Agile Testing Methods, Techniques, and Tools

Agile Testing Methods

There are certain testing practices that can be followed in every development project (agile or not) to produce quality products. These include writing tests in advance to express proper behaviour, focusing on early defect prevention, detection, and removal, and ensuring that the right test types are run at the right time and as part of the right test level. Agile practitioners aim to introduce these practices early. Testers in Agile projects play a key role in guiding the use of these testing practices throughout the lifecycle. 

Test-Driven Development, Acceptance Test-Driven Development, and Behaviour-Driven Development

Test-driven development, acceptance test-driven development, and behaviour-driven development are three complementary techniques in use among Agile teams to carry out testing across the various test levels. Each technique is an example of a fundamental principle of testing, the benefit of early testing and QA activities, since the tests are defined before the code is written. 

Test-Driven Development

Test-driven development (TDD) is used to develop code guided by automated test cases. The process for test-driven development is:

  • Add a test that captures the programmer’s concept of the desired functioning of a small piece of code
  • Run the test, which should fail since the code doesn’t exist
  • Write the code and run the test in a tight loop until the test passes
  • Refactor the code after the test is passed, re-running the test to ensure it continues to pass against the refactored code
  • Repeat this process for the next small piece of code, running the previous tests as well as the added tests

The tests written are primarily unit level and are code-focused, though tests may also be written at the integration or system levels. Test-driven development gained its popularity through Extreme Programming, but is also used in other Agile methodologies and sometimes in sequential lifecycles. It helps developers focus on clearly defined expected results. The tests are automated and are used in continuous integration.

Acceptance Test-Driven Development

Acceptance test-driven development defines acceptance criteria and tests during the creation of user stories. Acceptance test-driven development is a collaborative approach that allows every stakeholder to understand how the software component has to behave and what the developers, testers, and business representatives need to ensure this behaviour.

Acceptance test-driven development creates reusable tests for regression testing. Specific tools support creation and execution of such tests, often within the continuous integration process. These tools can connect to data and service layers of the application, which allows tests to be executed at the system or acceptance level. Acceptance test-driven development allows quick resolution of defects and validation of feature behaviour. It helps determine if the acceptance criteria are met for the feature.

Behaviour-Driven Development

Behaviour-driven development allows a developer to focus on testing the code based on the expected behaviour of the software. Because the tests are based on the exhibited behaviour from the software, the tests are generally easier for other team members and stakeholders to understand.

Specific behaviour-driven development frameworks can be used to define acceptance criteria based on the given/when/then format:

Given some initial context,

When an event occurs,

Then ensure some outcomes. 

From these requirements, the behaviour-driven development framework generates code that can be used by developers to create test cases. Behaviour-driven development helps the developer collaborate with other stakeholders, including testers, to define accurate unit tests focused on business needs. 

The Test Pyramid

A software system may be tested at different levels. Typical test levels are, from the base of the pyramid to the top, unit, integration, system, and acceptance. The test pyramid emphasises having a large number of tests at the lower levels (bottom of the pyramid) and, as development moves to the upper levels, the number of tests decreases (top of the pyramid). Usually unit and integration level tests are automated and are created using API-based tools. At the system and acceptance levels, the automated tests are created using GUI-based tools. The test pyramid concept is based on the testing principle of early QA and testing (i.e., eliminating defects as early as possible in the lifecycle). 

Testing Quadrants, Test Levels, and Testing Types

Testing quadrants, align the test levels with the appropriate test types in the Agile methodology. The testing quadrants model, and its variants, helps to ensure that all important test types and test levels are included in the development lifecycle. This model also provides a way to differentiate and describe the types of tests to all stakeholders, including developers, testers, and business representatives. 

In the testing quadrants, tests can be business (user) or technology (developer) facing. Some tests support the work done by the Agile team and confirm software behaviour. Other tests can verify the product. Tests can be fully manual, fully automated, a combination of manual and automated, or manual but supported by tools. The four quadrants are as follows: 

  • Quadrant Q1 is unit level, technology facing, and supports the developers. This quadrant contains unit tests. These tests should be automated and included in the continuous integration process.
  • Quadrant Q2 is system level, business facing, and confirms product behaviour. This quadrant contains functional tests, examples, story tests, user experience prototypes, and simulations. These tests check the acceptance criteria and can be manual or automated. They are often created during the user story development and thus improve the quality of the stories. They are useful when creating automated regression test suites.
  • Quadrant Q3 is system or user acceptance level, business facing, and contains tests that critique the product, using realistic scenarios and data. This quadrant contains exploratory testing, scenarios, process flows, usability testing, user acceptance testing, alpha testing, and beta testing. These tests are often manual and are user-oriented.
  • Quadrant Q4 is system or operational acceptance level, technology facing, and contains tests that critique the product. This quadrant contains performance, load, stress, and scalability tests, security tests, maintainability, memory management, compatibility and interoperability, data migration, infrastructure, and recovery testing. These tests are often automated.

During any given iteration, tests from any or all quadrants may be required. The testing quadrants apply to dynamic testing rather than static testing.

The Role of a Tester

Throughout this article, general reference has been made to Agile methods and techniques, and the role of a tester within various Agile lifecycles. This subsection looks specifically at the role of a tester in a project following a Scrum lifecycle. 

Teamwork 

Teamwork is a fundamental principle in Agile development. Agile emphasises the whole-team approach consisting of developers, testers, and business representatives working together. The following are organisational and behavioural best practices in Scrum teams:

  • Cross-functional: Each team member brings a different set of skills to the team. The team works together on test strategy, test planning, test specification, test execution, test evaluation, and test results reporting.
  • Self-organising: The team may consist only of developers, but, as noted before, ideally there would be one or more testers.
  • Co-located: Testers sit together with the developers and the product owner.
  • Collaborative: Testers collaborate with their team members, other teams, the stakeholders, the product owner, and the Scrum Master.
  • Empowered: Technical decisions regarding design and testing are made by the team as a whole (developers, testers, and Scrum Master), in collaboration with the product owner and other teams if needed.
  • Committed: The tester is committed to question and evaluate the product’s behaviour and characteristics with respect to the expectations and needs of the customers and users.
  • Transparent: Development and testing progress is visible on the Agile task board.
  • Credible: The tester must ensure the credibility of the strategy for testing, its implementation, and execution, otherwise the stakeholders will not trust the test results. This is often done by providing information to the stakeholders about the testing process.
  • Open to feedback: Feedback is an important aspect of being successful in any project, especially in Agile projects. Retrospectives allow teams to learn from successes and from failures.
  • Resilient: Testing must be able to respond to change, like all other activities in Agile projects.

These best practices maximise the likelihood of successful testing in Scrum projects.

Sprint Zero 

Sprint zero is the first iteration of the project where many preparation activities take place. The tester collaborates with the team on the following activities during this iteration:

  • Identify the scope of the project (i.e., the product backlog)
  • Create an initial system architecture and high-level prototypes
  • Plan, acquire, and install needed tools (e.g., for test management, defect management, test automation, and continuous integration)
  • Create an initial test strategy for all test levels, addressing (among other topics) test scope, technical risks, test types, and coverage goals
  • Perform an initial quality risk analysis
  • Define test metrics to measure the test process, the progress of testing in the project, and product quality
  • Specify the definition of “done”
  • Create the task board
  • Define when to continue or stop testing before delivering the system to the customer

Sprint zero sets the direction for what testing needs to achieve and how testing needs to achieve it throughout the sprints.

Integration 

In Agile projects, the objective is to deliver customer value on a continuous basis (preferably in every sprint). To enable this, the integration strategy should consider both design and testing. To enable a continuous testing strategy for the delivered functionality and characteristics, it is important to identify all dependencies between underlying functions and features.

Test Planning

Since testing is fully integrated into the Agile team, test planning should start during the release planning session and be updated during each sprint. Test planning for the release and each sprint should address the issues.

Sprint planning results in a set of tasks to put on the task board, where each task should have a length of one or two days of work. In addition, any testing issues should be tracked to keep a steady flow of testing.

Agile Testing Practices

Many practices may be useful for testers in a scrum team, some of which include: 

  • Pairing: Two team members (e.g., a tester and a developer, two testers, or a tester and a product owner) sit together at one workstation to perform a testing or other sprint task.
  • Incremental test design: Test cases and charters are gradually built from user stories and other test bases, starting with simple tests and moving toward more complex ones.
  • Mind mapping: Mind mapping is a useful tool when testing. For example, testers can use mind mapping to identify which test sessions to perform, to show test strategies, and to describe test data.

