Skip to Main Content
A service level agreement (SLA) is a formal contract between the service provider (SP) and the customer. Among various specifications, the SLA states an availability requirement and a penalty the SP pays if this requirement is violated. Traditional approaches to protection use a fixed deployment of backup resources, or do not consider the status of a service. As more heterogeneous applications use optical wavelength-division multiplexing (WDM) backbone network services of the future internet, these service-agnostic approaches are unable to accommodate the differentiated availability requirements. This results in substantial revenue loss either by refusing requests, or by accepting them and then violating their SLAs. In this paper, we propose and investigate the characteristics of a centralized control and management framework for service-aware, admission, and re-provisioning (SAR) in WDM backbone networks of the future internet. Our framework aggressively admits service requests, and dynamically reprovisions existing services by exploiting knowledge of their status. We first propose a GMPLS-based architecture and a revenue model for our analysis. Then we introduce a priority index, called urgency level (UL). An UL is assigned to an existing service indicating its status and relative importance, technically and economically. We reprovision resources by shifting them from low-UL services to provide backups for higher-UL services, which are more at risk of violating their SLAs or which have higher penalties. Our framework increases the revenue for the SP by creating more sales revenue and reducing potential penalties. Numerical results show improved performance in terms of 1) lower resource overbuild; 2) lower bandwidth blocking ratio; 3) higher SLA satisfaction ratio; 4) more balanced SLA violations; and 5) higher overall net revenue. Our framework is also shown to be effective for a range of failure models and penalty strategies.