Designing Integration Tests, Integration Strategies for Procedures and Functions - Essay Example

Integration Test December, 13, 4 Introduction 5 Goals 5 Integration Strategies for Procedures and Functions 5 Integration strategies for classes 6 Designing integration test 6 Integration test plan 7 Reference 8 Abstract Integration testing has been as one of the way that systems subcomponents are tested after being integrated. The paper tries to research on the design of integrated testing to ensure testing is carried out to the full. The testing strategies for classes and components have also been looked at to ensure they tally with integration testing. Integration testing has been seen to eliminate errors that would not be seen during the unit testing. These errors include functional errors and interface errors. Introduction Software needs testing either testing throughout its lifecycle or at the end of the whole process. Testing is done to establish errors and improvements that need to be done to the software. Unit testing is done independently on each component of a system before integration to ensure they contain no bugs and function within the requirements. After integrating these units a second form of testing should be carried out to ensure compatibility of the different units in performing their functionalities. Testing is known to eliminate a larger percent of errors in software which would be catastrophic if left in the final product. Different modes of testing should be carried on the same software. This ensures errors not discovered by one method will be identified through the other. Developers should also include software testing as one of the phase in software development. Goals The main objective of integration testing is to ensure after integration, different components of a system work as required. This is because most system are developed as independent components which are later integrated into one. Integration testing is supposed to establish reliability of the integrated components. Reliability will look at whether the final system is able to deliver the needed functionalities required by the users. The system can be fed with dummy data to see the output. The matching of the output with the expected results signifies the system is reliable. Functional requirements of the system should be verified in the final system. These requirements might include carrying analysis on data to give a certain output. Some other functional requirements of the system might provide an alert message to users after a certain period of time. The integrated system should provide these requirements on timely basis and in the format prescribed by the user. Integration testing is used to test the interface of a system after integration. This is done to the interface after unit testing has been carried out. The interface after integration will interact with interfaces of other units via command calls. The testing should ensure the communication between the interfaces is bug free(Thelma, Weslety, Silvia, and Aurora, 2006). Integration Strategies for Procedures and Functions Strategies used in procedures and functions are top down and bottom up integration. These strategies involve adding a single unit to the system each at a time. The addition continues up to the when the last component is tested. The bottom up approach starts testing by units that exist at the bottom layer up to the one at the top. Case scenario of bottom up would be carrying a test at a graph made by integrating units from M0 to M4. This would involve first carrying a test at the bottom layer which starts form M1 to M2. Completion of testing of this layer would need integration of the M3 to M4 layer which is dependent of the first layer that had been tested. The testing will continue with adding new layers to the current layer up to when all the layers have been tested. The testing of the top layer marks the end of bottom up testing approach. Each layer of tested units represents a module of the final system. The top down approach begins with modules at the highest level of system moving towards the bottom. This approach begins with a complete system unlike the bottom up approach which begins by building up the system. This model is seen to start with the control and main models in a system. Testing of some models cannot be carried using only bottom up or top down approach. These require integration of the two approaches coming up the hybrid approach. The approach combines characteristics of top down and bottom up approach in carrying out the test (Christian , Andreas, Judith, Martin, Sabine, and Doris, 2007). The big bang integration can also be used. This approach integrates units haphazardly and then continues to carry integration testing on them. This method complicates integration testing because of the haphazard integration. Incremental integration is another way by which all components of a system are prepared and then integrated at once. This method has to take care of the components that depend on each other while carrying out the integration. The method assumes that all units will be prepared during the time of integration. The developers should carry out a census of all the components before starting the integration process to ensure the process does not get into a deadlock(Thelma, Weslet, Silvia, and Aurora, 2006)). Integration strategies for classes Different classes are integrated either through inheritance generalization or dependency. This integration leads to the need of developing strategies to carry out testing on the integrated component. One of the proposed strategies is the ordering class integration. Different mode of ordering has been proposed to avoid stubbing in testing. The first mode of ordering is assigning levels to classes depending on their aggregation and