These practices are in addition to other practices discussed in this article and previous articles on the basics pages.

Assessing Quality Risks and Estimating Test Effort

A typical objective of testing in all projects, Agile or traditional, is to reduce the risk of product quality problems to an acceptable level prior to release. Testers in Agile projects can use the same types of techniques used in traditional projects to identify quality risks (or product risks), assess the associated level of risk, estimate the effort required to reduce those risks sufficiently, and then mitigate those risks through test design, implementation, and execution. However, given the short iterations and rate of change in Agile projects, some adaptations of those techniques are required.

Assessing Quality Risks in Agile Projects

One of the many challenges in testing is the proper selection, allocation, and prioritisation of test conditions. This includes determining the appropriate amount of effort to allocate in order to cover each condition with tests, and sequencing the resulting tests in a way that optimises the effectiveness and efficiency of the testing work to be done. Risk identification, analysis, and risk mitigation strategies can be used by the testers in Agile teams to help determine an acceptable number of test cases to execute, although many interacting constraints and variables may require compromises.

Risk is the possibility of a negative or undesirable outcome or event. The level of risk is found by assessing the likelihood of occurrence of the risk and the impact of the risk. When the primary effect of the potential problem is on product quality, potential problems are referred to as quality risks or product risks. When the primary effect of the potential problem is on project success, potential problems are referred to as project risks or planning risks.

In Agile projects, quality risk analysis takes place at two places.

  • Release planning: business representatives who know the features in the release provide a high-level overview of the risks, and the whole team, including the tester(s), may assist in the risk identification and assessment.
  • Iteration planning: the whole team identifies and assesses the quality risks.

Examples of quality risks for a system include:

  • Incorrect calculations in reports (a functional risk related to accuracy)
  • Slow response to user input (a non-functional risk related to efficiency and response time)
  • Difficulty in understanding screens and fields (a non-functional risk related to usability and understandability)

As mentioned earlier, an iteration starts with iteration planning, which culminates in estimated tasks on a task board. These tasks can be prioritised in part based on the level of quality risk associated with them. Tasks associated with higher risks should start earlier and involve more testing effort. Tasks associated with lower risks should start later and involve less testing effort.

An example of how the quality risk analysis process in an Agile project may be carried out during iteration planning is outlined in the following steps:

  1. Gather the Agile team members together, including the tester(s).
  2. List all the backlog items for the current iteration (e.g., on a task board).
  3. Identify the quality risks associated with each item, considering all relevant quality
    characteristics.
  4. Assess each identified risk, which includes two activities: categorising the risk and determining its level of risk based on the impact and the likelihood of defects.
  5. Determine the extent of testing proportional to the level of risk.
  6. Select the appropriate test technique(s) to mitigate each risk, based on the risk, the level of risk, and the relevant quality characteristic.

The tester then designs, implements, and executes tests to mitigate the risks. This includes the totality of features, behaviours, quality characteristics, and attributes that affect customer, user, and stakeholder satisfaction. 

Throughout the project, the team should remain aware of additional information that may change the set of risks and/or the level of risk associated with known quality risks. Periodic adjustment of the quality risk analysis, which results in adjustments to the tests, should occur. Adjustments include identifying new risks, re-assessing the level of existing risks, and evaluating the effectiveness of risk mitigation activities.

Quality risks can also be mitigated before test execution starts. For example, if problems with the user stories are found during risk identification, the project team can thoroughly review user stories as a mitigating strategy.

Estimating Testing Effort Based on Content and Risk

During release planning, the Agile team estimates the effort required to complete the release. The estimate addresses the testing effort as well. A common estimation technique used in Agile projects is planning poker, a consensus-based technique. The product owner or customer reads a user story to the estimators. Each estimator has a deck of cards with values similar to the Fibonacci sequence (i.e., 0, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, …), or any other progression of choice (e.g., shirt sizes ranging from extra-small to extra-extra-large). The values represent the number of story points, effort days, or other units in which the team estimates. The Fibonacci sequence is recommended because the numbers in the sequence reflect that uncertainty grows proportionally with the size of the story. A high estimate usually means that the story is not well understood or should be broken down into multiple smaller stories. 

The estimators discuss the feature, and ask questions of the product owner as needed. Aspects such as development and testing effort, complexity of the story, and scope of testing play a role in the estimation. Therefore, it is advisable to include the risk level of a backlog item, in addition to the priority specified by the product owner, before the planning poker session is initiated. When the feature has been fully discussed, each estimator privately selects one card to represent his or her estimate. All cards are then revealed at the same time. If all estimators selected the same value, that becomes the estimate. If not, the estimators discuss the differences in estimates after which the poker round is repeated until agreement is reached, either by consensus or by applying rules (e.g., use the median, use the highest score) to limit the number of poker rounds. These discussions ensure a reliable estimate of the effort needed to complete product backlog items requested by the product owner and help improve collective knowledge of what has to be done. 

Techniques in Agile Projects

Many of the test techniques and testing levels that apply to traditional projects can also be applied to Agile projects. However, for Agile projects, there are some specific considerations and variances in test techniques, terminologies, and documentation that should be considered. 

Acceptance Criteria, Adequate Coverage, and Other Information for Testing

Agile projects outline initial requirements as user stories in a prioritised backlog at the start of the project. Initial requirements are short and usually follow a predefined format. Non-functional requirements, such as usability and performance, are also important and can be specified as unique user stories or connected to other functional user stories. Non-functional requirements may follow a predefined format or standard, such as [ISO25000], or an industry specific standard.

The user stories serve as an important test basis. Other possible test bases include:

  • Experience from previous projects
  • Existing functions, features, and quality characteristics of the system
  • Code, architecture, and design
  • User profiles (context, system configurations, and user behaviour)
  • Information on defects from existing and previous projects
  • A categorisation of defects in a defect taxonomy
  • Applicable standards (e.g., [DO-178B] for avionics software)
  • Quality risks

During each iteration, developers create code which implements the functions and features described in the user stories, with the relevant quality characteristics, and this code is verified and validated via acceptance testing. To be testable, acceptance criteria should address the following topics where relevant:

  • Functional behaviour: The externally observable behaviour with user actions as input operating under certain configurations.
  • Quality characteristics: How the system performs the specified behaviour. The characteristics may also be referred to as quality attributes or non-functional requirements. Common quality characteristics are performance, reliability, usability, etc.
  • Scenarios (use cases): A sequence of actions between an external actor (often a user) and the system, in order to accomplish a specific goal or business task.
  • Business rules: Activities that can only be performed in the system under certain conditions defined by outside procedures and constraints (e.g., the procedures used by an insurance company to handle insurance claims).
  • External interfaces: Descriptions of the connections between the system to be developed and the outside world. External interfaces can be divided into different types (user interface, interface to other systems, etc.).
  • Constraints: Any design and implementation constraint that will restrict the options for the developer. Devices with embedded software must often respect physical constraints such as size, weight, and interface connections.
  • Data definitions: The customer may describe the format, data type, allowed values, and default values for a data item in the composition of a complex business data structure (e.g., the ZIP code in a US mail address).

In addition to the user stories and their associated acceptance criteria, other information is relevant for the tester, including:

  • How the system is supposed to work and be used
  • The system interfaces that can be used/accessed to test the system
  • Whether current tool support is sufficient
  • Whether the tester has enough knowledge and skill to perform the necessary tests

Testers will often discover the need for additional information (e.g., code coverage) throughout the iterations and should work collaboratively with the rest of the Agile team members to obtain that information. Relevant information plays a part in determining whether a particular activity can be considered done. This concept of the definition of done is critical in Agile projects and applies in a number of different ways as discussed in the following sub-subsections. 

Test Levels

Each test level has its own definition of done. The following list gives examples that may be relevant for the different test levels.

