To adapt to continuously changing workloads in networks, components of therunning network services may need to be replicated (scaling the networkservice) and allocated to physical resources (placement) dynamically, alsonecessitating dynamic re-routing of flows between service components. In thispaper, we propose JASPER, a fully automated approach to jointly optimizingscaling, placement, and routing for complex network services, consisting ofmultiple (virtualized) components. JASPER handles multiple network servicesthat share the same substrate network; services can be dynamically added orremoved and dynamic workload changes are handled. Our approach lets servicedesigners specify their services on a high level of abstraction using servicetemplates. From the service templates and a description of the substratenetwork, JASPER automatically makes scaling, placement and routing decisions,enabling quick reaction to changes. We formalize the problem, analyze itscomplexity, and develop two algorithms to solve it. Extensive empirical resultsshow the applicability and effectiveness of the proposed approach.
展开▼