relationships. Those classes placed at the lower level only have association relationship. The problem with the integration through inheritance is that some classes might not have inheritance relationship and therefore lead to a deadlock in the integration process. Another strategy is the use of directed graphs known as the object oriented relation diagrams. This method involves testing the classes on their own first. The aspect is then integrated in the classes which are then tested again. This method has been seen to have a number of problems like difficulties in adapting all factors used in stub creation. This strategy also reduces broken cycles as one of its main goal. These problems are overcome by having multiple algorithms in coming up with the optimal solution in a graph. One of the solutions is use of multi objective approach. This approach eliminates the need of use of weights in generating better solution for use by testers. Designing integration test The integration of units before testing requires design to be carried out carefully to ensure everything is in order before the testing period. The following are things that require to be done during the design of the software. Developers must prepare test cases for the entire program. This test case should be documented and reviewed to ensure its workability before implementation. The next step will involve creating test cases. The creation requires developers to come up with the test cases together with their test data. The test case should look at certain areas such as calls to other routines and passing of parameters between functions. The test case should also check on data dependency where one unit of a program requires output of another in order to function. The designed test case should also take into consideration the validation of data by the system. Validation can take place at the interface during execution of the data. The test case should include measures that will show weaknesses in these areas. The interrelationship of interfaces of the subsystems is another thing that should be included in designing a test case. Systems have interfaces that trigger others to do an action on their behalf. The test case should ensure that the triggered event occurs at the right time and in the right manner. Global data is data used by all subsystems. This data has to be checked for consistency during the entire testing period. The consistency will ensure timely reflection of changes made on the data by each subsystem. When developers finish designing the test case they should document everything that was done. Documentation will ensure future need of change on the test case will be carried out easily (Craig, 2007). Integration test plan The test plan involves multiple activities before executing the final test on the integrated system. The elements of a test plan include getting general information of the system which leads to the test definition. The general information includes the scope and purpose of the test plan. The scope of the test plan should be stated in relation to the system. The purpose will show the needs that have led to carrying out the test. The test definition comes up with the different tests that are to be carried out on the system. Some of the tests that are defined here are the functional testing and ease of use testing. The ease of use testing will determine whether the system interface followed usability principles. The functional testing determines whether the system is delivering the needed requirements by users. This level also involves establishing the requirements of the system that will be tested and expected results after testing. The testing team will also define the hierarchy that will be used in carrying out the different tests on the system. The test should include the external software to be used in carrying it out. Test might include usage of some data. The test data, input data controls and output data control should be established before carrying out the test. Recovery of the original data after carrying out the test should be looked at in the plan. This is important to establish validity of the data used after the test. Limitations that may exist while carrying out the test should be included in the plan. Some of the limitations might be the lack of qualified personnel to carry out the test. Another is the lack of corporation by developers of the system with the testers who might be external tester. The final step in test plan is implementing the test on the system. The implementation involves carrying the actual test on the integrated system (Craig, 2007). Reference Craig, B.(2007). Designing integration tests. Retrieved from http://it.toolbox.com/blogs/enterprise-solutions/designing-integration-tests-17371 Christian, P., Andreas, F., Judith, H., Martin, R., Sabine, R., and Doris, W.(2007). On the integration of design and test: A model-Based Approach for embedded systems. http://www.cs.uwaterloo.ca/~gweddell/cs446/ITandPC.pdf. Philippe, M., and Linda, B. (2006).Rapid Integration of Software Engineering Techniques. Retrieved from http://www.mendeley.com/research/aspectsclasses-integration-testing-strategy-incremental-approach/. Thelma, E., Weslety, G., Silvia, R., and Aurora, T.(2006). Generating Integration Test Orders for Aspect-Oriented software with multi-objective algorithms. Retrieved fromhttp://www2.dc.ufscar.br/~lawasp/2011/artigos/90224.pdf Briad, L.(2001). Revisiting strategies for Ordering class integration testing in the presence of dependency cycles. Retrieved from http://www.arnetminer.org/viewpub.do?pid=450202. Craig, B.(2007). Technical Architecture: Test Planning-Integration Testing Process. Retrieved from http://it.toolbox.com/blogs/enterprise-solutions/technical-architecture-test-planningintegration-testing-process-46589 Read More