Performance Awareness Example Clause Samples

Performance Awareness Example. ‌ In this section we demonstrate how we imagine the performance awareness of an application. We demonstrate this on an ad-hoc cloud case study, where a smart phone would offload computationally intensive applications to the nearby available computational nodes to improve battery lifetime [BBHK13]. Along that we show how the models and techniques from previous sections are put into practice. To elaborate the application migration example, we use the jDEECo component model [BGH+13], which realizes the concepts of the SCEL formalism for developing adaptive ensembles. As a major feature, the ensembles consist of components that communicate exclusively through shared knowledge – we therefore include performance measurements (e.g., results of SPL formulas evaluated on data obtained with DiSL-based instrumentation) among the knowledge elements. As a side note, including performance in knowledge is not entirely straightforward. In SCEL, the knowledge elements are exchanged between components throughout the lifetime of the ensemble – knowledge that is requested is assumed to exist and communicating it does not bring any additional cost. This does not apply to measurements expressed as performance knowledge – measurements are not collected before being requested, because measuring just to populate knowledge that may never be used would incur an excessive overhead. Once demanded, the instrumentation and measurement take certain time, and incur overhead on the observed rather than the observing component. Further penalty may be associated with communicating the measurements. The example is also described in [BBHK13].