Continuous Integration - Reducing Risks
There are chances that things will go wrong on a project. By effectively practicing CI, you find out what happens at every step along the way, rather than later when the project is into the development cycle. CI helps you identify and mitigate risks when they occur, making it easier to evaluate and report on the health of the project based on concrete evidence.
This section is going to concentrate on the risks that can be avoided by using Continuous Integration.
On any project, there are many risks that need to be managed. By eliminating the risks earlier in the development lifecycle, there are lesser chances of these risks developing into issues later on, when the system actually goes live.
Risk 1 – Lack of Deployable Software
“It works on my machine but does not work on another” – This is probably one of the most common phrases encountered in any software organization. Because of the number of changes done to software builds on a daily basis, sometimes there is little confidence on whether the build of the software actually works or not. This concern has the following three side effects.
Little or no confidence in whether we could even build the software.
Lengthy integration phases before delivering the software internally (i.e., test team) or externally (i.e., customer), during which time nothing else gets done.
Inability to produce and reproduce testable builds.
Solution
Eliminating tight coupling between the IDE and the build processes. Use a separate machine solely for integrating the software. Ensure that everything you need to build the software is contained in the version control repository. Finally, create a Continuous Integration system.
The Continuous Integration server can watch for changes in the version control repository and run the project build script when it detects a change to the repository. The capability of the Continuous Integration system can be increased to include having the build run through tests, perform inspections, and deploy the software in the development and test environments; this way you always have a working software.
“Inability to synchronize with the database” – Sometimes developers are unable to recreate the database quickly during development, and hence find it difficult to make changes. Often this is due to a separation between the database team and the development team. Each team will be focused on their own responsibilities and have little collaboration between each other. This concern has the following three side effects −
Fear of making changes or refactoring the database or source code.
Difficulty in populating the database with different sets of test data.
Difficulty in maintaining development and testing environments (e.g., Development, Integration, QA, and Test).
Solution
The solution to the above issue is to ensure that the placement of all database artifacts in the version control repository are carried out. This means everything that is required to recreate the database schema and data: database creation scripts, data manipulation scripts, stored procedures, triggers, and any other database assets are needed.
Rebuild the database and data from your build script, by dropping and recreating your database and tables. Next, apply the stored procedures and triggers, and finally, insert the test data.
Test (and inspect) your database. Typically, you will use the component tests to test the database and data. In some cases, you’ll need to write database-specific tests.
Risk 2 – Discovering Defects Late in the Lifecycle
Since there are so many changes which happen frequently by multiple developers to the source code, there are always chances that a defect can be introduced in the code that could only be detected at a later stage. In such cases, this can cause a big impact because the later the defect is detected in the software, the more expensive it becomes to remove the defect.
Solution
Regression Testing − This is the most important aspect of any software development cycle, test and test again. If there is any major change to the software code, it is absolutely mandatory to ensure that all the tests are run. And this can be automated with the help of the Continuous Integration server.
Test Coverage − There is no point in testing if the test cases do not cover the entire functionality of the code. It is important to ensure that the test cases created to test the application are complete and that all code paths are tested.
For example, if you have a login screen which needs to be tested, you just can’t have a test case that has the scenario of a successful login. You need to have a negative test case wherein a user enters a different combination of user names and passwords and then it is required to see what happens in such scenarios.
Risk 3 – Lack of Project Visibility
Manual communication mechanisms require a lot of coordination to ensure the dissemination of project information to the right people in a timely manner. Leaning over to the developer next to you and letting them know that the latest build is on the shared drive is rather effective, yet it doesn’t scale very well.
What if there are other developers who need this information and they are on a break or otherwise unavailable? If a server goes down, how are you notified? Some believe they can mitigate this risk by manually sending an e-mail. However, this cannot ensure the information is communicated to the right people at the right time because you may accidentally leave out interested parties, and some may not have access to their e-mail at the time.
Solution
The Solution to this issue is again the Continuous Integration server. All CI servers have the facility to have automated emails to be triggered whenever the builds fail. By this automatic notification to all key stakeholders, it is also ensured that everyone is on board on what is the current state of the software.
Risk 4 – Low Quality Software
There are defects and then there are potential defects. You can have potential defects when your software is not well designed or if it is not following the project standards, or is complex to maintain. Sometimes people refer to this as code or design smells — “a symptom that something may be wrong.”
Some believe that lower-quality software is solely a deferred project cost (after delivery). It can be a deferred project cost, but it also leads to many other problems before you deliver the software to the users. Overly complex code, code that does not follow the architecture, and duplicated code - all usually lead to defects in the software. Finding these code and design smells before they manifest into defects can save both time and money, and can lead to higher-quality software.
Solution
There are software components to carry out a code quality check which can be integrated with the CI software. This can be run after the code is built to ensure that the code actually conforms to proper coding guidelines.
