ABSTRACT
The efficient allocation of computational resources requires some understanding of the resources themselves and their availability, as well as the load that must be allocated to them. This chapter presents the requirements for workload and platform models that are suitable to support resource allocation mechanisms in embedded, high-performance and cloud computing. Such models can be used as internal representations, allowing resource allocation mechanisms to evaluate different allocation alternatives. The chapter also presents the requirements for such load modelling approach along four distinct categories: structure, temporal behaviour, resource constraints and load characterisation. The resourcing of applications defines which kind of resources a given job requires for its execution. This requires a taxonomy of resources over different types. A strictly periodic series of jobs can be characterised by its release time, the period after which a new job is released, and the time interval each job requires from a notional resource.
