The introduction to the architecture documentation should list the driving forces that software architects must consider in their decisions.

This includes on the one hand the fulfillment of functional requirements of the stakeholders, on the other hand the fulfillment of or compliance with required constraints, always in consideration of the architecture goals.

Short description of the functional requirements.

Digest (or abstract) of the requirements documents.

Reference to complete requirements documents (incl. version identification and location).

Contents

A compact summary of the functional environment of the system. Answers the following questions (at least approximately):

• What happens in the system's environment?
• Why should the system exist? Why is the system valuable or important? Which problems does the system solve?

Motivation

From the point of view of the end users a system is created or modified to improve execution of a business activity.

This essential architecture driver must not be neglected even though the quality of an architecture is mostly judged by its level of fulfillment of non-functional requirements.

Form

Short textual description, probably in tabular use-case format.

The business context should in any case refer to the corresponding requirements documents.

Examples

Short descriptions of the most important:

• business processes,
• functional requirements,
• non-functional requirements and other constraints (the most important ones must be covered as architecture goals or are listed in the „Constraints“ section), and/or
• quantity structures.
• Background information

Here you can reuse parts of the requirements documents - but keep these excerpts short and balance readability against avoidance of redundancy.

Contents

The top ten goals for the architecture and/or constraints whose fulfillment is of highest importance to the major stakeholders.

Goals that define the architecture's quality could be:

• availability
• modifiability
• performance
• security
• testability
• usability

Motivation

If you as an architect do not know how the quality of your work can be judged …

Form

Simple tabular representation, ordered by priorities

Background Information

NEVER start developing an architecture if these goals have not been put into writing and have been signed by the major stakeholders.

We have endured projects lacking defined quality goals much too often. We do not like to suffer, therefore we are by now highly convinced that the few hours spent on collecting quality goals are well invested.

PH & GS.

Sources

The DIN/ISO 9126 Standard contains an extensive set of possible quality goals.

For everybody who is not interested in this level of detail: a readable excerpt is contained in „Agile Software-Entwicklung für Embedded Real-Time Systems mit der UML“ (Hruschka, Rupp, Carl- Hanser-Verlag, 2002 on page 9. PH

Contents

A list of the most important persons or organizations that are affected by can contribute to the architecture.

Motivation

If you do not know the persons participating in or concerned with the project you may get nasty surprises later in the development process.

Form

Simple table with role names, person names, their knowledge as pertaining to architecture, their availability, etc.

Examples

see e.g. VOLERE-Stakeholder table in the downloads on www.arc42.de or see Chapter 5.2 in „Requirements- Engineering und -Management“ by Chris Rupp .

Contents

Binds that constrain software architects in their freedom of design or development process.

Motivation

Architects should know exactly where they are free in their design decisions and where they must adhere to constraints.

Constraints must always be dealt with; they may be negotiable, though.

Form

Informal lists, structured by the sub-sections of this section.

Examples

see sub-sections

Background Information

In the optimal case constraints are defined by requirements. In any case, at least the architects must be aware of constraints.

The influ

Contents

List all technical constraints in this section. This category covers hard- and software infrastructure, applied technologies (operating systems, middleware, databases, programming languages, …).

Examples

Contents

Enter all organizational, structural, and resource-related constraints. This category also covers standards and legal constraints that you must comply with.

Examples

Agreements|Requirements or standards related to availability or required service levels?|

Contents

List all conventions that are relevant for the software architecture's development.

Form

Either insert the conventions directly in this document or refer to relevant other documents.

Examples

• Coding guidelines
• Documentation guidelines
• Guidelines for version and configuration management
• Naming conventions

Contents

The context view defines the boundaries of the system under development to distinguish it from neighboring systems. It thereby identifies the system's relevant external interfaces.

Make sure that the interfaces are specified with all their relevant aspects (what is communicated, in which format is it communicated, what is the transport medium, …), even though some popular diagrams (such as the UML use case diagram) represent only a few aspects of the interface.

