Survivable virtual infrastructure mappings in multi-datacenter systems

With the increasingly popular virtualization of both computing and networking resources in a multi-datacenter system (called a physical substrate), multiple virtual infrastructures (VIs) can share the physical resources of the underlying substrate, and accordingly even a single failure in the substrate can affect a large number of VIs and the services they offer. Thus, the problem of efficiently mapping a VI to a substrate while guaranteeing the VI's survivability in the event of failures in the substrate becomes important. In this paper, we investigate various survivable virtual infrastructure mapping approaches and algorithms to recover from physical server failures, physical link failures and regional failures. To improve the backup resource efficiency, the paper further discusses the sharing schemes of multiple VI requests.