Aggregation Semantics for Requirements in Service Templates

Instances of the same Requirement could appear in different Node Templates in the same Service Template. (Note that this could be avoided by requiring the Service Template author perform the aggregation in specific cases where a single author controls all the content included in the Service Template, but not in the general case where all the content may not be known in advance and/or not controlled by the same author.)

A simple example would be a Requirement for memory. Multiple Node Templates might specify a memory Requirement. At deployment time, this Requirement would need to be aggregated up to the container which corresponds to the memory resource that would be allocated to resolve the requirement.

Requirements which can appear in multiple Node Templates in the same Service Template and interact, i.e. must be resolved with respect to all instances of the same Requirement, must specify some kind of aggregation semantic that the TOSCA Runtime implement in order to correctly resolve the Requirements (this semantic could instead be implied by the type of the Requirement).