  • Unit testing
    • 100% decision coverage where possible, with careful reviews of any infeasible paths
    • Static analysis performed on all code
    • No unresolved major defects (ranked based on priority and severity)
    • No known unacceptable technical debt remaining in the design and the code
    • All code, unit tests, and unit test results reviewed
    • All unit tests automated
    • Important characteristics are within agreed limits (e.g., performance)
  • Integration testing
    • All functional requirements tested, including both positive and negative tests, with the number of tests based on size, complexity, and risks
    • All interfaces between units tested
    • All quality risks covered according to the agreed extent of testing
    • No unresolved major defects (prioritised according to risk and importance)
    • All defects found are reported
    • All regression tests automated, where possible, with all automated tests stored in a common repository
  • System testing
    • End-to-end tests of user stories, features, and functions
    • All user personas covered
    • The most important quality characteristics of the system covered (e.g., performance, robustness, reliability)
    • Testing done in a production-like environment(s), including all hardware and software for all supported configurations, to the extent possible
    • All quality risks covered according to the agreed extent of testing
    • All regression tests automated, where possible, with all automated tests stored in a common repository
    • All defects found are reported and possibly fixed
    • No unresolved major defects (prioritised according to risk and importance)

User Story

The definition of done for user stories may be determined by the following criteria: 

  • The user stories selected for the iteration are complete, understood by the team, and have detailed, testable acceptance criteria
  • All the elements of the user story are specified and reviewed, including the user story acceptance tests, have been completed
  • Tasks necessary to implement and test the selected user stories have been identified and estimated by the team

Feature

The definition of done for features, which may span multiple user stories or epics, may include:

  • All constituent user stories, with acceptance criteria, are defined and approved by the customer
  • The design is complete, with no known technical debt
  • The code is complete, with no known technical debt or unfinished refactoring
  • Unit tests have been performed and have achieved the defined level of coverage
  • Integration tests and system tests for the feature have been performed according to the defined coverage criteria
  • No major defects remain to be corrected
  • Feature documentation is complete, which may include release notes, user manuals, and on-line help functions

Iteration 

The definition of done for the iteration may include the following:

  • All features for the iteration are ready and individually tested according to the feature level criteria
  • Any non-critical defects that cannot be fixed within the constraints of the iteration added to the product backlog and prioritised 
  • Integration of all features for the iteration completed and tested 
  • Documentation written, reviewed, and approved 

At this point, the software is potentially releasable because the iteration has been successfully completed, but not all iterations result in a release. 

Release

The definition of done for a release, which may span multiple iterations, may include the following areas:

  • Coverage: All relevant test basis elements for all contents of the release have been covered by testing. The adequacy of the coverage is determined by what is new or changed, its complexity and size, and the associated risks of failure.
  • Quality: The defect intensity (e.g., how many defects are found per day or per transaction), the defect density (e.g., the number of defects found compared to the number of user stories, effort, and/or quality attributes), estimated number of remaining defects are within acceptable limits, the consequences of unresolved and remaining defects (e.g., the severity and priority) are understood and acceptable, the residual level of risk associated with each identified quality risk is understood and acceptable.
  • Time: If the pre-determined delivery date has been reached, the business considerations associated with releasing and not releasing need to be considered.
  • Cost: The estimated lifecycle cost should be used to calculate the return on investment for the delivered system (i.e., the calculated development and maintenance cost should be considerably lower than the expected total sales of the product). The main part of the lifecycle cost often comes from maintenance after the product has been released, due to the number of defects escaping to production. 

Applying Acceptance Test-Driven Development

Acceptance test-driven development is a test-first approach. Test cases are created prior to implementing the user story. The test cases are created by the Agile team, including the developer, the tester, and the business representatives and may be manual or automated. The first step is a specification workshop where the user story is analysed, discussed, and written by developers, testers, and business representatives. Any incompleteness, ambiguities, or errors in the user story are fixed during this process. 

The next step is to create the tests. This can be done by the team together or by the tester individually. In any case, an independent person such as a business representative validates the tests. The tests are examples that describe the specific characteristics of the user story. These examples will help the team implement the user story correctly. Since examples and tests are the same, these terms are often used interchangeably. The work starts with basic examples and open questions. 

Typically, the first tests are the positive tests, confirming the correct behaviour without exception or error conditions, comprising the sequence of activities executed if everything goes as expected. After the positive path tests are done, the team should write negative path tests and cover non-functional attributes as well (e.g., performance, usability). Tests are expressed in a way that every stakeholder is able to understand, containing sentences in natural language involving the necessary preconditions, if any, the inputs, and the related outputs. 

The examples must cover all the characteristics of the user story and should not add to the story. This means that an example should not exist which describes an aspect of the user story not documented in the story itself. In addition, no two examples should describe the same characteristics of the user story. 

Functional and Non-Functional Black Box Test Design

In Agile testing, many tests are created by testers concurrently with the developers’ programming activities. Just as the developers are programming based on the user stories and acceptance criteria, so are the testers creating tests based on user stories and their acceptance criteria. (Some tests, such as exploratory tests and some other experience-based tests, are created later, during test execution) Testers can apply traditional black box test design techniques such as equivalence partitioning, boundary value analysis, decision tables, and state transition testing to create these tests. For example, boundary value analysis could be used to select test values when a customer is limited in the number of items they may select for purchase. 

In many situations, non-functional requirements can be documented as user stories. Black box test design techniques (such as boundary value analysis) can also be used to create tests for non-functional quality characteristics. The user story might contain performance or reliability requirements. For example, a given execution cannot exceed a time limit or a number of operations may fail less than a certain number of times. 

Exploratory Testing and Agile Testing

Exploratory testing is important in Agile projects due to the limited time available for test analysis and the limited details of the user stories. In order to achieve the best results, exploratory testing should be combined with other experience-based techniques as part of a reactive testing strategy, blended with other testing strategies such as analytical risk-based testing, analytical requirements-based testing, model-based testing, and regression-averse testing. Test strategies and test strategy blending is discussed in the basics Level pages. 

In exploratory testing, test design and test execution occur at the same time, guided by a prepared test charter. A test charter provides the test conditions to cover during a time-boxed testing session. During exploratory testing, the results of the most recent tests guide the next test. The same white box and black box techniques can be used to design the tests as when performing pre-designed testing. 

A test charter may include the following information: 

  • Actor: intended user of the system
  • Purpose: the theme of the charter including what particular objective the actor wants to achieve, i.e., the test conditions
  • Setup: what needs to be in place in order to start the test execution
  • Priority: relative importance of this charter, based on the priority of the associated user story or the risk level
  • Reference: specifications (e.g., user story), risks, or other information sources
  • Data: whatever data is needed to carry out the charter
  • Activities: a list of ideas of what the actor may want to do with the system (e.g., “Log on to the system as a super user”) and what would be interesting to test (both positive and negative tests)
  • Oracle notes: how to evaluate the product to determine correct results (e.g., to capture what happens on the screen and compare to what is written in the user’s manual)
  • Variations: alternative actions and evaluations to complement the ideas described under activities

To manage exploratory testing, a method called session-based test management can be used. A session is defined as an uninterrupted period of testing which could last from 60 to 120 minutes. Test sessions include the following:

  • Survey session (to learn how it works)
  • Analysis session (evaluation of the functionality or characteristics)
  • Deep coverage (corner cases, scenarios, interactions)

The quality of the tests depends on the testers’ ability to ask relevant questions about what to test. Examples include the following:

  • What is most important to find out about the system?
  • In what way may the system fail?
  • What happens if…..?
  • What should happen when…..?
  • Are customer needs, requirements, and expectations fulfilled?
  • Is the system possible to install (and remove if necessary) in all supported upgrade paths?

During test execution, the tester uses creativity, intuition, cognition, and skill to find possible problems with the product. The tester also needs to have good knowledge and understanding of the software under test, the business domain, how the software is used, and how to determine when the system fails.

A set of heuristics can be applied when testing. A heuristic can guide the tester in how to perform the testing and to evaluate the results [Hendrickson]. Examples include:

  • Boundaries
  • CRUD (Create, Read, Update, Delete)
  • Configuration variations
  • Interruptions (e.g., log off, shut down, or reboot)

It is important for the tester to document the process as much as possible. Otherwise, it would be difficult to go back and see how a problem in the system was discovered. The following list provides examples of information that may be useful to document:

  • Test coverage: what input data have been used, how much has been covered, and how much remains to be tested
  • Evaluation notes: observations during testing, do the system and feature under test seem to be stable, were any defects found, what is planned as the next step according to the current observations, and any other list of ideas
  • Risk/strategy list: which risks have been covered and which ones remain among the most important ones, will the initial strategy be followed, does it need any changes
  • Issues, questions, and anomalies: any unexpected behaviour, any questions regarding the efficiency of the approach, any concerns about the ideas/test attempts, test environment, test data, misunderstanding of the function, test script or the system under test
  • Actual behaviour: recording of actual behaviour of the system that needs to be saved (e.g., video, screen captures, output data files)

The information logged should be captured and/or summarised into some form of status management tools (e.g., test management tools, task management tools, the task board), in a way that makes it easy for stakeholders to understand the current status for all testing that was performed.

