Cloud services provide computing resources (for example, infrastructure, platform, and software) as services. Utilizing these cloud resources requires a complex life cycle process that involves select, describe, configure, deploy, and manage steps. Cloud services are classified by these resource management techniques, called cloud resource orchestration. The cloud resource orchestration techniques are analyzed using five dimensions: 1) resource dimensions to analyze resource representation, entity modeling, and access techniques; 2) orchestration capabilities that define different resource life cycle actions and automation strategies; 3) user types that analyze who manages cloud resources; 4) runtime environment that distinguishes different cloud services in the virtualization or execution models; and 5) knowledge reuse that distinguishes cloud services depending on whether or not resource description templates, snapshots, or community-based knowledge sharing are available. Based on these five dimensions, a taxonomy of cloud resource orchestration techniques is developed, and it is applied to distinguish 11 cloud services from each other.

The proposed taxonomy that focuses on resource orchestration techniques is useful to understand the differences and commonalities of various cloud services. However, it is not clear whether it could help end users (for example, organizations) to decide which cloud services to choose for their needs. In order for the taxonomy to be more actionable, the end users should be able to pick and choose from these dimension values to create an individualized cloud service. In addition, the end users need a substantial amount of support for the resource management and monitoring capabilities and for the policy specification. These end-user control capabilities seem to be missing in the taxonomy. Also, major cloud services such as Azure and IBM Bluemix are not included in the analysis.