That is a question which most environment managers will hear all too often. So what is the answer? Yes, would be obvious, but I usually counter with an array of questions such as: What do you understand by “an environment”? What do you want it for? What will be tested on it? Who will manage the testing for the environment? Who will use the environment? When do you need it and how long for? Where do you want it? What should be in the environment? Which interfaces should be used? Do you need stubs? How flexible does it need to be? Who will manage the testdata? How many instances does it make sense to have? Does it need to reflect production? How well does it need to perform? What is the budget? and the list goes on…
Many of the requesters will start to look vague by the third question, and then totally lost by the last. The simple request has become a lot more complex and testing will not likely start within the next hour because the test environment manager wants to understand what is actually required, as opposed to what the tester thought he or she might like.
It is not that I wish to be a pain nor waste time unnecessarily, but getting it right the first time can save money and time in the future. It should be the aim of any environment manager to build stable, manageable and well prepared environments which are fit for purpose and do not contain excess baggage.
As I have previously stated, effective testing relies on reproduction. This can be achieved by keeping changes to a minimum and properly managed.
Knowing exactly what is in the environment and why it is there will allow the manager to better coordinate change. Assessing the impact of environmental changes with the testing team and development whilst understanding the possible effects it might have on the rest of the environment, or other environments for that matter, should reduce the number of unforeseen issues.
Environments have a tendency to get very messy very quickly if inefficiently managed, especially when used in parallel. Timelines, requirements, software, versioning, databuilds, servers, storage, databases and testdata can get out of hand when not well coordinated. People, especially those in development, are quick to question the environments when bugs are found in testing. And not knowing what was where when the software was tested and not being able to reproduce a defect can result in longwinded and expensive analysis phases, giving test environments their bad name.
I have to laugh when requests for test environments to support complex and costly projects are proposed over coffee, without an ounce of thought going into what it is that is actually needed. I also enjoy watching the requester deflate as I question the needs. That is priceless and makes up for a lot of the stress which follows 🙂