Tools in Agile Projects

Tools described in the basics Level pages are relevant and used by testers on Agile teams. Not all tools are used the same way and some tools have more relevance for Agile projects than they have in traditional projects. For example, although the test management tools, requirements management tools, and incident management tools (defect tracking tools) can be used by Agile teams, some Agile teams opt for an all-inclusive tool (e.g., application lifecycle management or task management) that provides features relevant to Agile development, such as task boards, burn-down charts, and user stories. Configuration management tools are important to testers in Agile teams due to the high number of automated tests at all levels and the need to store and manage the associated automated test artefacts.

In addition to the tools described in the basic Level pages, testers on Agile projects may also utilise the tools described in the following subsections. These tools are used by the whole team to ensure team collaboration and information sharing, which are key to Agile practices.

Task Management and Tracking Tools

In some cases, Agile teams use physical story/task boards (e.g., whiteboard, cork-board) to manage and track user stories, tests, and other tasks throughout each sprint. Other teams will use application lifecycle management and task management software, including electronic task boards. These tools serve the following purposes:

  • Record stories and their relevant development and test tasks, to ensure that nothing gets lost during a sprint
  • Capture team members’ estimates on their tasks and automatically calculate the effort required to implement a story, to support efficient iteration planning sessions
  • Associate development tasks and test tasks with the same story, to provide a complete picture of the team’s effort required to implement the story
  • Aggregate developer and tester updates to the task status as they complete their work, automatically providing a current calculated snapshot of the status of each story, the iteration, and the overall release
  • Provide a visual representation (via metrics, charts, and dashboards) of the current state of each user story, the iteration, and the release, allowing all stakeholders, including people on geographically distributed teams, to quickly check status
  • Integrate with configuration management tools, which can allow automated recording of code check-ins and builds against tasks, and, in some cases, automated status updates for tasks

Communication and Information Sharing Tools

In addition to e-mail, documents, and spoken communication, Agile teams often use three additional types of tools to support communication and information sharing: wikis, instant messaging, and desktop sharing.

Wikis allow teams to build and share an online knowledge base on various aspects of the project, including the following:

  • Product feature diagrams, feature discussions, prototype diagrams, photos of whiteboard discussions, and other information
  • Tools and/or techniques for developing and testing found to be useful by other members of the team
  • Metrics, charts, and dashboards on product status, which is especially useful when the wiki is integrated with other tools such as the build server and task management system, since the tool can update product status automatically
  • Conversations between team members, similar to instant messaging and email, but in a way that is shared with everyone else on the team

Instant messaging, audio teleconferencing, and video chat tools provide the following benefits:

  • Allow real time direct communication between team members, especially distributed teams
  • Involve distributed teams in standup meetings
  • Reduce telephone bills by use of voice-over-IP technology, removing cost constraints that could reduce team member communication in distributed settings

Desktop sharing and capturing tools provide the following benefits:

  • In distributed teams, product demonstrations, code reviews, and even pairing can occur
  • Capturing product demonstrations at the end of each iteration, which can be posted to the team’s wiki

These tools should be used to complement and extend, not replace, face-to-face communication in Agile teams.

Software Build and Distribution Tools

As discussed earlier in this article, daily build and deployment of software is a key practice in Agile teams. This requires the use of continuous integration tools and build distribution tools. The uses, benefits, and risks of these tools was described earlier on the basics of agile page. 

Configuration Management Tools

On Agile teams, configuration management tools may be used not only to store source code and automated tests, but manual tests and other test work products are often stored in the same repository as the product source code. This provides traceability between which versions of the software were tested with which particular versions of the tests, and allows for rapid change without losing historical information. The main types of version control systems include centralised source control systems and distributed version control systems. The team size, structure, location, and requirements to integrate with other tools will determine which version control system is right for a particular Agile project.

Test Design, Implementation, and Execution Tools

Some tools are useful to Agile testers at specific points in the software testing process. While most of these tools are not new or specific to Agile, they provide important capabilities given the rapid change of Agile projects.

  • Test design tools: Use of tools such as mind maps have become more popular to quickly design and define tests for a new feature.
  • Test case management tools: The type of test case management tools used in Agile may be part of the whole team’s application lifecycle management or task management tool.
  • Test data preparation and generation tools: Tools that generate data to populate an application’s database are very beneficial when a lot of data and combinations of data are necessary to test the application. These tools can also help re-define the database structure as the product undergoes changes during an Agile project and refactor the scripts to generate the data. This allows quick updating of test data as changes occur. Some test data preparation tools use production data sources as a raw material and use scripts to remove or anonymise sensitive data. Other test data preparation tools can help with validating large data inputs or outputs.
  • Test data load tools: After data has been generated for testing, it needs to be loaded into the application. Manual data entry is often time consuming and error prone, but data load tools are available to make the process reliable and efficient. In fact, many of the data generator tools include an integrated data load component. In other cases, bulk-loading using the database management systems is also possible.
  • Automated test execution tools: There are test execution tools which are more aligned to Agile testing. Specific tools are available via both commercial and open source avenues to support test first approaches, such as behaviour-driven development, test-driven development, and acceptance test-driven development. These tools allow testers and business staff to express the expected system behaviour in tables or natural language using keywords.
  • Exploratory test tools: Tools that capture and log activities performed on an application during an exploratory test session are beneficial to the tester and developer, as they record the actions taken. This is useful when a defect is found, as the actions taken before the failure occurred have been captured and can be used to report the defect to the developers. Logging steps performed in an exploratory test session may prove to be beneficial if the test is ultimately included in the automated regression test suite.

Cloud Computing and Virtualisation Tools

Virtualisation allows a single physical resource (server) to operate as many separate, smaller resources. When virtual machines or cloud instances are used, teams have a greater number of servers available to them for development and testing. This can help to avoid delays associated with waiting for physical servers. Provisioning a new server or restoring a server is more efficient with snapshot capabilities built into most virtualisation tools. Some test management tools now utilise virtualisation technologies to snapshot servers at the point when a fault is detected, allowing testers to share the snapshot with the developers investigating the fault.

Testing Throughout the Software Development Lifecycle

Software Development Lifecycle Models

A software development lifecycle model describes the types of activity performed at each stage in a software development project, and how the activities relate to one another logically and chronologically. There are a number of different software development lifecycle models, each of which requires different approaches to testing.

Software Development and Software Testing

It is an important part of a tester’s role to be familiar with the common software development lifecycle models so that appropriate test activities can take place.

In any software development lifecycle model, there are several characteristics of good testing:

  • For every development activity, there is a corresponding test activity
  • Each test level has test objectives specific to that level
  • Test analysis and design for a given test level begin during the corresponding development activity
  • Testers participate in discussions to define and refine requirements and design, and are involved in reviewing work products (e.g., requirements, design, user stories, etc.) as soon as drafts are available

No matter which software development lifecycle model is chosen, test activities should start in the early stages of the lifecycle, adhering to the testing principle of early testing.

This article categories common software development lifecycle models as follows:

  • Sequential development models
  • Iterative and incremental development models

A sequential development model describes the software development process as a linear, sequential flow of activities. This means that any phase in the development process should begin when the previous phase is complete. In theory, there is no overlap of phases, but in practice, it is beneficial to have early feedback from the following phase.

In the Waterfall model, the development activities (e.g., requirements analysis, design, coding, testing) are completed one after another. In this model, test activities only occur after all other development activities have been completed.

Unlike the Waterfall model, the V-model integrates the test process throughout the development process, implementing the principle of early testing. Further, the V-model includes test levels associated with each corresponding development phase, which further supports early testing. In this model, the execution of tests associated with each test level proceeds sequentially, but in some cases overlapping occurs.

Sequential development models deliver software that contains the complete set of features, but typically require months or years for delivery to stakeholders and users.

