What are the top ALM considerations related to cloud testing?
As applications move to the cloud, they create new challenges for the software development lifecycle (SDLC). But, as is usually the case, the new paradigm is more about the platform needs than the processes and best practices that have evolved.
Many aspects of deploying to the cloud can become routine: disaster recovery (DR), security, authentication, access control, scalability and performance. A new skill in testing is in choosing a vendor with the right credentials, reputation and financial stability. Insisting on talking to existing customers, visiting their data center locations and carrying out detailed due diligence into their finances are now part of the testing regimen. Set the bar high; don't compromise on any of your needs, and demand excellence.
When you have a good cloud vendor, you don't have to invest as much in testing DR, performance or those other elements. The difference is, or should be, that the outcomes are much more predictable and much closer to your expectations. And when they are not, you should have strong leverage to get things as you need them. Setting up stringent Service Level Agreements (SLAs) are essential, particularly on turnaround times for issue resolution.
If development of your application remains on premise, or even if this too has moved to the cloud, the release management process and the goals of release management are unaffected. Once again, it is the tools one uses to get code into the cloud that change what we do.
Release automation continues to be the key to minimizing disruption in the development process. And in a cloud context, it is absolutely essential. As we develop the code and pass through the gates of the SDLC, we need to ensure that the handoffs, and the code moves, builds, packaging and deployments are handled in an automated manner. When a new deployment platform comes along, it merely requires the retiring of the automation to ensure consistent results.
All other things being equal, testing for the cloud means your SDLC can remain intact, but you do need to make sure that you have the right tooling, that the built-in time frames for the phases and milestones is carefully reviewed and that expectations are correctly set.
This was first published in November 2011