Simulation and testing
To wrap up this chapter, let's reiterate the importance of being able to test your application on realistic infrastructure. Your test environments should be as live-like as possible. If you don't test on equivalent DBs and networks, then you may get a nasty surprise come deployment time.
When using a cloud hosting provider (and if you automate your server builds), then this is easy. You can simply spin up a staging system that matches production. You don't have to provision it to the exact same scale as long as all the parts are there and in the same place. To reduce costs further, you only need to keep it around for as long as your test.
Alternatively, you could create a new live environment, deploy and test it, then switch over, and destroy or reuse the old live environment. This swapping technique is known as blue-green deployment. Another option is to deploy new code behind a feature switch, which allows you to toggle the feature at runtime, and only for some users...
