Over a million developers have joined DZone.

Sustainable Automated Testing

DZone's Guide to

Sustainable Automated Testing

· DevOps Zone
Free Resource

Download “The DevOps Journey - From Waterfall to Continuous Delivery” to learn learn about the importance of integrating automated testing into the DevOps workflow, brought to you in partnership with Sauce Labs.

Unit testing over integration testing

Have you ever had a feeling that your automated tests are becoming overhead for the project? That the effort to keep them green seems to be higher than the added value? Then maybe what you do is integration testing and not unit testing.

I had the following discussion last week: “Do you have unit tests?”, “Sure. We do continuous delivery, unit tests are essential for us.” ...... “I’ve looked at your tests and I think that they are more integration tests than unit tests.”, “That’s just a matter of terminology. You know. Our application is heavily data-driven. There’s practically nothing to unit test. We must go to database.”

Hmm. Heavily data-driven. What does that mean? Isn’t every piece of software data driven? Ok, stop philosophising.

First of all, let’s make it clear what is typical for unit testing:

  • Don’t access database
  • Don’t access file system
  • Don’t depend on system time
  • Don’t access active directory
  • Don’t call external services
  • Don’t need real HTTP session to exist

This is not a definition of what unit test is. It’s rather a practical checklist. If you break any of these rules, your test is an integration test. That’s how I see it.

First problem is that integration tests need an environment to run. What’s worse, many of them need different environments. By environment I mean database, files, specific system time, session state, access permissions, … That’s a lot of work to do and maintain.

Second problem is that integration tests are not isolated. They don’t test one specific feature of one specific component. Their purpose is to validate that the whole flow through individual components works.

My intention is not to explain what integration testing is. What I would like to point to is that unit tests should not be replaced by integration tests. You’ll end up in hell where one small change in implementation will lead to many failed integration tests. To find out what is the real cause even of one failed integration test takes much more time than to fix a few unit tests. A lot more time. A unit test is focused to test one particular case of one specific method. If such unit test fails, it’s a piece of cake to find the reason.

How? Write classes that are (unit) testable. Let every dependency of a class to be an interface. Don’t be lazy, write interfaces. Once your class accesses the outer world though interfaces only, you are the winner. Then you can use techniques like stubbing and mocking. Writing and maintaining proper unit tests will not be a problem anymore. Long live heavily data-driven applications.

Discover how to optimize your DevOps workflows with our cloud-based automated testing infrastructure, brought to you in partnership with Sauce Labs


Opinions expressed by DZone contributors are their own.


Dev Resources & Solutions Straight to Your Inbox

Thanks for subscribing!

Awesome! Check your inbox to verify your email so you can start receiving the latest in tech news and resources.


{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}