Optimizing QoS, performance, and power efficiency of backup services

For most businesses, backup is a daily operation that needs to reliably protect diverse digital assets distributed across the enterprise. Efficiently processing ever increasing amounts of data residing on multiple desktops, servers, laptops, etc., and providing dynamic recovery capabilities becomes a high priority task for many IT departments. Driven by the advances in cloud computing and Software as a Service (SaaS) delivery model, IT departments are transitioning from providing highly customized services to offering on demand services which can be requested and canceled instantly. Backup service providers must be able to react efficiently to on-demand requests and cannot afford labor intensive resource planning and manual adjustments of schedules. Our goal is to automate the design of a backup schedule that minimizes the overall completion time for a given set of backup jobs. This problem can be formulated as a resource constrained scheduling problem where a set of n jobs should be scheduled on m machines with given capacities. In this work, we compare the outcome of the integer programming formulation with a heuristic-based job scheduling algorithm, called FlexLBF. The FlexLBF schedule produces close to optimal results (reducing backup time 20%-60%) while carrying no additional computing overhead and scaling well to efficiently process large datasets compared to the IP-based solution. Moreover, FlexLBF can be easily analyzed in a simulation environment to further tune a backup server configuration for achieving given performance objectives while saving power (up to additional 50% in our experiments). It helps to avoid guess-based configuration efforts by system administrators and significantly increase the quality and reliability of implemented solutions.