SLAs for cross-layer adaptation and monitoring of service-based applications

Cross-layer adaptation and monitoring (CLAM) is an approach to the run-time quality assurance of service-based applications (SBAs). The aim of CLAM is to monitor the different layers of an SBA and correlate the monitoring results, such that in the event that a problem occurs an effective adaptation strategy is inferred for enacting a coordinated adaptation across all layers of the SBA. An important aspect of CLAM is the definition of the appropriate Service-Level Agreements (SLAs) for third party services utilised in the different layers of the SBAs. In this paper, we present insights into how to define SLAs for CLAM, by analysing SBAs in order to differentiate the third party business, software and infrastructure services utilised by the SBA. As a case study, we apply the analytical approach to an existing platform-as-a-service framework, which has been developed as an SBA and could benefit from CLAM. The analysis reveals the different third party services and their characteristics, as a precursor to defining SLAs. The case study successfully demonstrates how distinct SLAs for business, software and infrastructure services may be applied respectively in the BPM, SCC and SI layers of an SBA, to provide a flexible monitoring and adaptation response across layers.