Incremental development involves establishing requirements, designing, building, and testing a system in pieces, which means that the software’s features grow incrementally. The size of these feature increments varies, with some methods having larger pieces and some smaller pieces. The feature increments can be as small as a single change to a user interface screen or new query option.

Iterative development occurs when groups of features are specified, designed, built, and tested together in a series of cycles, often of a fixed duration. Iterations may involve changes to features developed in earlier iterations, along with changes in project scope. Each iteration delivers working software which is a growing subset of the overall set of features until the final software is delivered or development is stopped.

Examples include: 

  • Rational Unified Process: Each iteration tends to be relatively long (e.g., two to three months), and the feature increments are correspondingly large, such as two or three groups of related features
  • Scrum: Each iteration tends to be relatively short (e.g., hours, days, or a few weeks), and the feature increments are correspondingly small, such as a few enhancements and/or two or three new features
  • Kanban: Implemented with or without fixed-length iterations, which can deliver either a single enhancement or feature upon completion, or can group features together to release at once
  • Spiral: Involves creating experimental increments, some of which may be heavily re-worked or even abandoned in subsequent development work

Components or systems developed using these methods often involve overlapping and iterating test levels throughout development. Ideally, each feature is tested at several test levels as it moves towards delivery. In some cases, teams use continuous delivery or continuous deployment, both of which involve significant automation of multiple test levels as part of their delivery pipelines. Many development efforts using these methods also include the concept of self-organising teams, which can change the way testing work is organised as well as the relationship between testers and developers.

These methods form a growing system, which may be released to end-users on a feature-by-feature basis, on an iteration-by-iteration basis, or in a more traditional major-release fashion. Regardless of whether the software increments are released to end-users, regression testing is increasingly important as the system grows.

In contrast to sequential models, iterative and incremental models may deliver usable software in weeks or even days, but may only deliver the complete set of requirements product over a period of months or even years.

Software Development Lifecycle Models in Context

Software development lifecycle models must be selected and adapted to the context of project and product characteristics. An appropriate software development lifecycle model should be selected and adapted based on the project goal, the type of product being developed, business priorities (e.g., time-to-market), and identified product and project risks. For example, the development and testing of a minor internal administrative system should differ from the development and testing of a safety-critical system such as an automobile’s brake control system. As another example, in some cases organisational and cultural issues may inhibit communication between team members, which can impede iterative development.

Depending on the context of the project, it may be necessary to combine or reorganise test levels and/or test activities. For example, for the integration of a commercial off-the-shelf (COTS) software product into a larger system, the purchaser may perform interoperability testing at the system integration test level (e.g., integration to the infrastructure and other systems) and at the acceptance test level (functional and non-functional, along with user acceptance testing and operational acceptance testing).

In addition, software development lifecycle models themselves may be combined. For example, a V-model may be used for the development and testing of the backend systems and their integrations, while an Agile development model may be used to develop and test the front-end user interface (UI) and functionality. Prototyping may be used early in a project, with an incremental development model adopted once the experimental phase is complete. 

Internet of Things (IoT) systems, which consist of many different objects, such as devices, products, and services, typically apply separate software development lifecycle models for each object. This presents a particular challenge for the development of Internet of Things system versions. Additionally the software development lifecycle of such objects places stronger emphasis on the later phases of the software development lifecycle after they have been introduced to operational use (e.g., operate, update, and decommission phases). 

Reasons why software development models must be adapted to the context of project and product characteristics can be:

  • Difference in product risks of systems (complex or simple project)
  • Many business units can be part of a project or program (combination of sequential and agile development)
  • Short time to deliver a product to the market (merge of test levels and/or integration of test types in test levels)

Test Levels

Test levels are groups of test activities that are organised and managed together. Each test level is an instance of the test process, consisting of the activities described in the basics of testing article, performed in relation to software at a given level of development, from individual units or components to complete systems or, where applicable, systems of systems. Test levels are related to other activities within the software development lifecycle. The test levels used in this article are:

  • Component testing
  • Integration testing
  • System testing
  • Acceptance testing

Test levels are characterised by the following attributes:

  • Specific objectives
  • Test basis, referenced to derive test cases
  • Test object (i.e., what is being tested)
  • Typical defects and failures
  • Specific approaches and responsibilities

For every test level, a suitable test environment is required. In acceptance testing, for example, a production-like test environment is ideal, while in component testing the developers typically use their own development environment.

Component Testing

Objectives of component testing

Component testing (also known as unit or module testing) focuses on components that are separately testable. Objectives of component testing include:

  • Reducing risks
  • Verifying whether the functional and non-functional behaviours of the component are as designed and specified
  • Building confidence in the component’s quality
  • Finding defects in the component
  • Preventing defects from escaping to higher test levels

In some cases, especially in incremental and iterative development models (e.g., Agile) where code changes are ongoing, automated component regression tests play a key role in building confidence that changes have not broken existing components.

Component testing is often done in isolation from the rest of the system, depending on the software development lifecycle model and the system, which may require mock objects, service virtualisation, harnesses, stubs, and drivers. Component testing may cover functionality (e.g., correctness of calculations), non-functional characteristics (e.g., searching for memory leaks), and structural properties (e.g., decision testing).

Test basis

Examples of work products that can be used as a test basis for component testing include:

  • Detailed design
  • Code
  • Data model 
  • Component specifications

Test objects

Typical test objects for component testing include:

  • Components, units or modules
  • Code and data structures
  • Classes
  • Database modules

Typical defects and failures

Examples of typical defects and failures for component testing include:

  • Incorrect functionality (e.g., not as described in design specifications)
  • Data flow problems
  • Incorrect in the code and/or logic

Defects are typically fixed as soon as they are found, often with no formal defect management. However, when developers do report defects, this provides important information for root cause analysis and process improvement.

Specific approaches and responsibilities

Component testing is usually performed by the developer who wrote the code, but it at least requires access to the code being tested. Developers may alternate component development with finding and fixing defects. Developers will often write and execute tests after having written the code for a component. However, in Agile development especially, writing automated component test cases may precede writing application code. 

For example, consider test driven development (TDD). Test driven development is highly iterative and is based on cycles of developing automated test cases, then building and integrating small pieces of code, then executing the component tests, correcting any issues, and re-factoring the code. This process continues until the component has been completely built and all component tests are passing. Test driven development is an example of a test-first approach. While test driven development originated in eXtreme Programming (XP), it has spread to other forms of Agile and also to sequential lifecycles.

Integration Testing

Objectives of integration testing

Integration testing focuses on interactions between components or systems. Objectives of integration testing include:

  • Reducing risks
  • Verifying whether the functional and non-functional behaviours of the interfaces are as designed and specified
  • Building confidence in the quality of the interfaces
  • Finding defects (which may be in the interfaces themselves or within the components or systems)
  • Preventing defects from escaping to higher test levels

As with component testing, in some cases automated integration regression tests provide confidence that changes have not broken existing interfaces, components, or systems.

There are two different levels of integration testing described in this article, which may be carried out on test objects of varying size as follows:

  • Component integration testing focuses on the interactions and interfaces between integrated components. Component integration testing is performed after component testing, and is generally automated. In iterative and incremental development, component integration tests are usually part of the continuous integration process.
  • System integration testing focuses on the interactions and interfaces between systems, packages, and micro-services. System integration testing can also cover interactions with, and interfaces provided by external organisations (e.g., web services). In this case, the developing organisation does not control the external interfaces, which can create various challenges for testing (e.g., ensuring that test-blocking defects in the external organisation’s code are resolved, arranging for test environments, etc.). System integration testing may be done after system testing or in parallel with ongoing system test activities (in both sequential development and iterative and incremental development).

