文摘
Recent evolutions of virtualization technologies allow carriers to optimize and monetize their network infrastructures in new ways, acting as virtual infrastructure providers. By assuming an underlying GMPLS-enabled network infrastructure connecting a number of geographically dispersed data centers, in this paper we define an architectural framework that allows infrastructure providers to optimally use their resources to provide Virtual Infrastructures on demand. The architecture we propose is designed as an extension of the standard Path Computation Element (PCE) architecture. A centralized entity, named VRO, is responsible for optimally allocating the physical resources needed to deploy a requested Virtual Infrastructure. In the paper, we also present how it is possible to apply our framework to pursue green management objectives so that OPEX expenditures can be reduced, while preserving contractual SLAs. We also describe a prototype of our framework that is able to configure GMPLS-enabled network nodes and Cloud-enabled data centers in order to create Virtual Infrastructures.