Motivation

The interfaces to neighboring systems belong to the most critical aspects of a project. Ensure early on that you have understood them in their entirety.

Form

• Various context diagram (see below)
• Lists of neighboring systems and their interfaces.

Contents

Identify all¹⁾ neighboring systems and specify all logical/business data that is exchanged with the system under development. Add data formats and communication protocols with neighboring systems and the general environment if these are not specified in detail with the relevant components.

Motivation

Understanding of the information exchange with neighboring systems.

Form

Logical context diagram,

in UML e.g. simulated by class diagrams, use case diagrams, communications diagrams - i.e. all diagrams that represent the system as a black box and explain its interfaces to neighboring systems (in varying degrees of detail).

Examples

Contents

Specification of the communications channels between your system, its neighboring systems, and the environment.

Motivation

Understanding of the media used for information exchange with neighboring systems, and the environment.

Form

E.g. UML deployment diagram describing channels to neighboring systems

Examples

Contents

A summary and explanation of the fundamental solution ideas and strategy.

Motivation

Most architectures are based upon some specific solution ideas or strategies. These ideas should be familiar to everyone involved into the architecture.

Form

• Diagrams and / or text, as appropriate

Contents

Static decomposition of the system into building blocks (modules, components, subsystems, subsidiary systems, classes, interfaces, packages, libraries, frameworks, layers, partitions, tiers, functions, macros, operations, data structures, …) and the relationships thereof.

Motivation

This is the most important view, that must be part of each architecture documentation. In building construction this would be the floor plan.

Form

The building block view is a hierarchical collection of black box and white box descriptions as shown in the following diagram:

graphics1

Level 1 contains the white box description of the overall system (system under development / SUD) made up of black box descriptions of the system's building blocks.

Level 2 zooms into the building blocks of Level 1 and is thus made up of the white box descriptions of all building blocks of Level 1 together with the black box descriptions of the building blocks of Level 2.

Level 3 zooms into the building blocks of Level 2, etc.

The section is structured as follows:

White Box Template:

Contains multiple building blocks with corresponding black box descriptions.

One or more black box templates:

Each building block appearing in the white box template should be described as follows:

• Purpose / Responsibility:
• Interface(s):
• Implemented requirements:
• Variability:
• Performance attributes:
• Repository / Files:
• Other administrative information: Author, Version, Date, Revision History
• Open issues:

Here you describe the white box view of level 1 according to the white box template. The structure is given below.

The overview diagram describes the inner structure of the overall system in terms of building blocks 1 - n, as well as their relationships and interdependencies.

It is also useful to list the most important reasons that led to this structure, esp. as relevant to the interdependencies / relationships among the building blocks at this level.

You should also mention rejected alternatives incl. reasons for their rejection.

The following diagram shows the main building blocks of the system and their interdependencies:

<insert overview diagram here>

Comments regarding structure and interdependencies at Level 1:

Structure according to black box template:

• Purpose / Responsibility:
• Interface(s):
• Implemented requirements:
• Variability:
• Performance attributes:
• Repository / Files:
• Other administrative information: Author, Version, Date, Revision History
• Open issues:

<insert the building block's black box template here>

<insert the building block's black box template here>

<insert the building block's black box template here>

<insert the building block's black box template here>

Describe all building blocks comprising level 1 as a series of white box templates. The structure is given below for three building blocks and should be duplicated as needed.

Shows the inner workings of the building block in form of a diagrams with local building blocks 1 - n, as well as their relationships and interdependencies.

It is also useful to list the most important reasons that led to this structure, esp. as relevant to the interdependencies / relationships among the building blocks at this level.

You should also mention rejected alternatives incl. reasons for their rejection.