Test basis

Examples of work products that can be used as a test basis for integration testing include:

  • Software and system design
  • Sequence diagrams
  • Interface and communication protocol specifications
  • Use cases
  • Architecture at component or system level
  • Workflows
  • External interface definitions

Test objects

Typical test objects for integration testing include:

  • Subsystems
  • Databases
  • Infrastructure
  • Interfaces
  • APIs
  • Microservices

Typical defects and failures

Examples of typical defects and failures for component integration testing include:

  • Incorrect data, missing data, or incorrect data encoding
  • Incorrect sequencing or timing of interface calls
  • Interface mismatch
  • Failures in communication between components
  • Unhandled or improperly handled communication failures between components
  • Incorrect assumptions about the meaning, units, or boundaries of the data being passed between components

Examples of typical defects and failures for system integration testing include: 

  • Inconsistent message structures between systems
  • Incorrect data, missing data, or incorrect data encoding
  • Interface mismatch
  • Failures in communication between systems
  • Unhandled or improperly handled communication failures between systems
  • Incorrect assumptions about the meaning, units, or boundaries of the data being passed between systems
  • Failure to comply with mandatory security regulations

Specific approaches and responsibilities

Component integration tests and system integration tests should concentrate on the integration itself. For example, if integrating module A with module B, tests should focus on the communication between the modules, not the functionality of the individual modules, as that should have been covered during component testing. If integrating system X with system Y, tests should focus on the communication between the systems, not the functionality of the individual systems, as that should have been covered during system testing. Functional, non-functional, and structural test types are applicable.

Component integration testing is often the responsibility of developers. System integration testing is generally the responsibility of testers. Ideally, testers performing system integration testing should understand the system architecture, and should have influenced integration planning.

If integration tests and the integration strategy are planned before components or systems are built, those components or systems can be built in the order required for most efficient testing. Systematic integration strategies may be based on the system architecture (e.g., top-down and bottom-up), functional tasks, transaction processing sequences, or some other aspect of the system or components. In order to simplify defect isolation and detect defects early, integration should normally be incremental (i.e., a small number of additional components or systems at a time) rather than “big bang” (i.e., integrating all components or systems in one single step). A risk analysis of the most complex interfaces can help to focus the integration testing.

The greater the scope of integration, the more difficult it becomes to isolate defects to a specific component or system, which may lead to increased risk and additional time for troubleshooting. This is one reason that continuous integration, where software is integrated on a component-by-component basis (i.e., functional integration), has become common practice. Such continuous integration often includes automated regression testing, ideally at multiple test levels.

System Testing

Objectives of system testing

System testing focuses on the behaviour and capabilities of a whole system or product, often considering the end-to-end tasks the system can perform and the non-functional behaviours it exhibits while performing those tasks. Objectives of system testing include:

  • Reducing risks
  • Verifying whether the functional and non-functional behaviours of the system are as designed and specified
  • Validating that the system is complete and will work as expected
  • Building confidence in the quality of the system as a whole
  • Finding defects
  • Preventing defects from escaping to higher test levels or production

For certain systems, verifying data quality may also be an objective. As with component testing and integration testing, in some cases automated system regression tests provide confidence that changes have not broken existing features or end-to-end capabilities. System testing often produces information that is used by stakeholders to make release decisions. System testing may also satisfy legal or regulatory requirements or standards.

The test environment should ideally correspond to the final target or production environment.

Test basis

Examples of work products that can be used as a test basis for system testing include:

  • System and software requirement specifications (functional and non-functional)
  • Risk analysis reports
  • Use cases
  • Epics and user stories
  • Models of system behaviour
  • State diagrams
  • System and user manuals

Test objects

Typical test objects for system testing include:

  • Applications
  • Hardware/software systems
  • Operating systems
  • System under test (SUT)
  • System configuration and configuration data

Typical defects and failures

Examples of typical defects and failures for system testing include:

  • Incorrect calculations
  • Incorrect or unexpected system functional or non-functional behaviour
  • Incorrect control and/or data flows within the system
  • Failure to properly and completely carry out end-to-end functional tasks
  • Failure of the system to work properly in the system environment(s)
  • Failure of the system to work as described in system and user manuals

Specific approaches and responsibilities

System testing should focus on the overall, end-to-end behaviour of the system as a whole, both functional and non-functional. System testing should use the most appropriate techniques (see test techniques) for the aspect(s) of the system to be tested. For example, a decision table may be created to verify whether functional behaviour is as described in business rules.

System testing is typically carried out by independent testers who rely heavily on specifications. Defects in specifications (e.g., missing user stories, incorrectly stated business requirements, etc.) can lead to a lack of understanding of, or disagreements about, expected system behaviour. Such situations can cause false positives and false negatives, which waste time and reduce defect detection effectiveness, respectively. Early involvement of testers in user story refinement or static testing activities, such as reviews, helps to reduce the incidence of such situations.

Acceptance Testing

Objectives of acceptance testing

Acceptance testing, like system testing, typically focuses on the behaviour and capabilities of a whole system or product. Objectives of acceptance testing include:

  • Establishing confidence in the quality of the system as a whole
  • Validating that the system is complete and will work as expected
  • Verifying that functional and non-functional behaviours of the system are as specified

Acceptance testing may produce information to assess the system’s readiness for deployment and use by the customer (end-user). Defects may be found during acceptance testing, but finding defects is often not an objective, and finding a significant number of defects during acceptance testing may in some cases be considered a major project risk. Acceptance testing may also satisfy legal or regulatory requirements or standards.

Common forms of acceptance testing include the following:

  • User acceptance testing
  • Operational acceptance testing
  • Contractual and regulatory acceptance testing
  • Alpha and beta testing.

Each is described in the following four subsections.

User acceptance testing (UAT)

User acceptance testing of the system is typically focused on validating the fitness for use of the system by intended users in a real or simulated operational environment. The main objective is building confidence that the users can use the system to meet their needs, fulfil requirements, and perform business processes with minimum difficulty, cost, and risk.

Operational acceptance testing (OAT)

The acceptance testing of the system by operations or systems administration staff is usually performed in a (simulated) production environment. The tests focus on operational aspects, and may include:

  • Testing of backup and restore
  • Installing, uninstalling and upgrading
  • Disaster recovery
  • User management
  • Maintenance tasks
  • Data load and migration tasks
  • Checks for security vulnerabilities
  • Performance testing

The main objective of operational acceptance testing is building confidence that the operators or system administrators can keep the system working properly for the users in the operational environment, even under exceptional or difficult conditions.

Contractual and regulatory acceptance testing

Contractual acceptance testing is performed against a contract’s acceptance criteria for producing custom-developed software. Acceptance criteria should be defined when the parties agree to the contract. Contractual acceptance testing is often performed by users or by independent testers.

Regulatory acceptance testing is performed against any regulations that must be adhered to, such as government, legal, or safety regulations. Regulatory acceptance testing is often performed by users or by independent testers, sometimes with the results being witnessed or audited by regulatory agencies.

The main objective of contractual and regulatory acceptance testing is building confidence that contractual or regulatory compliance has been achieved.

Alpha and beta testing

Alpha and beta testing are typically used by developers of commercial off-the-shelf (COTS) software who want to get feedback from potential or existing users, customers, and/or operators before the software product is put on the market. Alpha testing is performed at the developing organisation’s site, not by the development team, but by potential or existing customers, and/or operators or an independent test team. Beta testing is performed by potential or existing customers, and/or operators at their own locations. Beta testing may come after alpha testing, or may occur without any preceding alpha testing having occurred. 

One objective of alpha and beta testing is building confidence among potential or existing customers, and/or operators that they can use the system under normal, everyday conditions, and in the operational environment(s) to achieve their objectives with minimum difficulty, cost, and risk. Another objective may be the detection of defects related to the conditions and environment(s) in which the system will be used, especially when those conditions and environment(s) are difficult to replicate by the development team.

Test basis

Examples of work products that can be used as a test basis for any form of acceptance testing include:

  • Business processes
  • User or business requirements
  • Regulations, legal contract and/or standards
  • Use cases and/or user stories
  • System requirements
  • System or user documentation
  • Installation procedures
  • Risk analysis reports

In addition, as a test basis for deriving test cases for operational acceptance testing, one or more of the following work products can be used:

  • Backup and restore procedures
  • Disaster recovery procedures
  • Non-functional requirements
  • Operations documentation
  • Deployment and installation instructions
  • Performance targets
  • Database packages
  • Security standards or regulations

Typical test objects

Typical test objects for any form of acceptance testing include:

  • System under test
  • System configuration and configuration data
  • Business processes for a fully integrated system
  • Recovery systems and hot sites (for business continuity and disaster recovery testing)
  • Operational and maintenance processes
  • Forms
  • Reports
  • Existing and converted production data

Typical defects and failures

Examples of typical defects for any form of acceptance testing include:

  • System workflows do not meet business or user requirements
  • Business rules are not implemented correctly
  • System does not satisfy contractual or regulatory requirements
  • Non-functional failures such as security vulnerabilities, inadequate performance efficiency under high loads, or improper operation on a supported platform

Specific approaches and responsibilities

Acceptance testing is often the responsibility of the customers, business users, product owners, or operators of a system, and other stakeholders may be involved as well.

