One of the first sets of applications you should consider for the public cloud is development and testing. In the absence of virtualization, application and database servers may be occupying one physical server each, with levels of utilization as low as a woeful 10 percent. Even with virtualization, servers may be underutilized, since the amount of test data in use pales in comparison to the amount of production data.
Synthetic test data may be used with these development and testing servers, but that data can be moved comfortably to the public cloud. Moreover, you will pay only when the cloud services are being used. Agile development methodologies, code branching and continuous integration, which requires many code builds and versions, all require the rapid spin up and down of a large number of application and database servers in parallel. Moving all these servers to the public clouds makes sense. Not only would you pay only when you use these public cloud services, but network latency, storage expenses and performance will also be less of a concern.
2. Development Platform Services
As organizations embrace the principle of DevOps, they increasingly use design, wireframing and prototyping, mash-up, agile project management, automated test tools and development platforms for continuous integration. As noted, these services belong in the public cloud, as that makes it easy for programmers to spin them up when needed and spin them down when not needed. Likewise, it is unlikely for these services to store any sensitive internal data on the public cloud.
3. Training Servers
Training servers have become immensely easier to set up at the beginning of training and tear down at the end. They are also likely to contain synthetic data rather than real data. In other words, they are a natural for the public cloud.
From a console, cloud provisioning tools can set up or tear down servers in the public cloud in a matter of minutes. These tools are also ripe for setting up self-service options that the training groups can handle these themselves.
4. One-time Big Data Projects
When The New York Times needed to convert its entire archives into PDF format a few years ago, the newspaper used the public cloud. Using 100 servers, the job was done in just 24 hours.