IT is evolving from deploying services by statically assigning them to resources, to dynamically allocating services to available resources. The ability to map services to the infrastructure, based on the required SLOs, requires a model of the resources, their capabilities, and overall topology. This model ultimately enables the computing utility to be dynamic and self-regulating, thereby facilitating scale-out of services.

The model serves as a foundation for effective management of the IT facility [4]. It is typically divided into a static and dynamic component. The static model typically maintains these:

1. Asset information. This comprises information on all the elements within the IT facility (e.g., hardware, OS, and software).
2. Static topology. This maps the relationships both physical and logical between devices (e.g., network, compute, storage, clients, OS, and applications).

The dynamic model captures the current state of the IT environment at an instance in time. Typically, the dynamic model may include dynamic topology that maps the current relationship (typically logical) in the environment, such as between devices and between logical platforms and currently deployed services.

Benefits of the model

The ability to deploy new builds to classes of servers rather than to individual servers will also ease the task of change and release management, which will be concerned with an inventory of elements and class relationships, rather than specific allocations of resources. This effectively changes the meaning of deployment. In addition, the ability to deploy to a logical representation of a computing resource makes it possible to refactor the infrastructure, giving additional flexibility to the implementers, while providing stability through virtualization. An example of such refactoring might be changing from an enclosure in which there were many server blades, each having a complement of processors, memory, storage, and network components, into an enclosure that separates compute, storage, and network functionally.

Without the model, decisions on how a service should be managed over its lifecycle could not be made. The model is used to analyze whether a service is meeting its desired state by collecting metrics from the service and the resources hosting that service. It can also be used to make informed decisions on available capacity to deploy the service in such a manner that it performs within the required SLOs.