Acceptance testing is often thought of as the last test level in a sequential development lifecycle, but it may also occur at other times, for example:

  • Acceptance testing of a COTS software product may occur when it is installed or integrated
  • Acceptance testing of a new functional enhancement may occur before system testing

In iterative development, project teams can employ various forms of acceptance testing during and at the end of each iteration, such as those focused on verifying a new feature against its acceptance criteria and those focused on validating that a new feature satisfies the users’ needs. In addition, alpha tests and beta tests may occur, either at the end of each iteration, after the completion of each iteration, or after a series of iterations. User acceptance tests, operational acceptance tests, regulatory acceptance tests, and contractual acceptance tests also may occur, either at the close of each iteration, after the completion of each iteration, or after a series of iterations.

Test Types

A test type is a group of test activities aimed at testing specific characteristics of a software system, or a part of a system, based on specific test objectives. Such objectives may include:

  • Evaluating functional quality characteristics, such as completeness, correctness, and appropriateness
  • Evaluating non-functional quality characteristics, such as reliability, performance efficiency, security, compatibility, and usability
  • Evaluating whether the structure or architecture of the component or system is correct, complete, and as specified
  • Evaluating the effects of changes, such as confirming that defects have been fixed (confirmation testing) and looking for unintended changes in behaviour resulting from software or environment changes (regression testing)

Functional Testing

Functional testing of a system involves tests that evaluate functions that the system should perform. Functional requirements may be described in work products such as business requirements specifications, epics, user stories, use cases, or functional specifications, or they may be undocumented. The functions are “what” the system should do. 

Functional tests should be performed at all test levels (e.g., tests for components may be based on a component specification), though the focus is different at each level. 

Functional testing considers the behaviour of the software, so black-box techniques may be used to derive test conditions and test cases for the functionality of the component or system. 

The thoroughness of functional testing can be measured through functional coverage. Functional coverage is the extent to which some functionality has been exercised by tests, and is expressed as a percentage of the type(s) of element being covered. For example, using traceability between tests and functional requirements, the percentage of these requirements which are addressed by testing can be calculated, potentially identifying coverage gaps.

Functional test design and execution may involve special skills or knowledge, such as knowledge of the particular business problem the software solves (e.g., geological modelling software for the oil and gas industries). 

Non-functional Testing

Non-functional testing of a system evaluates characteristics of systems and software such as usability, performance efficiency or security. Non-functional testing is the testing of “how well” the system behaves. 

Contrary to common misperceptions, non-functional testing can and often should be performed at all test levels, and done as early as possible. The late discovery of non-functional defects can be extremely dangerous to the success of a project. 

Black-box techniques may be used to derive test conditions and test cases for non-functional testing. For example, boundary value analysis can be used to define the stress conditions for performance tests. 

The thoroughness of non-functional testing can be measured through non-functional coverage. Non-functional coverage is the extent to which some type of non-functional element has been exercised by tests, and is expressed as a percentage of the type(s) of element being covered. For example, using traceability between tests and supported devices for a mobile application, the percentage of devices which are addressed by compatibility testing can be calculated, potentially identifying coverage gaps. 

Non-functional test design and execution may involve special skills or knowledge, such as knowledge of the inherent weaknesses of a design or technology (e.g., security vulnerabilities associated with particular programming languages) or the particular user base (e.g., the personas of users of healthcare facility management systems).

White-box Testing

White-box testing derives tests based on the system’s internal structure or implementation. Internal structure may include code, architecture, work flows, and/or data flows within the system. 

The thoroughness of white-box testing can be measured through structural coverage. Structural coverage is the extent to which some type of structural element has been exercised by tests, and is expressed as a percentage of the type of element being covered. 

At the component testing level, code coverage is based on the percentage of component code that has been tested, and may be measured in terms of different aspects of code (coverage items) such as the percentage of executable statements tested in the component, or the percentage of decision outcomes tested. These types of coverage are collectively called code coverage. At the component integration testing level, white-box testing may be based on the architecture of the system, such as interfaces between components, and structural coverage may be measured in terms of the percentage of interfaces exercised by tests. 

White-box test design and execution may involve special skills or knowledge, such as the way the code is built, how data is stored (e.g., to evaluate possible database queries), and how to use coverage tools and to correctly interpret their results.

Change-related Testing

When changes are made to a system, either to correct a defect or because of new or changing functionality, testing should be done to confirm that the changes have corrected the defect or implemented the functionality correctly, and have not caused any unforeseen adverse consequences.

  • Confirmation testing: After a defect is fixed, the software may be tested with all test cases that failed due to the defect, which should be re-executed on the new software version. The software may also be tested with new tests to cover changes needed to fix the defect. At the very least, the steps to reproduce the failure(s) caused by the defect must be re-executed on the new software version. The purpose of a confirmation test is to confirm whether the original defect has been successfully fixed.
  • Regression testing: It is possible that a change made in one part of the code, whether a fix or another type of change, may accidentally affect the behaviour of other parts of the code, whether within the same component, in other components of the same system, or even in other systems. Changes may include changes to the environment, such as a new version of an operating system or database management system. Such unintended side-effects are called regressions. Regression testing involves running tests to detect such unintended side-effects.

Confirmation testing and regression testing are performed at all test levels.

Especially in iterative and incremental development lifecycles (e.g., Agile), new features, changes to existing features, and code refactoring result in frequent changes to the code, which also requires change-related testing. Due to the evolving nature of the system, confirmation and regression testing are very important. This is particularly relevant for Internet of Things systems where individual objects (e.g., devices) are frequently updated or replaced.

Regression test suites are run many times and generally evolve slowly, so regression testing is a strong candidate for automation. Automation of these tests should start early in the project.

Test Types and Test Levels

It is possible to perform any of the test types mentioned above at any test level. To illustrate, examples of functional, non-functional, white-box, and change-related tests will be given across all test levels, for a banking application, starting with functional tests: 

  • For component testing, tests are designed based on how a component should calculate compound interest.
  • For component integration testing, tests are designed based on how account information captured at the user interface is passed to the business logic.
  • For system testing, tests are designed based on how account holders can apply for a line of credit on their checking accounts.
  • For system integration testing, tests are designed based on how the system uses an external micro-service to check an account holder’s credit score.
  • For acceptance testing, tests are designed based on how the banker handles approving or declining a credit application.

The following are examples of non-functional tests:

  • For component testing, performance tests are designed to evaluate the number of CPU cycles required to perform a complex total interest calculation. 
  • For component integration testing, security tests are designed for buffer overflow vulnerabilities due to data passed from the user interface to the business logic. 
  • For system testing, portability tests are designed to check whether the presentation layer works on all supported browsers and mobile devices. 
  • For system integration testing, reliability tests are designed to evaluate system robustness if the credit score micro-service fails to respond. 
  • For acceptance testing, usability tests are designed to evaluate the accessibility of the banker’s credit processing interface for people with disabilities.

The following are examples of white-box tests:

  • For component testing, tests are designed to achieve complete statement and decision coverage for all components that perform financial calculations. 
  • For component integration testing, tests are designed to exercise how each screen in the browser interface passes data to the next screen and to the business logic.
  • For system testing, tests are designed to cover sequences of web pages that can occur during a credit line application.
  • For system integration testing, tests are designed to exercise all possible inquiry types sent to the credit score micro-service.
  • For acceptance testing, tests are designed to cover all supported financial data file structures and value ranges for bank-to-bank transfers.

Finally, the following are examples for change-related tests:

  • For component testing, automated regression tests are built for each component and included within the continuous integration framework.
  • For component integration testing, tests are designed to confirm fixes to interface-related defects as the fixes are checked into the code repository.
  • For system testing, all tests for a given workflow are re-executed if any screen on that workflow changes.
  • For system integration testing, tests of the application interacting with the credit scoring micro-service are re-executed daily as part of continuous deployment of that micro-service.
  • For acceptance testing, all previously failed tests are re-executed after a defect found in acceptance testing is fixed.