<insert diagram of building block 1 here>

Building Block Name 1.1 (Black Box Description)

Structure according to black box template:

• Purpose / Responsibility:
• Interface(s):
• Implemented requirements:
• Variability:
• Performance attributes:
• Repository / Files:
• Other administrative information: Author, Version, Date, Revision History
• Open issues:

Building Block Name 1.2 (Black Box Description)

Structure according to black box template

…

Building Block Name 1.n (Black Box Description)

Structure according to black box template

Description of Relationships

Open Issues

…

<insert diagram of building block 2 here>

Building Block Name 2.1 (Black Box Description)

Structure according to black box template

Building Block Name 2.2 (Black Box Description)

Structure according to black box template

…

Building Block Name 2.n (Black Box Description)

Structure according to black box template

Description of Relationships

Open Issues

…

<insert diagram of building block 3 here>

Building Block Name 3.1 (Black Box Description)

Structure according to black box template

Building Block Name 3.2 (Black Box Description)

Structure according to black box template

…

Building Block Name 3.n (Black Box Description)

Structure according to black box template

Description of Relationships

Open Issues

Describe all building blocks comprising level 2 as a series of white box templates. The structure is identical to the structure of level 2. Duplicate the corresponding sub-sections as needed.

Simply use this section structure for any additional levels you would like to describe.

Contents

alternative terms:

• Dynamic view
• Process view
• Workflow view

This view describes the behavior and interaction of the system's building blocks as runtime elements (processes, tasks, activities, threads, …).

Select interesting runtime scenarios such as:

• How are the most important use cases executed by the architectural building blocks?
• Which instances of architectural building blocks are created at runtime and how are they started, controlled, and stopped.
• How do the system's components co-operate with external and pre-existing components?
• How is the system started (covering e.g. required start scripts, dependencies on external systems, databases, communications systems, etc.)?

Note: The main criterion for the choice of possible scenarios (sequences, workflows) is their architectural relevancy. It is not important to describe a large number of scenarios. You should rather document a representative selection.

Candidates are:

• The top 3 - 5 use cases
• System startup
• The system's behavior on its most important external interfaces
• The system's behavior in the most important error situations

Motivation

Esp. for object-oriented architectures it is not sufficient to specify the building blocks with their interfaces, but also how instances of building blocks interact during runtime.

Form

Document the chosen scenarios using UML sequence, activity or communications diagrams. Enumerated lists are sometimes feasible.

Using object diagrams you can depict snapshots of existing runtime objects as well as instantiated relationships. The UML allows to distinguish between active and passive objects.

• Runtime diagram (or other adequate description of scenario!)
• Description of the notable aspects of the interactions between the building block instances depicted in this diagram.

• Runtime diagram (or other adequate description of scenario!)
• Description of the notable aspects of the interactions between the building block instances depicted in this diagram.

• Runtime diagram (or other adequate description of scenario!)
• Description of the notable aspects of the interactions between the building block instances depicted in this diagram.

Contents

This view describes the environment within which the system is executed. It describes the geographic distribution of the system or the structure of the hardware components that execute the software. It documents workstations, processors, network topologies and channels, as well as other elements of the physical system environment. The deployment view shows the system from the operator's point of view.

Please explain how the systems' building blocks are aggregated or packaged into deployment artifacts or deployment units.

Motivation

Software is not much use without hardware. The minimum that is needed by you as a software architect is sufficient detail of the underlying (hardware) deployment so that you can assign each software building block that is relevant for the system's operations to some hardware element. (This also holds for any COTS that is a prerequisite for the operations of the overall system.) These models should enable the operator to properly install the software.

Form

The UML provides deployment diagrams for describing this view. Use these - possibly in a nested manner if necessary. (The top level deployment diagram should already be part of your context view, showing your infrastructure as a single black box. Here you are zooming into this black box with additional deployment diagrams.)

Diagrams by your hardware-oriented colleagues who describe processors and channels are also usable. You should abstract these to aspects relevant for software deployment.

