Through AWS, we saw how cloud providers typically offer compute and storage platforms on a pay-as-you-go model, whereby users (that is, us) only pay for the resources that they actually use.

These platforms can be of significant help in both the development phase as well as in operations for our distributed applications. This is especially true for scalability tests, for instance, and in all situations where ordering and provisioning a set of machines ourselves would take too long or represent too expensive an upfront cost. Not to mention the fact that being able to leverage the vast infrastructure and uptime guarantees of a large cloud provider is no small thing.

At the same time, however, one should be aware of the fact that cloud services are not free. They most definitely are not free economically (apart from maybe the first year of use for AWS). Also, they are not free in terms of time and effort, in that they do not free us from knowing how to administer our resources, install...