While this section provides examples of every test type across every level, it is not necessary, for all software, to have every test type represented across every level. However, it is important to run applicable test types at each level, especially the earliest level where the test type occurs.

Maintenance Testing

Once deployed to production environments, software and systems need to be maintained. Changes of various sorts are almost inevitable in delivered software and systems, either to fix defects discovered in operational use, to add new functionality, or to delete or alter already-delivered functionality. Maintenance is also needed to preserve or improve non-functional quality characteristics of the component or system over its lifetime, especially performance efficiency, compatibility, reliability, security, and portability. 

When any changes are made as part of maintenance, maintenance testing should be performed, both to evaluate the success with which the changes were made and to check for possible side-effects (e.g., regressions) in parts of the system that remain unchanged (which is usually most of the system). Maintenance can involve planned releases and unplanned releases (hot fixes). 

A maintenance release may require maintenance testing at multiple test levels, using various test types, based on its scope. The scope of maintenance testing depends on: 

  • The degree of risk of the change, for example, the degree to which the changed area of software communicates with other components or systems
  • The size of the existing system
  • The size of the change

Triggers for Maintenance

There are several reasons why software maintenance, and thus maintenance testing, takes place, both for planned and unplanned changes. 

We can classify the triggers for maintenance as follows: 

  • Modification, such as planned enhancements (e.g., release-based), corrective and emergency changes, changes of the operational environment (such as planned operating system or database upgrades), upgrades of COTS software, and patches for defects and vulnerabilities
  • Migration, such as from one platform to another, which can require operational tests of the new environment as well as of the changed software, or tests of data conversion when data from another application will be migrated into the system being maintained
    • Retirement, such as when an application reaches the end of its life. When an application or system is retired, this can require testing of data migration or archiving if long data-retention periods are required. 
    • Testing restored/retrieve procedures after archiving for long retention periods may also be needed. 
    • Regression testing may be needed to ensure that any functionality that remains in service still works. 

For Internet of Things systems, maintenance testing may be triggered by the introduction of completely new or modified things, such as hardware devices and software services, into the overall system. The maintenance testing for such systems places particular emphasis on integration testing at different levels (e.g., network level, application level) and on security aspects, in particular those relating to personal data. 

Impact Analysis for Maintenance

Impact analysis evaluates the changes that were made for a maintenance release to identify the intended consequences as well as expected and possible side effects of a change, and to identify the areas in the system that will be affected by the change. Impact analysis can also help to identify the impact of a change on existing tests. The side effects and affected areas in the system need to be tested for regressions, possibly after updating any existing tests affected by the change. 

Impact analysis may be done before a change is made, to help decide if the change should be made, based on the potential consequences in other areas of the system. 

Impact analysis can be difficult if: 

  • Specifications (e.g., business requirements, user stories, architecture) are out of date or missing
  • Test cases are not documented or are out of date
  • Bi-directional traceability between tests and the test basis has not been maintained
  • Tool support is weak or non-existent
  • The people involved do not have domain and/or system knowledge
  • Insufficient attention has been paid to the software’s maintainability during development

Testing throughout the software development lifecycle

A software development lifecycle model describes the types of activity performed at each stage in a software development project, and how the activities relate to one another logically and chronologically. There are a number of different software development lifecycle models, each of which requires different approaches to testing.

Software development and software testing

It is an important part of a tester’s role to be familiar with the common software development lifecycle models so that appropriate test activities can take place.

In any software development lifecycle model, there are several characteristics of good testing:

  • For every development activity, there is a corresponding test activity
  • Each test level has test objectives specific to that level
  • Test analysis and design for a given test level begin during the corresponding development activity
  • Testers participate in discussions to define and refine requirements and design, and are involved in reviewing work products (e.g., requirements, design, user stories, etc.) as soon as drafts are available

No matter which software development lifecycle model is chosen, test activities should start in the early stages of the lifecycle, adhering to the testing principle of early testing.

This categorizes common software development lifecycle models as follows:

  • Sequential development models
  • Iterative and incremental development models

A sequential development model describes the software development process as a linear, sequential flow of activities. This means that any phase in the development process should begin when the previous phase is complete. In theory, there is no overlap of phases, but in practice, it is beneficial to have early feedback from the following phase.

In the Waterfall model, the development activities (e.g., requirements analysis, design, coding, testing) are completed one after another. In this model, test activities only occur after all other development activities have been completed.

Unlike the Waterfall model, the V-model integrates the test process throughout the development process, implementing the principle of early testing. Further, the V-model includes test levels associated with each corresponding development phase, which further supports early testing. In this model, the execution of tests associated with each test level proceeds sequentially, but in some cases overlapping occurs.

Sequential development models deliver software that contains the complete set of features, but typically require months or years for delivery to stakeholders and users.

Incremental development involves establishing requirements, designing, building, and testing a system in pieces, which means that the software’s features grow incrementally. The size of these feature increments varies, with some methods having larger pieces and some smaller pieces. The feature increments can be as small as a single change to a user interface screen or new query option.

Iterative development occurs when groups of features are specified, designed, built, and tested together in a series of cycles, often of a fixed duration. Iterations may involve changes to features developed in earlier iterations, along with changes in project scope. Each iteration delivers working software which is a growing subset of the overall set of features until the final software is delivered or development is stopped.

Examples include:

  • Rational Unified Process: Each iteration tends to be relatively long (e.g., two to three months), and the feature increments are correspondingly large, such as two or three groups of related features
  • Scrum: Each iteration tends to be relatively short (e.g., hours, days, or a few weeks), and the feature increments are correspondingly small, such as a few enhancements and/or two or three new features
  • Kanban: Implemented with or without fixed-length iterations, which can deliver either a single enhancement or feature upon completion, or can group features together to release at once
  • Spiral: Involves creating experimental increments, some of which may be heavily re-worked or even abandoned in subsequent development work

Components or systems developed using these methods often involve overlapping and iterating test levels throughout development. Ideally, each feature is tested at several test levels as it moves towards delivery. In some cases, teams use continuous delivery or continuous deployment, both of which involve significant automation of multiple test levels as part of their delivery pipelines. Many development efforts using these methods also include the concept of self-organizing teams, which can change the way testing work is organized as well as the relationship between testers and developers.

These methods form a growing system, which may be released to end-users on a feature-by-feature basis, on an iteration-by-iteration basis, or in a more traditional major-release fashion. Regardless of whether the software increments are released to end-users, regression testing is increasingly important as the system grows.

In contrast to sequential models, iterative and incremental models may deliver usable software in weeks or even days, but may only deliver the complete set of requirements product over a period of months or even years.

Software development lifecycle models in context

Software development lifecycle models must be selected and adapted to the context of project and product characteristics. An appropriate software development lifecycle model should be selected and adapted based on the project goal, the type of product being developed, business priorities (e.g., time-to- market), and identified product and project risks. For example, the development and testing of a minor internal administrative system should differ from the development and testing of a safety-critical system such as an automobile’s brake control system. As another example, in some cases organizational and cultural issues may inhibit communication between team members, which can impede iterative development.

Depending on the context of the project, it may be necessary to combine or reorganize test levels and/or test activities. For example, for the integration of a commercial off-the-shelf (COTS) software product into a larger system, the purchaser may perform interoperability testing at the system integration test level (e.g., integration to the infrastructure and other systems) and at the acceptance test level (functional and non-functional, along with user acceptance testing and operational acceptance testing).

In addition, software development lifecycle models themselves may be combined. For example, a V- model may be used for the development and testing of the backend systems and their integrations, while an Agile development model may be used to develop and test the front-end user interface (UI) and functionality. Prototyping may be used early in a project, with an incremental development model adopted once the experimental phase is complete.

Internet of Things (IoT) systems, which consist of many different objects, such as devices, products, and services, typically apply separate software development lifecycle models for each object. This presents a particular challenge for the development of Internet of Things system versions. Additionally the software development lifecycle of such objects places stronger emphasis on the later phases of the software development lifecycle after they have been introduced to operational use (e.g., operate, update, and decommission phases).

Reasons why software development models must be adapted to the context of project and product characteristics can be:

  • Difference in product risks of systems (complex or simple project)
  • Many business units can be part of a project or program (combination of sequential and agile development)
  • Short time to deliver a product to the market (merge of test levels and/or integration of test types in test levels)