The best way to avoid confusion in making decisions about cloud computing is to start with a good definition. The best one we’ve found is by the National Institute for Science and Technology (NIST). It describes cloud computing’s essential characteristics, service models and deployment models. With it, you can distinguish cloud computing from hosting and other entities “in the cloud.”

The definition presents five essential characteristics, which distinguish cloud computing from other outsourcing to the cloud. They include:

• On-demand self-service
• Broad network access
• Resource pooling
• Rapid elasticity
• Measured Service

Considering these characteristics can begin to indicate when IT managers should choose cloud computing over other alternatives. For example, measured service may appeal to development and testing projects that can be easily put up and torn down, without requiring the purchase of a server.

Service Models

The definition offers three service models that varieties of cloud computing fulfill. The first is Cloud Software as a Service, epitomized by applications like salesforce.com and hosted Microsoft Exchange. The second is Cloud Platform as a Service, perhaps best illustrated in the Google App Engine. The third is Cloud Infrastructure as a Service, which is where Amazon Web Services and Microsoft Azure best fit.

These service models can also be applied to traditional hosting. The difference is that the examples here display those essential characteristics mentioned above.

Deployment Models

The NIST definition includes four deployment models:

• Private cloud
• Community cloud
• Public cloud
• Hybrid cloud

Most of these deployment models have received a lot of attention. The one that interests us at the moment is the community cloud, where the cloud infrastruture is shared by several organizations and supports a community with shared concerns. An interesting example is the United Nations International Computing Centre (UNICC). The fascinating thought is going beyond providing infrastructure to shared management of say, databases, and even applications.

A couple of subsequent posts on cloud computing will discuss how nonprofits can use this definition to respond to the drivers towards cloud computing while avoiding some roadblocks.