• Shows the deployment of the overall system to 1 - n processors or sites as well as the physical connections among these elements.
• Lists the most important reasons that led to this deployment structure, i.e. the specific selection of nodes and channels.
• Should also mention rejected alternatives incl. reasons for their rejection.

Structure according to node template:

• Description
• Performance attributes
• Assigned software building blocks
• Other administrative information
• Open issues

Structure according to node template:

Structure according to node template:

Contents

Specification of the channel's attributes, as relevant for software architecture.

Motivation

Specify at least those attributes of the communications channels that you need for proving fulfillment of non-functional requirements such as maximal throughput, probability for faults, etc.

Form

Use a structure similar to the node template.

Often you will refer to a standard (e.g. CAN-Bus, 10Mbit Ethernet, IEEE 1394, …).

Contents

Additional deployment diagrams with similar structure as above.

Motivation

To describe additional details of the infrastructure, as needed by software deployment.

Sometimes a hierarchical decomposition of building blocks is insufficient for giving an overview of detailed interdependencies between individual building blocks. The following sections are intended to describe generic or specific dependencies among any set of building blocks - possibly even across different levels.

We call a dependency generic if it appears more than once in the architecture, and specific if it is unique.

Form:

Use building block models (class diagrams, package diagrams, component diagrams, etc.) and related descriptions in the same way as in the hierarchical decomposition.

Often it is pracital to support understandability by adding specific rruntime views to these recurring structures.

<insert diagram and descriptions here>

<insert diagram and descriptions here>

Contents

The following chapters cover examples of frequent cross-cutting concerns or aspects.

Fill in these chapters if there is NO building block that covers this aspect. If some of the aspects are not relevant for your project mention this fact instead of removing the section.

Motivation

Some aspects cannot be „factored“ into a separate building block of the architecture (e.g. the topic „security“). This section of the template is the location where you can cover all concepts for such topics in a central place.

Form

.. can be varied. Some concept articles with free structure, some wide-ranging models/scenarios using notations that are also applied in architecture views.

Persistency means moving data from (volatile) memory to a durable storage medium (and back).

Some of the data that a software system is processing must be written to and read from persistent storage media.

• Volatile storage media (main memory or cache) are not designed for permanent storage. Data is lost if the hardware is switched off.
• The amount of data processed by commercial software systems normally exceeds the capacity of main memory.
• Hard disks, optical media and tapes often contain large amounts of existing business data that represent a significant investment.

Persistency is a technical issue that normally does not appear as part of the actual business functionality. An architect must deal with this issue nevertheless because most software systems require efficient access to persistently stored data. This is relevant for essentially all commercial and most technical systems; embedded systems on the other hand often differ in their data management requirements.

Software systems that are used interactively by (human) users require a user interface. These can be graphical, textual, or voice user interfaces.

Ergonomics of software systems deals with the improvement (optimization) of their usability with respect to objective and subjective factors. Key ergonomic factors are user interface, reactivity (subjective performance) as well as availability and robustness of the system.

Flow control of software systems is related to visible flows (on the - graphical - user interface) as well as the flow of background activities. Therefore this section should cover control of the user interface as well as control of workflows.

A transactions is a sets of operations or activities that must be processed either in its entirety or not at all. The term is especially relevant in the database area with the important notion of ACID-transactions (atomic, consistent, isolated, durable).

A session identifies an active connection between a client and a server. The session state must be preserved, which is esp. important if stateless protocols such as HTTP are used for communications. Session handling is a critical challenge esp. for intra- and internet-systems and can strongly influence the performance of a system.

The security of software systems deals with mechanisms that ensure data confidentiality, integrity, and availability.

Typical issues are:

• How can data be protected during transport (e.g. via open networks such as the internet)?
• How can communicating entities ensure mutual trust?
• How can communicating entities identify each other and be protected against faked identities?
• How can communicating entities prove data provenience or certify validity of data?

The topic of IT-security often touches upon legal aspects, sometimes even international law.

