Resource Pooling
Service providers rent their computing resources in an automated manner. They also have to be capable of supporting multiple users simultaneously, and there must be enough resources available for the customers.
At the same time, the resources used by the customers are the resources of the data centers, which are not unlimited. To solve this situation, the NIST defines the resource pooling characteristic (see Figure 21-4). It describes the need for the utilization of the resources to be most effective. This requirement also means that the resources’ utilization will become more flexible as well. The virtualized resources provided from a physical server have to be available to multiple users, as a single user might not utilize them all. And this becomes even more effective when we scale it to the level of a data center and beyond. The end user does not know on which exact physical resources their workloads are running—and does not need to know this information, as long as the resources are available and the services are up and running and accessible.
Figure 21-4 Resource Pooling Essential Characteristic
For the purposes of compliance and to stay close to the customers, the cloud service providers use the approach of dividing their resources into regions, countries, continents, and so on.
Rapid Elasticity
When resources are not being used by an end user, they need to go back into the pools to be available for new provisioning. This is the rapid elasticity characteristic (see Figure 21-5), and it creates the impression that the cloud resources are infinite. However, the more important aspect is that when the resources are managed in such an elastic manner, the cloud providers are capable of supporting horizontal and vertical scaling of the capacities, to support the needs of the users. As an example, some businesses see spikes in the services they offer during certain periods of the year. For these time slots, the companies will need more resources to keep their services up and running, and during the rest of the year, these resources will be idle. If these companies were to buy and build their own infrastructure, the cost would be too high for something that’s used for a short period of time. By using cloud resources, these companies can be developing and testing their services, using far fewer resources, while during the peak hours, days, weeks, they can automatically, or manually, scale up and down. In this way, they will be capable of providing only the needed resources, and thus expenses will be optimized.
Figure 21-5 Rapid Elasticity Essential Characteristic