Efficiently scheduling Cloud resources for peak or urgent events in telecom system

Call Detail Record or Charging Data Record (CDR) processing is a computation-intensive task, which is traditionally handled by costly high-performance machines. If extra computing resources are not allocated beforehand, it is fragile to handle peak or urgent events. Cloud computing, which offers elastic computing resources, can cure this problem. In this paper, we investigated several Cloud scheduling strategies for minimizing cost in peak and/or urgent CDR processing scenarios while preserving designated Service Level Agreements (SLAs). To be realistic, we assumed that incoming event is unknown to scheduler until it arrives. Results show that a well-performing strategy in normal scenario could be the worst strategy in peak & urgent scenario. Scheduling strategies should be carefully chosen to minimize cost when handling unexpected events in Cloud.