Communication: Exchange of data between system components. Covers communications within one process or address space, between different processes (inter-process communication - IPC), and between different systems.

Integration: Combination of existing systems in a new context. Also known as: (Legacy) Wrapper, Gateway, Enterprise Application Integration (EAI).

Distribution: Design of software systems whose parts are executed on different - physically separated - hardware systems.

Distribution covers issues such as calling methods on remote systems (remote procedure call - RPC or remote method invocation - RMI), the transfer of data or documents among distributed parties, the choice of optimal modes of interaction or communications patterns (such as synchronous / asynchronous, publish-subscribe, peer-to-peer).

How are exceptions and errors handled systematically and consistently?

How can the system reach a consistent state after an error? Is this done automatically or is manual interaction required?

This aspect is also related to logging and tracing,

Which kind of exceptions and errors are handled by the system? Which kind of errors are forwarded to which external interface and which are handled fully internally?

How are the exception handling mechanisms of your programming language used? Do you use checked or unchecked exceptions?

Larger software systems are often executed in controlled environments (data centers) under oversight of operators or administrators. These stakeholders require specific information on the applications' states during runtime as well as special means of control and configuration.

There are two ways of documenting an application's status during runtime: Logging and Tracing. In both cases the application is extended with function or method calls that write state information, but there is a difference in their usage:

• Logging can cover business or technical aspects or any combination of both.
• Business logs are normally prepared for end users, administrators or operators. They contain information on the business processes that are executed by the application. This kind of logging may also be related to auditing.
• Technical logs contain information for operators or developers. These are used for error detection and system optimization.
• Tracing is intended to provide debugging information for developers or support personnel. It is primarily used for error detection and analysis.

How do you deal with business logic and business rules? Is business logic implemented in the corresponding business classes or is it handled in a central component? Do you use a rule engine for the interpretation of business rules (production system, forward-/backward-chaining)?

The flexibility of a software system is influenced by its configurability, i.e. the possibility to make certain decisions about usage of the system at a late point in time.

Configurability can occur at the following events:

• During development: For example server, file, or directory names could be stored directly in the code („hard-coded“).
• During deployment or installation: Configuration information for a specific installation (such as the installation path) can be given.
• At system startup: Information can be read dynamically before or during system startup.
• During application execution: Configuration information is queried or read during runtime.

Applications can be executed in parallel processes or threads. This creates a need for synchronization points. The theory of parallel processing serves as a foundation for this aspect. The architecture and implementation of parallel systems needs to consider many technical details such as address spaces, applied mechanisms for synchronization - guards, semaphores, etc. - processes and threads, parallelism in the operating system, parallelism in virtual machines. etc.

This section covers support for usage of the system in different countries, i.e. adjusting the system to country specific attributes. Internationalization (often abbreviated as „i18n“ where „18“ refers to the eighteen characters between the I and the n) covers translation of text, usage of character encodings, display of fonts, writing of numbers and dates, and other (external) aspects.

In many cases a new software system is intended to replace an existing legacy system. As an architect you should not only consider your shiny new architecture but also all organizational and technical aspects that must be considered for the introduction or migration of the architecture.

Examples:

• Concept, process, or tools for data transfer and initial data creation.
• Concept for system introduction or temporary parallel operations of legacy system and new system.

Is it necessary to migrate existing data? How do you execute any needed syntactic or semantic transformations?

Support for simple (and if possible automated) tests. This aspect is the basis for the important implementation pattern of „continuous integration“. Projects should support at least daily build-and-test cycles. Important keywords for this aspect are unit tests and mock objects.

How and where do you check plausibility and validity of (input) data, esp. user inputs?

How and where do you use code generators to create parts of the system from models or domain specific languages (DSL's)?

How is the overall system created from is (source code) building blocks? Which repositories contain source code, where are configuration files, test cases, test data and build scripts (make, ant, maven) stored?

Contents

Document all important design decisions and their reasons!

