A data integration platform is, by definition, at the center of the information system. There are going to be many opportunities to test it.
At a minimum, testing will be required before each new release of the platform itself. Depending on the changes made, there may be different testing activities:
This type of test is necessary for example in the following cases:
- addition of a new application
- addition or modification a business process
- modification of an application causing an update in a flow/mapping.
In all these cases, the development must be validated by an exploratory test phase, during which the tester will make sure that the addition or modification behaves as expected. The tester will define the different test cases according to the specifications. He will take care to take into account the boundary cases, as well as the cases of errors: for example a bad data format or incomplete data.
functional changes or additions requiring validation tests may also involve a regression testing phase if they might have side effects on existing features. For example, in the case of modifying a mapping used by several flows, it will be necessary to ensure the absence of regression for all the impacted flows. On the other hand, a mapping used by a single flow that would be added or modified will not require regression tests on the other flows.
Another case requiring regression testing is when there is a change in the software or hardware used for the integration platform. Here, it is important to make sure of the absence of non-functional regression: are the performances, the load and the safety still at the expected level? Such an update can however have a functional impact: it is sometimes necessary to modify existing mappings following a modification of the API of the platform. In this case, the same test activities must be performed as described in the previous point.