpadding=&quot;4&quot; width=&quot;58%&quot; bordercolorlight=&quot;#808080&quot; border=&quot;1&quot;&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;th width=&quot;40%&quot;&gt;
&lt;p align=&quot;center&quot; scope=&quot;col&quot;&gt;
&lt;b&gt;Number of Classes&lt;/b&gt;
&lt;/p&gt;
&lt;/th&gt;
&lt;th width=&quot;60%&quot;&gt;
&lt;p align=&quot;center&quot; scope=&quot;col&quot;&gt;
&lt;b&gt;Number of Layers&lt;/b&gt;
&lt;/p&gt;
&lt;/th&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td width=&quot;40%&quot;&gt;
&lt;p align=&quot;center&quot;&gt;
0 - 10
&lt;/p&gt;
&lt;/td&gt;
&lt;td width=&quot;60%&quot;&gt;
&lt;blockquote&gt;
&lt;blockquote&gt;
&lt;p&gt;
No layering needed
&lt;/p&gt;
&lt;/blockquote&gt;
&lt;/blockquote&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td width=&quot;40%&quot;&gt;
&lt;p align=&quot;center&quot;&gt;
10 - 50
&lt;/p&gt;
&lt;/td&gt;
&lt;td width=&quot;60%&quot;&gt;
&lt;blockquote&gt;
&lt;blockquote&gt;
&lt;p&gt;
2 layers
&lt;/p&gt;
&lt;/blockquote&gt;
&lt;/blockquote&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td width=&quot;40%&quot;&gt;
&lt;p align=&quot;center&quot;&gt;
25 - 150
&lt;/p&gt;
&lt;/td&gt;
&lt;td width=&quot;60%&quot;&gt;
&lt;blockquote&gt;
&lt;blockquote&gt;
&lt;p&gt;
3 layers
&lt;/p&gt;
&lt;/blockquote&gt;
&lt;/blockquote&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td width=&quot;40%&quot;&gt;
&lt;p align=&quot;center&quot;&gt;
100 - 1000
&lt;/p&gt;
&lt;/td&gt;
&lt;td width=&quot;60%&quot;&gt;
&lt;blockquote&gt;
&lt;blockquote&gt;
&lt;p&gt;
4 layers
&lt;/p&gt;
&lt;/blockquote&gt;
&lt;/blockquote&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;
&lt;/div&gt;
&lt;p&gt;
Failure to restrict dependencies according to Visibility criteria mentioned above can cause architectural degradation
and make the system difficult to extend and maintain.
&lt;/p&gt;
&lt;p&gt;
Exceptions include cases where subsystems need direct access to lower-layer services. Make a decision about how to
handle primitive services that are needed throughout the system, such as printing, sending messages, and so forth.
There is little value in restricting messages to lower layers if the solution is to effectively implement call
pass-throughs in the intermediate layers.
&lt;/p&gt;
&lt;h4&gt;
&lt;a id=&quot;PartitioningPatterns&quot; name=&quot;PartitioningPatterns&quot;&gt;Partitioning patterns&lt;/a&gt;
&lt;/h4&gt;
&lt;p&gt;
Within the top layers of the system, additional partitioning may help organize the model. The following guidelines for
partitioning present different issues to consider:
&lt;/p&gt;
&lt;p&gt;
&lt;b&gt;User organization&lt;/b&gt;&lt;strong&gt;:&lt;/strong&gt; Subsystems may be organized along lines that mirror the organization of
functionality in the business organization (partitioning occurs along departmental lines). This partitioning often
occurs early in the design because an existing enterprise model that is strongly partitioned according to the structure
of the organization. This pattern usually affects only the top few layers of application-specific services and often
disappears as the design evolves.
&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;
Partitioning along user-organization lines can be a good starting point for the model.
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;
The structure of the user organization is not stable over a long period of time because business
reorganizations occur; therefore, it is not a good long-term basis for system partitioning. The internal
organization of the system should enable the system to evolve and be maintained independently of the
organization of the business that it supports.
&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;
&lt;b&gt;Areas of competence and skills:&lt;/b&gt; Subsystems may be organized to partition responsibilities for parts of the model
among different groups within the development organization. Typically, this occurs in the middle and lower layers of
the system, and reflects the need for specialization in skills during the development and support of an infrastructure
based on complex technology. Examples of such technologies include network and distribution management, database
management, communication management, and process control, among others. Partitioning along competence lines may also
occur in upper layers, where special competency in the problem domain is required to understand and support key
business functionality. Examples include telecommunication call management, securities trading, insurance claims
processing, and air traffic control, to name a few.
&lt;/p&gt;
&lt;p&gt;
&lt;b&gt;System distribution:&lt;/b&gt; Within any of the layers of the system, the layers may be further partitioned horizontally,
so to speak, to reflect the distribution of functionality.
&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;
Partitioning to reflect distribution of functionality can help you visualize the network communication that
will occur as the system runs.
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;
Partitioning to reflect distribution can also, however, make the system more difficult to change if the
deployment model changes significantly.
&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;
&lt;b&gt;Secrecy areas&lt;/b&gt;&lt;strong&gt;:&lt;/strong&gt; Some applications, especially those requiring special security clearance to
develop or support, require additional partitioning according to security access privileges. Software that controls
access to secrecy areas must be developed and maintained by personnel with appropriate clearance. If the number of
people with this background on the project is limited, the functionality requiring special clearance must be
partitioned into subsystems that will be developed independently from other subsystems, with the interfaces to the
secrecy areas the only visible aspect of these subsystems.
&lt;/p&gt;
&lt;p&gt;
&lt;b&gt;Variability areas:&lt;/b&gt; Functionality that is likely to be optional, and therefore delivered only in some variants of
the system, should be organized into independent subsystems that are developed and delivered independently from the
mandatory functionality of the system.
&lt;/p&gt;</mainDescription>
</org.eclipse.epf.uma:ContentDescription>