Motivation

It is advantageous if all important design decisions can be found in one place. It is up to you to decide if a decision should be documented here or rather locally (e.g. in the white box descriptions of building blocks). In any case avoid redundancies.

Form

Informal list, if possible ordered by the decisions' importance for the reader.

This chapter summarizes all you (or other stakeholders) might need to systematically evaluate the architecture against the quality requirements.

Content

The quality tree ( as defined in ATAM) with quality / evaluation scenarios as leafs.

Motivation

When you want to evaluate the quality (especially risks to certain quality attributes) with methods like ATAM, you need to systematically refine your quality goals (from chapter 1.2). The quality tree shows the top-down refinement of the stakeholder-specific notion of quality.

Form

We personally prefer mindmaps to a pure tree-like structure, as mindmaps allow arbitrary cross-references between scenarios, attributes and intermediate nodes.

Often it is difficult to assign scenarios to single quality attributes, as the scenario refers to several qualities at once. Simply draw references from such scenarios to all affected nodes!

Contents

Scenarios describe a system's reaction to a stimulus in a certain situation. They thus characterize the interaction between stakeholders and the system. Scenarios operationalize quality criteria and turn them into measurable quantities.

Two scenarios are relevant for most software architects:

• Usage scenarios (also called application scenarios or use case scenarios) the system's runtime reaction to a certain stimulus. This also includes scenarios that describe the system's efficiency or performance. Example: The system reacts to a user's request within one second.
• Change scenarios describe a modification of the system or of its immediate environment. Example: Additional functionality is implemented or requirements for a quality attribute change.

If you design safety critical systems a third type of scenarios is important for you:

• Boundary or stress scenarios describe how the system reacts to exceptional conditions. Examples: How does the system react to a complete power outage, a serious hardware failure, etc.

Figure: Schematic depiction of scenarios (cf. [Bass+03])

Scenarios comprise the following major parts (according to [Starke05], original structure from [Bass+03]):

• Stimulus: Describes a specific interaction between the (stimulating) stakeholder and the system. Example: A user calls a functions, a developer implements an extension, an administrator installs or configures the system.
• Source of the stimulus: Describes where the stimulus comes from. Examples: internal or external, user, operator, attacker, manager.
• Environment: Describes the system's state at the time of arrival of the stimulus. This should list all preconditions that are necessary for comprehension of the scenario. Examples: Is the system under normal or maximal load? Is the data base available or down? Are any users online?
• System artifact: Describes the part of the system is affected by the stimulus. Examples: The whole system, the data base, the web server.
• System response: Describes the system's reaction to the stimulus as determined by the architecture. Examples: Is the function called by the user executed. How long does the developer need for implementation? Which parts of the system are affected by the installation / configuration?
• Response measure: Describes how the response can be measured or evaluated. Examples: Downtime in hours, correctness yes/no, time for code change in person days, reaction time in seconds.

Motivation

You need scenarios for the evaluation and review of architectures. They take the role of a „benchmark“ and aid in measuring the architecture's achievement of its objectives regarding the non-functional requirements and quality attributes.

Form

Tabular or free text. Explicitly highlight the scenario's elements (source, environment, artifact, response, measure).

Background Information

There are relations between scenarios and the runtime view. Often you can use scenarios of the runtime view fully or as a basis for evaluation. Evaluation scenarios additionally contain response measures that are often not considered in the pure execution focus of runtime scenarios.

Contents

A list of identified technical risks, ordered by priority

Motivation

„Risk management is project management for grown-ups“ (Tim Lister, Atlantic Systems Guild.) This should be your motto for systematic detection and evaluation of technical risks in the architecture, which will be needed by project management as part of the overall risk analysis.

Form

List of risks with probability of occurrence, amount of damage, options for risk avoidance or risk mitigation, …

Contents

The most important terms of the software architecture in alphabetic order.

Motivation

It should not be necessary to explain the usefulness of a glossary …

Form

A simple table with columns <Term> and <Definition>